While the quality expected of eggs retrieved at my age means the chances of achieving a live birth from these eggs is quite high, I want to give myself the greatest possible certainty. With that in mind, I plan to do a second round, with the aim of reaching at least 20 eggs in total.

The timing of my freezing is quite good, because in the coming days, an article I co-authored with for is coming out in the next few days. There, we demistify the entire process, showing that egg freezing works when done at the right age and at the right clinic, explaining the underpinnings of female fertility decline with age and giving some advice on the steps to start the process. It’s in many ways a follow-up on my earlier article with them, Fertility on Demand, where I make the case that one way to allow women to pursue the careers they want is through the aid of reproductive technologies.

The main article, however, doesn't cover every practical detail of the process in depth. This piece contains some of these details.

Practical considerations

The most important considerations are cost; the optimal time to freeze one’s eggs; choosing a clinic; the choice between freezing eggs or embryos, as well as considering the influence of any reproductive conditions.

Costs

Table 1. Average costs of the egg freezing procedure across three different countries. There might be additional optional costs for add-ons like PGT-A. Note that there is also substantial variation between clinics, especially in the US.

To avoid unexpected costs, we advise inquiring about all prices with the clinic you are choosing. It is particularly common for the cost of medications, those of additional tests like ultrasounds and blood tests, or annual storage to be excluded from the total. Clinics also often offer discounts for purchasing multiple cycles upfront so you might want to plan to do two cycles depending on your expected number of eggs per cycle.

Egg freezing is expensive in the US. In total, an egg freezing cycle including tests and medications will cost around $17,000. The UK is more affordable, at roughly $8,000 per cycle and Spain is even better, averaging $5,000. Luckily, many international clinics offer the option to conduct consultations online and travel only for the final retrieval procedure to the clinic. But bear in mind that they might need more than one cycle to get enough eggs.

If you’re not able to pay this out-of-pocket, some companies offer egg freezing as part of their insurance policies. Another alternative is to sell off half your eggs through services like Cofertility and Freeze and Share, which covers the cost of the procedure.

The optimal time to egg freeze

If cost is not a constraint, the optimal age for egg freezing is around 25 to 27, when a single cycle is more likely to yield enough eggs and a greater proportion of those eggs are euploid.

That said, egg freezing is a significant financial undertaking, and for many women that reality cannot be set aside. Since most women can still conceive naturally even in their late thirties, it may make sense to wait a little longer and factor in personal circumstances — relationship status, whether and when children feel like a realistic prospect, and so on.

Weighing all of this together, somewhere between 29 and 32 seems like a reasonable sweet spot. Official guidance typically cites anything under 35 as acceptable, but this reflects the older assumption that fertility drops sharply at that age. The evidence now suggests meaningful egg quality decline begins well before 35 — so from the standpoint of egg freezing specifically, doing so before 32 is likely the better course.

One useful data point when weighing up the timing is your antral follicle count (AFC), which can be measured via ultrasound without committing to a full cycle. As discussed in the main article, the number of eggs retrieved in a cycle tends to be roughly 80% of your AFC, making it a meaningful predictor of outcomes. If your AFC is low for your age, that is a strong signal to act sooner rather than later — and possibly to plan for more than one cycle from the outset.

Alternatively, you might measure your anti-mullerian hormone, or AMH, through a blood test. While not as precise as a follicle count for predicting the outcome of a single cycle, it can be a useful warning for low ovarian reserve. If you are in the slightly low or low categories, you might start freezing sooner rather than later.

Table 2. Egg reserve and AMH levels. From here.

Choosing the right clinic

Not all IVF clinics are created equal. While you might worry that comparing success rates between clinics is tricky due to selection bias, in practice it seems like they actually provide a good sense of clinic quality. Firstly, some of the best clinics accept difficult cases more often, but are so much better that they still achieve higher success rates. Secondly, IVF patients are a relatively homogeneous group with high socio-economic status.

We can also tell that technical quality varies widely among clinics. For example, low oxygen levels are important in embryo culture because early embryos are not adapted to the high oxygen concentration found in room air. In vivo, mammalian embryos develop in the fallopian tube and uterus at oxygen concentrations around two to eight percent, far lower than the 21 percent oxygen of atmospheric air. Yet many clinics still fail to use natural oxygen levels in their embryo culture chambers.

Overall, the difference in success rates between clinics should give you important information about the clinic’s quality, as long as you subset for your age group. It is generally advisable to double check the numbers you see on clinics’ websites against official statistics. The most comprehensive reports can be found in the US, through the CDC website. There are also websites that provide rankings of fertility clinics through an evaluation process that included a nationwide online survey of reproductive endocrinologists, infertility specialists, obstetricians and gynecologists. In the UK, the UK’s fertility regulator provides a Choose your clinic tool where you can see success rates for each clinic compared to the national average. In Spain, the equivalent is Reproduccion Asistida ORG.

What should I look for?

There are three important metrics: success per embryo transfer, success per IVF cycle and the frozen egg survival rate. Unfortunately, no official agency records frozen egg survival and only the CDC reports both success per embryo transfer and success per IVF cycle.

Success per embryo transfer measures the live birth rate once a viable embryo has been transferred into the uterus. It’s an indicator of the clinic’s skill at the final step of the process. Success per cycle, on the other hand, is a more comprehensive metric that measures the live birth rate from the start of an egg retrieval cycle. This includes all the attrition points along the way, such as egg retrieval, fertilization, and embryo development.

Success per embryo transfer is a useful metric because different women have different ovarian reserves which is independent of clinic. But it does not capture how good a clinic is at the steps before embryo transfer, including the key bottleneck of turning eggs into embryos, and it can be easily influenced by what the clinic considers to be a transferable embryo. If only very high quality embryos are considered for transfer, then the success per embryo transferred can be high, but the woman might overall have fewer embryos available.

This brings us to the question of how clinics choose what a high quality embryo is. The most objective test is for aneuploidy, whether the embryo has the right number of chromosomes. Most clinics also check whether an embryo is morphologically normal, meaning that its visible physical characteristics align with established developmental benchmarks.

Euploid embryos have a roughly 65 percent chance of resulting in a live birth. Preimplantation tests for aneuploidy, which are more common in the US than in the UK, will therefore push up the success rate when measured per embryo transfer, since a clinic using them will implant only euploid embryos..

Figure 1. Clinic success rates for patients under 35 using their own eggs (2020–2022 from US fertility clinics pooled). Each point represents a U.S. fertility clinic, showing the relationship between the number of IVF cycles performed (log scale) and success rate (percentage of embryo transfers resulting in live births). The lighter (darker) shaded area indicates where we would expect clinics to fall 99% (95%) of the time if the clinic had a “true” underlying success rate of 49%, the average across all cycles The circled region marks the high-performing clinics with both high success rates and substantial cycle volumes.

Source: NASS data provided by the CDC for 2020, 2021, and 2022, own analysis

What about egg survival?

Unfortunately, clinics do not have to report the success of egg freezing itself (how many eggs survive the freezing and thawing process) to central regulators.

Some clinics have peer-reviewed studies that include egg survival post thawing: for example, NYU Langone in US (79 percent survival rate), IVI Valencia (85 percent survival rate) in Spain and Guy’s & St Thomas’ NHS (UK) (74 percent survival rate). Already from these studies we can see variability. However, these are older studies, carried out around the time when vitrification was first being introduced. Furthermore, the women in these studies have a median age of 36-38 and survival of older eggs tends to be lower post freezing and thawing.

The European Society of Human Reproduction and Embryology now believes that 95 to 100 percent of eggs from donors under 35 years old should survive, and labs have long been able to reach this rate when using donor eggs. IVI Valencia reported a 96.7 percent survival rate for donor eggs in 2007. Since clinics pay for eggs from donors, they have an even stronger incentive than usual to maximize the success of vitrification and warming, and so may have optimized these processes. You may wish to use a clinic that also does egg donation.

Freezing eggs versus embryos

If you are in a stable relationship or decided upon the father of your child(ren), consider freezing embryos, as they have slightly higher survival rates post-thaw (95 versus 90 percent). Another advantage is lower variability between clinics and between women. You also reduce the amount of uncertainty once you know how many embryos you have: Although failures are always possible, fewer than five percent of women cannot achieve pregnancy after implanting three euploid embryos.

Endometriosis

Endometriosis is a chronic, very poorly understood disease characterized by the growth of endometrial (uterus-lining-like) tissue outside the uterus, affecting an estimated seven to ten percent of reproductive-aged women. Adenomyosis describes the condition when the endometrial tissue is in the uterine wall.

Endometriosis and adenomyosis have been associated with reduced ovarian reserve and lower egg quality. Given that both endometriosis and adenomyosis can worsen with time, freezing eggs at a young age could be an important pre-emptive measure to counteract an accelerated decline of egg quality with age.

However, women with adenomyosis should be careful about delaying pregnancy because adenomyosis worsens implantation rates independently of egg quality. A retrospective study conducted between 2016 and 2022 on 456 patients shows that the live birth rate after euploid embryo transfer was 25 percent in patients with adenomyosis compared to 47 percent in controls. (The latest studies using donor eggs, which allow for studying live birth rates while controlling for confounders like ovarian response, suggest that the same is not the case for endometriosis.) Considering the above evidence that adenomyosis can progress with age, egg freezing may not provide good insurance against infertility.

Polycystic ovary syndrome

Polycystic ovary syndrome is one of the most common endocrine disorders, affecting approximately 15 percent of women of reproductive age. It is characterized by irregular ovulation or anovulation (no ovulation), hyperandrogenism (higher male hormones), and polycystic ovarian morphology (this refers to arrested follicles visible on an ultrasound). Women with PCOS typically have a higher antral follicle count (AFC) and often produce a larger number of oocytes during stimulation cycles. This makes egg banking more efficient. On the other hand, the risk of ovarian hyperstimulation for women with PCOS is higher. here are also other important caveats: Although egg quantity is typically high, several studies have shown that women with PCOS that have a hyperandrogenic phenotype can experience more pregnancy complications. Optimizing metabolic health prior to implantation, through lifestyle interventions or, in some cases, medications such as metformin, can help.

Distribution of number of eggs collected and fertilized

We also had some fascinating plots that didn’t make it into the final article, drawing on data from the HFEA, the UK’s fertility regulator. These cover the distribution of eggs retrieved per woman across a single cycle, as well as the downstream distributions of fertilised eggs and euploid embryos.

The euploid embryo distribution is particularly striking. As a rough guide, 2 euploid embryos translates to around an 80% chance of a live birth, and 3 is close to a guarantee — only around 5% of women with 3 euploid embryos see no pregnancy after transfer. Among those for whom cycles fail, implantation issues are the likely culprit. Taken together, the data suggest that for most women between 18 and 34, a single cycle is enough to make a live birth achievable.

These distributions are also worth keeping in mind in the context of embryo selection services, such as those offered by Herasight, which are growing in popularity. The more embryos available, the more powerful these tools become, so having a realistic sense of what a single cycle is likely to yield is genuinely useful when deciding how to proceed.

Figure 2. Distribution of the number of eggs collected for women between 18 and 34 undergoing egg or embryo freezing across clinics in the UK. There is considerable variation between women in the number of eggs collected in an egg or embryo freezing cycle. Source: HFEA register data, 2012-2016, ’s analysis.

Figure 3. Distribution of the number of fertilized embryos and expected euploid embryos for women between 18 and 34 undergoing embryo freezing across clinics in the UK. Source: HFEA register data, 2012-2016. Own analysis. Note: To calculate the expected number of euploid embryos, we use the distribution of fertilised eggs and and simulate each egg having a 50% chance of turning into a blastocyst and each blastocyst having a 65% chance of being euploid, i.e. drawing from a binomial distribution.

Whether the treatment itself was curative and how much of an improvement it represents over state-of-the art is not the main focus of this essay. Others have already debated that question at length, and I recommend following their discussions1.

What interests me instead is the bureaucratic absurdity the dog’s owner encountered while trying to pursue the treatment. He described the long and frustrating process required simply to test the drug in his dog: “The red tape was actually harder than the vaccine creation, and I was trying to get an Australian ethics approval and run a dog trial on Rosie. It took me three months, putting two hours aside every single night, just typing the 100 page document.” Even in a small and urgent case, where the owner was fully willing to fund the treatment himself, the effort was slowed by layers of procedure.

Of course, this kind of red tape is not confined to Australia, nor to veterinary medicine. In fact, in the US, the red tape is even worse, at least for in-human trials.

In the United States, GitLab co-founder Sid Sijbrandij found himself in a similar position after the relapse of his osteosarcoma. When the ordinary doors of medicine closed, he entered what he called “founder mode on his cancer.” Like many entrepreneurs confronted with a difficult problem, he began trying to build his own path forward by self-funding his exploration of experimental therapies.

Even then, he ran into the same maze of regulatory and institutional barriers that not only delayed him, but also unnecessarily raised the price of his experimental therapies. These are obstacles that only someone with extraordinary resources could hope to navigate, often by assembling an entire team to deal with them and navigate the opacity. In the end, Sijbrandij prevailed: he has been relapse free since 2025, after doctors had told him he was at the end of his options.

Around the same time, writer Jake Seliger faced a similar situation while battling advanced throat cancer. Like Sid Sijbrandij, he was willing to try anything that might help. The difference was that Seliger was not a billionaire. He could not hire a team to navigate the system on his behalf, and he struggled even to enroll in the clinical trials that might have offered him a chance.

A system originally conceived to safeguard patients has gradually produced a strange and troubling outcome: the mere chance of survival is effectively reserved for the very few who possess the means to assemble an army of experts capable of navigating its labyrinthine procedures.

What makes these stories particularly frustrating is that we already know clinical trials — especially small, early-stage ones like the ones Sijbrandij enrolled in for himself— can be conducted far more cheaply and with far less bureaucracy than is currently required. Ironically, the original article cites Australia as a bad example, yet clinical trials there are conducted 2.5–3× cheaper and faster than in the U.S., at least for human trials, without any increase in safety events—a genuine free lunch.

Removing unnecessary barriers has long been important. That is why I co-founded the Clinical Trial Abundance initiative in 2024, a policy effort aimed at increasing both the number and efficiency of in-human drug trials and have consistently argued about the importance of making this crucial but often neglected part of the drug discovery process more efficient.

Since then, the issue has only become more urgent with the rise of AI. One of the central promises of the AI revolution is that it will accelerate medical progress. Organizations such as the OpenAI Foundation list curing disease as a core goal, and researchers like Dario Amodei of Anthropic have argued that AI could dramatically speed up biomedical innovation. But, as I have written before in response to an interview between Dario and Dwarkesh Patel, AI will not automatically accelerate a key bottleneck in making these dreams a reality: clinical trials. Conyngham’s observation that navigating the red tape to start a trial for his dog took longer than designing the drug itself only underscores the point.

Clinical trials themselves vary widely. At one end are small, bespoke trials involving one or a few patients testing highly experimental therapies—like the treatment in the Australian dog story or the experimental therapy Sijbrandij pursued. At the other end are large-scale trials involving thousands of participants, designed to confirm earlier findings and support regulatory approval.

Different types of trials require different reforms. In this essay, I will focus on the former: small, exploratory trials, which will be called early-stage small n trials for the purpose of this essay. These are often the fastest way to test promising ideas in humans and learn from them. They represent our best chance at a meaningful “right-to-try,” form the top of the funnel that generates proof-of-concept evidence, and may be the only viable path for personalized medicine and treatments for ultra-rare diseases. Understanding why these trials have been made unnecessarily difficult—and how we might change that—is essential if medical innovation is to keep pace with our growing ability to design new therapies.

The real problem: We make early stage trials prohibitively hard

When the story first circulated on X, many people interpreted it as evidence that a cure already exists but simply hasn’t been used due to bureaucracy. That isn’t quite true, as I explained.

The type of mRNA vaccine that the owner pursued looks promising, but he did not know a priori whether it worked or not, as it had not been tested before. So it was not a cure, but “a chance at a cure”. I hesitate to call it an “experimental treatment”, since this term evokes fears of potential safety issues while we generally can predict safety quite well now. The inaccuracy of whether this was a cure or not, however, does not make the story of the bureaucratic red tape that Conyngham encountered any less infuriating. More and more promising treatments are accumulating in the pipeline, fueled by an explosion of new therapeutic modalities, ranging from mRNA to better peptides and more recently, by AI. Yet we are not taking full advantage of them.

To better understand these points, it is helpful to briefly outline the clinical development process—the sequence of in-human trials through which a promising scientific idea is gradually translated into a therapy.

Drug development is often described as a funnel: many ideas enter at the top, but only a few become approved treatments. Early human studies, known as Phase I trials, sit at the entrance of this process. They involve small numbers of patients and are designed to quickly test whether a new therapy is safe and shows early signs of effectiveness.

If the results look promising, the therapy moves to larger and more complex studies, including Phase III trials that enroll large numbers of patients to confirm whether the treatment truly works. Most people gain access to new therapies only after these large randomized trials are completed.

On average, moving from a promising idea to Phase III results takes seven to ten years and costs roughly $1.2 billion. Accelerated approval pathways in areas such as cancer or rare diseases can shorten this timeline by relying on surrogate endpoints, but the process remains slow. As a result, many discoveries that make headlines today will take close to a decade before they become treatments that patients can widely access.

Part of this delay is unavoidable. Observing how a drug affects the human body simply takes time. But much of it is not. Layers of unnecessary bureaucracy, regulatory opacity, and rising trial costs add years to the process without clearly improving patient safety, which is why I started Clinical Trial Abundance.

Allowing a higher volume of small-n early stage trials, the focus of this essay, is a rare “win-win” for both public health and scientific progress. For patients, it transforms a terminal diagnosis from a closed door into a “chance at a cure,” providing legal, supervised access to cutting-edge medicine that currently sits idle in labs. For researchers and society, it unclogs the drug discovery funnel; by lowering the barrier to entry for new ideas, we ensure that the next generation of mRNA, peptide and AI-driven therapies are tested in humans years sooner, ultimately accelerating the arrival of universal cures for everyone.

Next, I will explain why making it easier to run these early stage trials matters.

First, from a patient perspective, they often provide the closest practical equivalent to a right-to-try. In theory, right-to-try laws allow patients with serious illnesses to access treatments that have not yet been confirmed in large randomized Phase III trials. In practice, these pathways rarely function as intended. Pharmaceutical companies are often reluctant to provide experimental drugs outside formal trials, and treatments typically must have already passed Phase I testing. As a result, very few patients gain access through these mechanisms. Early-stage trials offer a more workable alternative. They allow experimental therapies to be tested in structured clinical environments—often in academic settings or academia–industry collaborations—where patients can be monitored and meaningful data can be collected.

Second, early-stage small-n trials are essential for personalized medicine and the treatment of ultra-rare diseases. Many emerging therapies—such as personalized cancer vaccines, gene therapies, and other individualized interventions—do not fit easily into the traditional model of large randomized trials involving thousands of participants. By their nature, these treatments target very small patient populations and often require flexible, adaptive clinical designs.

From a societal perspective, these trials play a crucial learning role. As I argued in my earlier essay Clinic-in-the-Loop, early-stage trials are not simply regulatory checkpoints on the path to approval. They are part of the discovery process itself, creating a feedback loop between laboratory hypotheses and human biology. Later-stage studies, particularly Phase III trials, are designed mainly for validation: they test whether a treatment works under defined conditions and produce the evidence needed for approval.

Early-stage trials, by contrast, are oriented toward learning. Conducted with small patient groups and often using exploratory designs, they allow researchers to observe how a therapy behaves in the human body and how the disease responds. In this way, they close the gap between theory and real-world biology. In the Clinic-in-the-Loop essay, I explain how these trials were crucial to the discovery of Kymriah, the first curative cell therapy for blood cancer.

Lastly, these trials play an important role in maintaining U.S. leadership in biotechnology and, over the long run, in safeguarding biosecurity. In recent years, China has been advancing rapidly in biotechnology, in part because it is easier to run early-stage clinical studies there. These studies, often referred to as Investigator-Initiated Trials (IITs), allow researchers to test new ideas quickly and at relatively low cost.

As a result, more U.S. biotech firms are beginning to move parts of their clinical development to China. A Time magazine headline from May 2025 captures what many industry experts have been warning for years: “The US can’t afford to lose the biotech race with China.” This is no longer a hypothetical concern. Data is backing this fear up: US early-stage funding is deteriorating: dropping from $2.6 billion in Q1 to just $900 million in Q2 2025 — the lowest level in five quarters. If this trend continues, it could gradually shift the center of gravity for biomedical innovation abroad. In the long run, this poses risks not only to U.S. biotech competitiveness but also to biosecurity, much like the earlier offshoring of manufacturing supply chains created strategic vulnerabilities.

In the long run, we may need to rethink the entire sequence of requirements for drug approval, especially as personalized medicine becomes more common. For now, however, it is worth focusing on the unnecessary barriers that limit the expansion of early-stage trials to a larger group of patients.

Regulatory Barriers to Early-Stage Clinical Trials

A true technologist at heart, Sid Sijbrandij documented every medical interaction in extensive health notes, gathered and stored raw results from scans, blood tests, genomic sequencing, and tissue analysis, and assembled a team of concierge doctors, researchers, advisors, and operators to help manage and interpret everything and help him design his own therapies. You can read a summary of his journey here and here.

Although his determination was strong, he encountered numerous obstacles. One of them were the Institutional Review Boards (IRBs), committees responsible for reviewing the ethical aspects of clinical studies. In practice, however, the IRB process significantly delayed his efforts. As Sid explained in a Century of Bio article, they can function as “a ‘vetocracy’ where one member of the board can block treatment based on even the smallest concern.”

This situation is difficult to justify: here was someone with advanced cancer who was willing to self-fund the treatment and accept the risks, yet was still prevented from proceeding. It’s as if the system would rather have you dead than risk one imperfect overly long consent form.

In one of my policy proposals for Institute for Progress, part of the Clinical Trial Abundance initiative, I trace the history of IRBs and examine some of the dysfunctions and unintended consequences of the current system.

IRBs were created to safeguard the ethical treatment of human research participants. Over time, however, many institutional IRBs have drifted away from this core mission. What began as collegial ethical oversight has gradually evolved into a bureaucratic compliance system that prioritizes documentation, procedure, and institutional risk management over substantive ethical judgment.

Because universities typically require researchers to use their own institutional IRB, investigators often have no practical alternative when reviews become arbitrary or excessively slow. At the same time, the true costs of IRB administration are largely hidden within indirect grant funding, insulating IRBs from accountability and obscuring the financial impact of inefficiency. The result is a fragmented oversight landscape that imposes significant administrative burdens and delays on research while offering little evidence that these additional procedural layers meaningfully improve protections for human subjects.

What makes this situation particularly frustrating is that better models already exist. Independent IRBs regularly demonstrate faster review timelines and comparable or lower costs than many institutional IRBs, while holding the highest standards of ethical accreditation. Despite operating under the same federal regulatory framework, these boards compete on service, efficiency, and expertise. Yet investigators at hospitals and academic medical centers are typically prevented from accessing them by internal institutional policies. There is little justification for a system in which academic researchers are effectively locked into a single IRB provider when qualified, federally compliant alternatives are readily available.

One would expect they are more expensive but it turns out… they are not!

While IRBs are one example of frustrating delay in time, there are also regulatory requirements that make experimental therapies unnecessarily costly.

One of the largest cost drivers in early-stage trials comes from manufacturing regulations, specifically the requirements around Chemistry, Manufacturing, and Controls (CMC) and Good Manufacturing Practice (GMP) standards. CMC refers to the detailed process and validation of how a drug is produced, including the materials used, the production process, quality controls, and stability testing. These requirements are essential when a therapy is being produced at scale for widespread use. However, in the context of very small early-stage trials involving only a handful of patients, the same standards are often applied even though the risks and scale are entirely different.

GMP is the regulatory framework that governs how drugs must be manufactured to ensure safety, consistency, and quality. Full GMP compliance involves validated facilities, extensive documentation, batch testing, environmental monitoring, and strict process controls. These requirements make sense for large commercial manufacturing, where thousands or millions of patients may ultimately receive the product. But they are extremely expensive and time-consuming to implement, often adding millions of dollars and many months of delay before a therapy can even be tested in humans.

For small early-stage trials, a more proportionate approach is possible. Some regulatory systems already allow what could be called “GMP-light” manufacturing—production that follows core safety and quality principles but does not require the full set of industrial-scale validation steps. Under such a model, therapies can be manufactured in controlled research facilities using well-understood materials and processes, while still maintaining safeguards against contamination or gross quality failures. The key difference is that the level of documentation and process validation is scaled to the size and exploratory purpose of the trial.

How large could the decreases in cost be?

Experience in Australia shows that a lighter approach to Good Manufacturing Practice (GMP) requirements can substantially reduce costs. Clinical manufacturing conducted under Australia’s framework is roughly 2.5x cheaper for this stage of the process alone. Importantly, Australia has operated under this model for about three decades without any observed increase in safety events. This suggests that a 2.5x reduction should be seen as a conservative lower bound on the potential savings from regulatory reform.

Lower CMC requirements also reduce costs indirectly. By allowing much faster entry into the clinic, they enable biotechnology companies to obtain proof-of-concept earlier. This matters because the financial burn rate of a clinical-stage biotech is extremely high—often around $25 million per year. Delays in reaching proof-of-concept can therefore be existential for smaller firms. In practice, prolonged timelines can exhaust capital before meaningful clinical results are obtained.

For this reason, in recent years many smaller biotechs have shifted early clinical operations from the United States to Australia. This is bad in two important ways. Firstly, it deprives U.S. patients from access to state-of-the art therapies. Secondly, it threatens to weaken the biotech sector in the U.S as a whole. In addition, Australia itself is getting “crowded.” At a population of slightly less than 30 million, there are only so many trials it can support.

The direct 2.5x reduction in manufacturing costs is likely only the beginning. If regulatory reforms were combined with modern technologies for final product quality control — such as improved analytical assays and standardized validation platforms — experts I interviewed suggest that manufacturing costs could plausibly fall by 5–10× overall. Such improvements would not only lower the cost of individual trials but could also significantly expand the number of therapies that make it into clinical testing in the first place.

Lowering costs would effectively expand access to early-stage clinical trials. Today, the high cost of manufacturing and running these trials means that only a small number of patients can participate. As an academic immuno-oncologist working in cell therapy at a prestigious U.S. academic institution —who asked to remain anonymous — told me, this often leads to heartbreaking choices. Because the size of academic grants only allows treatment for a handful of patients, he is forced to decide which patients receive the therapy and which do not.

Solutions

I am currently working with Institute for Progress and OneDaySooner on concrete policy proposals for making early-stage clinical trials faster and cheaper by drafting and implementation frameworks that could realistically be adopted.

This work requires navigating significant opacity in the current regulatory and operational ecosystem. As I have written before here and here, the regulatory process is completely opaque and locked behind trade secret protections — which is a problem in and of itself.

Many of the key processes involved in clinical trial manufacturing and quality control are handled by specialized third-party vendors that operate under strict nondisclosure agreements (NDAs). As a result, even basic information about timelines, costs, and regulatory interpretation is fragmented and difficult to access. Despite these constraints, several promising reform directions are emerging.

Here I summarize some of the most promising proposals:

1. Improve ethics review by allowing investigator choice of IRB.

   Granting academic investigators the freedom to choose among accredited IRBs would better align incentives across the IRB ecosystem, introducing competition, cost transparency, and accountability. This could be achieved with two policy shifts:

   (A) Guaranteeing the right of federally funded investigators to select an external IRB as the IRB of record. This would be coupled with robust non-retaliation rules to protect researchers who choose an external IRB, and the elimination of duplicative internal review once an external IRB issues a determination. By allowing researchers to choose among compliant IRBs, and by making those options more visible, market discipline and accountability would be reintroduced into the oversight process.

   (B) Prohibiting institutions from funding IRB operations through indirect costs. Decoupling IRBs from Facilities and Administrative costs will make prices more transparent, enabling researchers to make more informed choices, and encouraging competition in terms of speed, clarity, and rigor.

   For more details on this, you can check the full proposal here.

2. Implement a notification pathway for early-stage trials.

   A regulatory structure modeled after Australia’s Clinical Trial Notification (CTN) framework offers a concrete example of the kind of policy push that could speed up these types of trials. There, most early-phase trials proceed after approval by a Human Research Ethics Committee (HREC), with notification rather than pre-approval by the regulator. The regulator retains inspection powers and the authority to halt unsafe studies, but does not duplicate the scientific review already conducted by the clinician-scientists and toxicologists embedded in HRECs. The result is that clinical trial sites can begin giving drugs to patients much sooner (about two times faster than in the United States, according to informal interviews with industry leaders).

   In the United States, by contrast, Phase I trials typically require submission of an Investigational New Drug (IND) application to the U.S. Food and Drug Administration before initiation. This dual review — by both an IRB and the federal regulator — creates redundancy that lengthens the feedback loop. A CTN-like model for Phase I trials could preserve safety oversight while shifting scientific and toxicological reviews to accredited, transparently governed IRBs with expanded expertise. The FDA would retain the power to inspect, impose clinical holds, and intervene in high-risk cases, such as for novel gene therapies. But for the majority of small-molecule first-in-human studies, the default could be notification rather than permission.

3. Modify statutory language to enable more flexible GMP requirements.

   Although in the U.S., Phase I investigational drugs are technically exempt from the detailed manufacturing rules in 21 CFR Part 211, the exemption offers far less practical relief than it appears. Part 211 is the section of the U.S. Code of Federal Regulations that specifies the current Good Manufacturing Practice (cGMP) requirements for finished pharmaceutical products. These rules govern how drugs are manufactured at a commercial scale and cover areas such as facility design, equipment validation, environmental controls, quality control systems, batch testing, stability studies, and detailed recordkeeping to ensure consistency and safety across large production runs.

   The regulation (21 CFR §210.2(c)) states that Phase I drugs do not need to comply with Part 211, but it simultaneously requires that they still satisfy the underlying statutory requirement that drugs be manufactured according to cGMP under 21 U.S.C. §351(a)(2)(B) which states that drugs not manufactured under GMP are “adulterated.”Being classified as adulterated carries serious consequences. Because the statute does not clearly specify what level of GMP is appropriate for a small exploratory trial versus a commercial product manufactured at scale and what adulterated, manufacturers typically adopt the safest interpretation and follow nearly the full set of commercial GMP standards.

   In practice, this ambiguity creates a de facto GMP floor, where companies default to the most expensive manufacturing standards simply to avoid regulatory risk. Clarifying or explicitly scaling GMP expectations for early-stage trials would allow proportionate standards. However, a key step is first modifying U.S.C. §351(a)(2)(B).

These would be just the beginnings.

The U.S. can do better than Australia. One option is to relax certain CMC (Chemistry, Manufacturing, and Controls) requirements and shift some verification steps to later in the process, conducting more checks at the end rather than requiring every safeguard upfront. This would reduce early-stage costs and speed up development without necessarily compromising safety.

We should also give patients greater autonomy in choosing their level of risk. In some cases, a manufacturing method might carry a slightly higher risk—say an additional 0.1% probability of an adverse event—but reduce costs by an order of magnitude. For a patient facing a terminal illness, that tradeoff may be entirely rational. A system that rigidly eliminates even small risks can inadvertently deny patients access to therapies that could meaningfully extend or improve their lives.

At the same time, we should invest in technologies that reduce the cost of manufacturing and regulatory compliance. Programs such as Advanced Research Projects Agency for Health (ARPA-H) could play an important role here by funding work to standardize assays, validation processes, and manufacturing protocols. These kinds of efforts often fall into what might be called the “boring public good trap.” They are enabling technologies that benefit the entire ecosystem but are too incremental for academic prestige and too diffuse in payoff for private investors. As a result, they are chronically underfunded despite having large systemic benefits.

Another important reform would be reducing regulatory opacity. The requirements that the U.S. Food and Drug Administration (FDA) applies to manufacturing and trial design are often interpreted through informal guidance, case-by-case feedback, and evolving expectations. This lack of transparency and consistency can distort the market, forcing companies to overbuild processes, hire expensive regulatory consultants, or pursue unnecessarily conservative strategies simply to avoid the risk of rejection. In many cases, the uncertainty created by opaque regulation imposes greater costs than the regulation itself. Greater clarity, standardized guidance, and more predictable decision-making could therefore significantly reduce development costs without weakening safety oversight.

And for more ideas, you can also check Sid Sijbrandij’ list of proposals informed by his own experience, at his website.

In his view, the structure of these proposals, particularly provisions that would treat founders’ voting control in private companies as taxable wealth, reveals an intention to weaken the authority of startup founders and entrepreneurs by forcing them to dilute or surrender control of their companies. Ultimately, Solana concludes that entrepreneurs and wealthy individuals should recognize this dynamic and respond accordingly: by investing more aggressively in media, education, technology, and local political institutions in order to preserve and extend their own capacity to shape the public sphere.

I am sympathetic to Solana’s argument in a few different ways: firstly, on the substance of it, I think that the proposed wealth taxes are bad. Secondly, I also agree that this is not mostly about money itself, but reflects other, more “spiritual” discontents of the public. And lastly, as I have written before, I believe that the old guard of intellectual elites, which largely overlap with the group of elites Solana takes issue with, has failed.

Yet I think Solana is mistaken in an important way. His proposed solution — that tech entrepreneurs should respond by funding more media outlets, educational institutions, and cultural platforms — carries a distinctly utilitarian tone. Of course, outlets and institutions matter. But culture cannot be built purely as a strategic response to political opposition. When institutions are created primarily as instruments in a struggle for influence, they often acquire a faintly artificial quality. They feel “engineered.”

This is the danger of what might be called a kind of cultural kitsch. According to French philosopher Oliver Roy, who has beautifully and cogently identified a crisis of culture in his book “Empire of Norms”, kitsch arises when symbols, traditions, and cultural forms are reproduced without the underlying spirit that once animated them. They mimic the appearance of culture but lack its organic depth. A media company founded merely to “counter the left,” or a university established to win ideological battles, risks becoming precisely this sort of hollow imitation. Cultural kitsch might feel like achieving your goals, but it is not long-lasting.

To travel down the path of cultural kitsch is ultimately self-defeating. In recent years I have come to think increasingly in the language of virtue ethics: that certain things must be done because they are right, and that in the long run what is right tends also to be what is also “useful”. Institutions, like individuals, cannot long survive when they betray the virtues that justify their existence. Intellectual life, in particular, rests on a fragile moral foundation — honesty, seriousness, and a devotion to truth that must remain independent of immediate political or strategic goals.

I remember quite clearly my own years in academia during the COVID pandemic. At the time, I warned colleagues—and wrote publicly—that the politicization of science would inevitably erode the authority of the very institutions they believed they were defending. By subordinating scientific judgment to momentary political needs, they were abandoning a duty owed not merely to their profession but to the public itself.

I was told, more than once, that such compromises were necessary — that certain facts needed to be managed or withheld for the sake of the greater good. Yet events that unfolded later confirmed what should have been obvious from the beginning: once an institution sacrifices the virtue that grounds its legitimacy, the loss of trust that follows is difficult, perhaps impossible, to repair.

The same principle applies to the cultural and intellectual institutions that some now propose to build as a counterweight to existing power. If they are created primarily for narrow, strategic purposes — as instruments designed to achieve predetermined political ends — they will inevitably acquire the hollow quality of artifice.

There is also, I think, a deeper misunderstanding in Solana’s diagnosis of the present moment. The growing hostility toward billionaires cannot be explained solely as a power struggle between political factions, nor can it be dismissed as the product of manipulative “leftist institutions” shaping public opinion. If one wants to seriously engage with this phenomenon, it must first be taken seriously on its own terms.

The California wealth tax, after all, enjoys the support of a majority of voters. Rather than reaching for easy and self-serving explanations about ideological brainwashing, it is more intellectually honest to ask why such policies resonate with so many people in the first place. Understanding that appeal requires looking beyond partisan narratives and examining the deeper moral and social frustrations that animate contemporary politics.

Much of the resentment directed at modern elites, including tech billionaires, has a moral undertone. Many people sense that those who hold great wealth and influence no longer occupy any recognizable civic or ethical role within the broader community. In earlier times, power was expected to carry visible obligations — duties of stewardship, patronage, and public virtue. Today those expectations are far less clearly articulated. As a result, dissatisfaction with the conduct of the powerful often finds expression in the only widely available moral language that remains legitimate in modern public life: equality. I see this as a structural change in modernity, not as some individual’s fault. Elites—whether intellectual, political, or financial—do not consistently embody virtue because societal expectations of their moral character have diminished. This development stems from broader shifts in our public philosophy, which I will examine further.

When I speak of equality here, I do not mean equality in the classical sense of a universal human right. What I have in mind instead is the contemporary impulse toward the equalization of outcomes, the growing discomfort with hierarchy, and the proliferation of oppression narratives that increasingly function as a central moral grammar in Western societies. The redistributive proposals now gaining prominence should be understood as part of this broader moral category of equality.

In a previous essay, I argued that the modern fixation with equality acts as a “consolation prize” in a secular era. As Western societies moved from a religious, “enchanted” worldview to a secular, “disenchanted” one, they lost the cosmic narratives that once gave every person—especially the weak or unsuccessful—a meaningful place in a divinely ordered hierarchy. In the modern world, individuals are understood as autonomous agents responsible for their own outcomes, which places a heavy burden on those who fail or fall behind. The language of oppression and equality, I argue, functions as a moral substitute for the lost religious framework: it preserves the idea that every life has dignity by attributing suffering to unjust systems rather than individual failure.

I want to expand this point by arguing that equality in a disenchanted world serves not only to reassure the “weak,” but also to constrain those who hold power. It offers a moral vocabulary through which we can judge when the powerful are acting improperly. In modern societies, power is often defined in terms of utility: those who generate wealth or exercise influence are regarded as successful or deserving. Yet without a shared conception of the good life, it becomes difficult to explain why such individuals might still be failing in their obligations to the community. In earlier eras, even the most powerful figures were understood to stand under a higher moral order. Kings, for instance, were accountable to God, and authority carried expectations of virtue, restraint, and stewardship.

Modern liberal societies are less comfortable articulating such expectations. Over the past century, political thought, influenced no doubt by the rise of secularism, has increasingly embraced the idea that the state should remain neutral regarding competing visions of the good life. As the philosopher Michael Sandel has argued, freedom came to be defined largely as the ability of individuals to pursue their own preferences without interference. The result is a society that carefully protects rights but grows hesitant to articulate shared ends.

One possible response is a return to a civic form of republicanism — the tradition that animated many of the Founding Fathers — in which the value of citizens is not measured solely by their economic utility but by their participation in a shared common good. Such a vision would require a society willing to articulate what that good consists of, and to speak openly about the virtues and forms of life that sustain it. Whether modern liberal societies still possess the confidence to define such shared ends — and whether disenchantment itself might be, even in part, reversible — remains an open question.

One possible response to this predicament is to look again to the example of the American Founding Fathers as models. They did not approach culture in a narrowly utilitarian way, nor did they see intellectual life merely as a tool for achieving political advantage. Many of them were highly educated and widely read, drawing deeply from classical history, Enlightenment philosophy, law, and theology. Figures such as John Adams, Thomas Jefferson, James Madison, and Alexander Hamilton immersed themselves in ancient authors like Cicero and Thucydides, as well as Enlightenment thinkers such as John Locke and Montesquieu. Their writings — most famously in The Federalist Papers — reflect a sustained engagement with the history of republics and the philosophical foundations of political order.

Just as importantly, the Founders understood that a republic could not survive on institutional design alone. John Adams wrote that “public virtue is the only foundation of republics,” insisting that a healthy political order depended on a genuine devotion to the common good. Benjamin Franklin expressed a similar sentiment when he observed that “only a virtuous people are capable of freedom; as nations become corrupt and vicious, they have more need of masters.” For the Founders, culture, education (which was supposed to instill moral virtues, too), and intellectual life were not instruments to be deployed in the pursuit of power, but essential foundations of a free society.

The ends we pursue now tend to cluster around three dominant aims: safety, utility, and equality. Of these, only equality still gestures toward something like a moral ideal; the others are entirely managerial and materialistic in nature.

Having noticed this, I began to wonder: why equality? Why the persistent fixation on oppression and the moral drama of victim and perpetrator? Why, out of all the possible goods, has this one become so central?

Nietzsche suggested that egalitarianism was the afterglow of Christianity—a moral residue lingering after faith itself had faded. But Christianity bequeathed many values to the modern world: humility, charity, forgiveness, self-sacrifice. Why did equality, in particular, survive as the organizing principle?

It seems to me that equality functions as a kind of consolation prize in a secular age. The shift from an “enchanted” premodern cosmos to a “disenchanted” secular order has undoubtedly expanded individual freedom. Yet in casting off the old metaphysical scaffolding, we have also stripped the so-called “losers” of society of any sense of inherent dignity rooted in a larger cosmic story.

At the same time, secular narratives place an immense weight of responsibility on the individual. In a world without providence and fate, outcomes appear to flow almost entirely from human choice and human systems. Nothing is “meant to be” or cosmically allotted.

The contemporary fixation on oppression narratives—closely allied with the language of equality and often emerging from the same moral circles—can be understood, in part, as a response to this strain. By locating suffering within structures of domination, it externalizes at least some portion of the otherwise crushing responsibility modernity places on the self.

Man shall not live by bread alone

“She appeared at daybreak one rainless morning, at the top of a hill on the road from Quijingue, carrying a wooden cross on her back. She was twenty years old, but she had suffered so much she looked ancient. She was a woman with a broad face, bruised feet, a shapeless body and mouse-colored skin (…)

She had cut all her hair herself after being raped for the fourth time.”

So begins the story of Maria Quadrado in The War of the End of the World, one of Mario Vargas Llosa’s most ambitious and unsettling novels. On her journey through the Brazilian backlands she is violated repeatedly — by a constable, a cowboy, two hunters, and finally a goatherd who had offered her shelter. The first three times she feels only revulsion and prays she will not be left pregnant. But the fourth time something more troubling occurs: she feels pity for her assailant. Horrified by this flicker of compassion, she punishes herself by cutting off her hair, remaking herself into something grotesque, as if to extinguish the last trace of tenderness within her.

Maria Quadrado is not an isolated tragedy in the novel. She is one among many broken figures who gather in the doomed settlement of Canudos. Set in late nineteenth-century Brazil, the novel recounts the rise and annihilation of this community in the arid sertão of the northeast. Around a wandering ascetic known as the Counselor, the destitute and displaced assemble: former slaves, bandits, prostitutes, the mutilated and abandoned — those for whom the promises of the modern republic – secular, rationalizing and impatient with religious fervor – mean nothing.

At the heart of the novel lies the most important question: what does the modern world offer to such people? The republic promises progress, order and rational administration. But these abstractions do not feed the souls of “the wretched refuse” that constitute the citizens of Canudos. In this social order, they would be losers anyway.

The Counselor offers something the Republic cannot. He does not promise comfort or prosperity, but dissolution — the surrender of the isolated self into a transcendent design. And he encourages this dissolution quite explicitly:

“Although he mentioned God and said that it was important for the salvation of a person’s soul that the person destroy his or her own will - a poison that gave everyone the illusion of being a little god who was superior to the other gods round about him – and put in its place the will of the Third person, the one that built, the one that labored, the Industrious Ant, and things of that sort, he spoke of these things in clear language, every word of which they understood”.

In Canudos, the wretched give away their will and in return, they are no longer superfluous. Instead, they inhabit a cosmic narrative, brought on Earth by the Counselor.

The newly established Brazilian Republic interprets Canudos as a threat to order and modernity and it dispatches military expeditions to crush the settlement. Yet again and again, these better-armed and professionally trained forces falter against the ragged defenders of the town. The defeats acquire an almost fantastical quality: a modern state, armed with technique and confidence, undone by a people who possess little but belief.

For a brief season, the annihilation of the isolated will — the surrender of individuality into a single, sacred purpose — seems to generate a force capable of holding even the republic’s army at bay. In losing themselves, the inhabitants of Canudos discover an intensity that they did not know existed.

And yet Canudos ultimately falls. Surrounded, starved, and overwhelmed, it is finally obliterated.

The disenchanted world

The inhabitants of Canudos are not equals in any modern sense. The settlement does not rest on the fiction of symmetry. It is hierarchical, ordered and shaped by charisma and submission.

One of the Counselor’s most impressive powers is precisely this: men who once lived by violence and appetite — the bandit Pajeú among them — bend before him without coercion or the need or persuasion. Their obedience arises as though it were the most natural fact in the world. What he offers to them is simple: in Canudos, each person occupies a place, and that place, however lowly, is intelligible within a sacred whole.

The disenchanted, secular world has killed Canudos for most people. So what is left for them? As a consolation prize, a sterile “equality.”

In a disenchanted world, a world that has lost its connection to transcendence, an obsession with equality is born out of a desperate structural necessity. This is the only remaining safety net preventing the autonomous self from sliding into total dehumanization.

In A Secular Age, the philosopher Charles Taylor offers a remarkable account of just how foreign that land of the past truly is. His ambition is sweeping: to illuminate the shift in the unspoken assumptions that structure our experience of the world, which happened during the transition to modernity. These assumptions lie prior to philosophical reflection. They are, as Taylor puts it, “the way we naively take things to be (…) the construal we live in without ever being aware of it as a construal.”

According to Charles Taylor, modernity is marked not merely by a decline in religious belief, but by a profound transformation in our imaginative horizon. What he calls the transition from an enchanted to a secular world is this shift in how the world is experienced.

In the pre-modern “enchanted” world, the world of “spirits, demons and moral forces”, reality was thick with agency and meaning. The cosmos was not neutral or inert, but constantly impinging upon humans. Human beings were vulnerable to forces beyond them, open to both corruption and elevation by powers that exceeded their control. The boundaries between self and world were permeable.

Taylor describes this as the condition of the porous self: a self that was embedded in a morally charged universe, susceptible to outside influence, and conscious of its dependence within a larger, divinely structured order. That dependence came with a recognition of a certain impotence, but it also conferred belonging. One’s life was situated within a cosmic order not of one’s own making.

Disenchantment dissolves this structure. What disappears is not simply belief in spirits, but the very sense that meaning resides in the fabric of the world. The modern cosmos becomes impersonal and governed by physical laws that can be scientifically deciphered. The locus of meaning is relocated from the world to solely inside the human mind. Thoughts, feelings, aspirations — what Taylor calls “fullness” — are now understood to exist only within human minds.

This gives rise to what Taylor terms the buffered self. The buffered self encounters the world from behind a kind of membrane. It is no longer vulnerable to cosmic forces; nature is no longer enchanted, but inert, available for explanation and manipulation. Even when we acknowledge the limits of our powers as humans, the dominant modern posture assumes that natural obstacles yield, eventually, to technology and science. We do not propitiate unseen powers for a good harvest. Instead, we now design irrigation systems and engineer crops.

The advantages of a buffered self and of living in a disenchanted world are clear: no longer do we feel a sense of helplessness in the face of the powers of nature. No longer do we feel like our very destinies are at the mercy of Gods. The modern, buffered self can control fate itself. We are Gods on Earth.

Yet, in my interpretation, with such immense empowerment comes a crushing burden: that of radical responsibility. Once we anchor meaning within the individual and we understand him to be radically able to improve his condition, there is no longer any refuge to be found in helplessness. The secular mind implicitly understands power and merit as matters of will. The loser is solely responsible for their condition.

This goes the other way too. In earlier times, even the powerful, kings included, understood themselves as subordinate to God. Now, it is one’s talent and will, or agency, if one is to speak in currently popular terms, that brings about their fortune.

There are many other ways, in my mind, in which the disenchanted world makes things hard for the “losers” of this world. One obvious loss is eschatological. In the older, enchanted imagination, earthly failure did not exhaust a life’s meaning. The poor, the humiliated, the obscure could inhabit a story whose resolution lay beyond mortal time. Salvation, not success, was the final measure. An inversion of status, even, was heavily hinted at: “Blessed are the poor in spirit, for theirs is the kingdom of heaven.”

Yet disenchantment also robs “the weak” of dignity in this life. In the enchanted world, society was as it was due to a divinely ordained order, as per Taylor: “A kingdom could only be conceived as grounded in something higher than mere human action in secular time. And beyond that, the life of various associations which made up society, parishes, boroughs, guilds and so on, were interwoven with ritual and worship (...) One could not help but encounter God everywhere”. In this divinely ordered world, one’s station, not matter how limited, was not arbitrary, but was understood to participate in a cosmic design.

Medieval society could thus be imagined as a hierarchy of complementary functions: the clergy praying for all, the lords defending all, the peasants laboring for all. The roles were unequal in honor, yet mutually necessary. Even the humblest life had intelligibility within a larger whole.The buffered, modern self stands in a different landscape. There is no divinely ratified structure into which one is fitted. There is only the immanent frame: a social world understood as the product of human construction, revisable and contingent.

What, then, consoles the weak? Very little. Without transcendence, there is no deferred compensation. Without a sacred order, there is no given place of inherent dignity that one might inhabit in the world. And with radical responsibility comes the realization that some might “deserve” to be the losers of this world.

And more disturbingly, without all these, it becomes difficult to resist the conclusion that some lives contribute less, matter less and perhaps even, are less necessary.

It is in erecting barriers to reaching this conclusion that post World War II modernity’s moral energy gathers. The relentless pursuit of equality, the acute sensitivity to oppression and the refusal to accept hierarchy as natural are not merely fashions. They are attempts to supply, through political means, what religion once provided: an assurance that no life is negligible. In a world where dignity no longer descends from heaven, it must be constructed on earth. Or else the weak are left with nothing at all.

The classical liberal dilemma

We still glimpse fragments of enchantment in modern life. Meaning has not entirely disappeared. In fact, it is now most accessible through work and especially work that presents itself as larger than the individual, as participation in a mission rather than the mere earning of a wage.

Nowhere is this more visible than in start-ups. In such spaces, even the pretense of equality is often suspended in practice. Employees willingly subordinate themselves to a founder, often investing that figure with a quasi-prophetic aura. An entire mythology around the figure is born. "Founder-mode” might be the closest we get in the modern era to a grand hero narrative.

This hierarchy is accepted, even celebrated, because it promises belonging to a grand narrative: building the future and bending the arc of history through innovation. In recent years, companies like Anthropic have inspired precisely this kind of near-religious devotion in their employees.

For the fortunate — the talented and the educated— this becomes a form of salvation. One may dissolve oneself into “the mission”, much like the inhabitants of Canudos.

But such salvation on Earth is rarely available to “the weak”. In modern America, where meritocracy has taken perhaps its most uncompromising form, the psychological weight of absolute personal responsibility grows even heavier. The Statue of Liberty may welcome the “wretched refuse,” but even that promise carries an implicit demand: that they transform themselves, that they succeed. The refuge it offers is also a challenge.

And so the final excuse disappears. For those less endowed by talent or circumstance, for those who chose poorly or were never given real choices at all, possibility itself becomes more of a burden.

The high-IQ classical liberal often fails to perceive the existential strain this creates. He speaks of growth, mobility and opportunity — of the long arc of history bending favorably. In doing so, he resembles the bewildered journalist at the end of The War of the End of the World, who confronts the doomed inhabitants of Canudos with incredulity: “I don’t understand, I don’t understand. What sort of creatures are you all anyway? What are you doing here, why didn’t all of you flee before they had you surrounded? What madness to wait in a rat trap like this for them to come kill you all!”

And the reply, given by the deformed Lion of Natuba, cuts through the incomprehension: “There isn’t anywhere to flee to. That’s what we kept doing before. That’s why we came here. That was the place we fled to.”

There is nowhere left to flee and retreat to. The secular age has dismantled the old sanctuaries that once shielded people from the full weight of responsibility. There is nowhere to hide from the self. And in the absence of a grand narrative, sterile equality is itself a rather pathetic consolation prize.

I would like to thank Patrick Collison for reading early drafts of this essay and other related writings.

During a recent interview, Dwarkesh Patel and the CEO of Anthropic, Dario Amodei, discussed whether clinical trials will remain a meaningful bottleneck for drug development in the age of AI. Patel said that “most clinical trials fail because the drug does not work.” In response, Amodei speculated that as AI models get better at designing drugs, “clinical trials will be much faster … let’s say, they will take one year.”

This is a commonly voiced sentiment, but flawed. The truth is that the most significant barriers to progress today are rarely a lack of intelligence. London has a housing crisis even though the technology to design and construct homes has existed for centuries. The bottleneck in housing is not a lack of knowhow, but rather the weaponization of environmental regulations, planning, and NIMBYism. Much the same is true for clinical trials.

AI models can help design more elegant molecules, in the same way an architect can use AI to design more efficient floor plans, but neither intervention guarantees an efficient use of institutional machinery to make that design in the real world. Even the most promising drug candidates must be tested in human bodies which, in turn, need time to metabolize those drugs and develop side effects. Patients must be recruited and followed over time, and regulators must be satisfied. None of this is easily accelerated with AI.

Although I’m optimistic that AI will design better drug candidates, this alone cannot ensure “therapeutic abundance,” for a few reasons. First, because the history of drug development shows that even when strong preclinical models exist for a condition, like osteoporosis, the high costs needed to move a drug through trials deters investment — especially for chronic diseases requiring large cohorts. And second, because there is a feedback problem between drug development and clinical trials. In order for AI to generate high-quality drug candidates, it must first be trained on rich, human data; especially from early, small-n studies.

Clinical Variables

The Amodei interview conflates two distinct variables: the success rate of a trial (based on the quality of a drug), and the speed of that trial, understood as an operational process.

The first variable — the success rate of a trial — is the probability that a drug candidate will be both efficacious and safe in humans. The current success rate for a drug entering clinical trials is only about ten percent, meaning 90 percent of all drugs fail. Most AI efforts in biology aim to boost this success rate.

The second variable is the speed of data generation — the calendar time required to run an experiment after it has started. A clinical trial is just an experiment in human subjects, and the duration of that experiment is determined by both operational and biological constraints that are largely independent of how confident we are in the drug itself. Recruiting 1000 patients across 10 sites takes time; understanding and satisfying unclear regulatory requirements is onerous and often frustrating; and shipping temperature-sensitive vials to research hospitals across multiple states takes both time and money.

Amodei’s prediction that clinical trials could be done in a single year seems to assume that improving the first variable will also compress the second; but this is not so. Even if AI can help design more effective drugs, timelines will not compress until we solve the operational and regulatory bottlenecks of trials.

Admittedly, there is a tempting counter-argument: If AI does generate better drug candidates, then perhaps clinical trials will cease to be a meaningful bottleneck. If a drug is almost certainly going to work, then trials may become a “formality,” even if, in general, they remain unnecessarily costly and long.¹ This argument is also wrong, but understanding why requires being clear about what clinical trials are actually for.

Trials serve two distinct functions: validation — confirming whether a drug works and is safe — and learning, or generating biological data to refine our understanding of a disease, a compound, and the relationship between the two.

Validation is the primary goal of large-scale Phase III trials, which come later in the process and are typically designed to support regulatory approval. While data from these studies can deepen our understanding of drugs, their main goal is to figure out whether a treatment works under defined conditions. Learning, by contrast, is the dominant aim of early-stage trials. Conducted in smaller patient populations and often using exploratory designs, these studies are not limited to simple “yes or no” outcomes. Instead, they are experiments in the fullest scientific sense: they seek to uncover how a drug behaves in the human body, and how the disease itself responds. As argued in my earlier essay, Clinic-in-the-Loop, this makes these early stage trials active engines of discovery that close the feedback loop between hypothesis and human biology.

For large “validation” trials, is it plausible that their cost will simply cease to matter in a (theoretical) world where AI makes drugs with a high probability of success? I think the answer is no, for a couple reasons.

First, unless we increase the pace and volume of the early-stage “learning trials,” it is unlikely that we will ever approach such a level of certainty in drug discovery. Today, most AI systems in drug development are trained predominantly on in vitro data and animal models. While valuable, these sources only imperfectly capture the complexity of human biology. Without large amounts of high-quality data from actual humans, we should not expect AI to generate predictions that approach near-certainty about trial outcomes.

Second, even if improved modeling could compress early-stage development timelines, every successful drug must still demonstrate benefit on an endpoint; either a clinical endpoint or a surrogate endpoint.

For many diseases, however, the relevant endpoints take a very long time to observe. This is especially true for chronic conditions, which develop and progress over years or decades. The outcomes that matter most — such as disability, organ failure, or death — take a long time to measure in clinical trials. Aging represents the most extreme case. Demonstrating an effect on mortality or durable healthspan would require following large numbers of patients for decades. The resulting trial sizes and durations are enormous, making studies extraordinarily expensive. This scale has been a major deterrent to investment in therapies that target aging directly.

Lastly, the duration of a clinical trial does not merely determine how fast an individual therapy reaches patients. It also shapes which diseases attract serious investment and which do not. In a scenario where AI produces better drug candidates, yet trials remain slow, medicines will become unevenly deployed. In that scenario, capital and innovation will flow toward indications with clear, rapidly measurable endpoints — such as oncology — where trials can be completed relatively quickly. By contrast, fields like aging, where meaningful outcomes take years or decades to observe, will continue to lag unless there is genuine innovation in endpoint development.

Osteoporosis, a progressive bone disease that primarily affects post-menopausal women, illustrates these dynamics well. Firstly, it benefits from an unusually strong preclinical model in the ovariectomized rat (OvX model). Unlike many other chronic diseases, where animal models have poor predictive validity, the OvX model reliably recapitulates post-menopausal bone loss and predicts drug response. This rat model is so good, in fact, that Phase III trials for osteoporosis succeed 83.7 percent of the time, substantially higher than the cross-indication average of roughly 57.8 percent at the same stage.

Given the existence of a good pre-clinical model that allows us to select higher quality candidates and the scale of unmet need in osteoporosis, one might expect it to attract sustained and substantial investment. But instead, the opposite has occurred. Today, only two drug candidates remain in late-stage clinical development for osteoporosis.

The primary reason is that Phase III osteoporosis trials are exceptionally large, long, and expensive to run. The core challenge lies in the endpoint: fracture reduction. Fractures are relatively infrequent events, even in high-risk populations, and they happen unpredictably. To demonstrate that a new therapy meaningfully lowers fracture rates compared with standard of care, trials must wait for enough fracture events to accumulate to produce statistical confidence.

Because the event rate is low and influenced by many factors beyond bone strength — such as fall risk, age, and comorbidities — the signal-to-noise ratio is modest. As a result, Phase III osteoporosis trials typically enroll 10,000–16,000 participants and follow them for three to five years. The sheer scale and duration of these trials push costs to between $500 million and $1 billion. Thus, investment into osteoporosis drugs slowed not because the biology failed or drug candidates lacked promise, but because the cost of proving benefit became prohibitively high.

Osteoporosis is just one example where trial size and costs deter investment. But there is broader empirical evidence in this direction. A 2015 study examining oncology R&D found that hematological cancers — where the FDA accepts short-term surrogate endpoints in roughly 92 percent of approvals, allowing for shorter trials — attracted 112 percent more private R&D investment than solid tumors, where surrogate endpoints are used in only about half of cases. The authors traced this disparity to commercialization timelines. The shorter trials used for the former preserve more of a drug’s effective patent life, improving expected returns and drawing capital. Each one-year reduction in bringing a new therapy to market was estimated to increase R&D investment by between 7 and 23 percent.

If we want AI models to actually accelerate “therapeutic abundance,” then, we must first find ways to speed up these large validation trials. And to design better drugs in the first place, we must find ways to collect in-human data in early-stage “learning” trials much faster, too.

Regulatory Friction

The best way forward is to reduce operational and regulatory friction. AI tools can already help at the margins by automating submission drafting, improving site selection, matching patients more efficiently, and streamlining data workflows. But without deep regulatory reform, this is unlikely to shrink trial timelines or costs at scale.

One regulatory lever we could pull is to implement more high-quality surrogate endpoints. A clinical endpoint directly measures how a patient feels, functions, or survives — such as prevention of stroke or a reduction in fractures. A surrogate endpoint, by contrast, is a measurable biological marker or intermediate outcome that reliably predicts such clinical benefit. Instead of waiting years to observe clinical outcomes, trials that rely on surrogate endpoints can measure signals much earlier.

AI tools can contribute to the development of better surrogate endpoints, such as by identifying promising biomarkers, analyzing cross-trial datasets, and modeling causal relationships between intermediate signals and clinical outcomes. But here, too, technical capability is only part of the story. Institutional reform is likely to be the binding constraint. As my case study of the 12-year effort to qualify bone mineral density (BMD) as an endpoint for osteoporosis trials illustrates, the bottleneck was not scientific capability. Instead, the core barriers to faster progress were fragmented trial data scattered across sponsors, weak funding incentives for what is effectively a public good, and an unnecessarily lengthy and opaque regulatory pathway.²

For AI to generate high-quality candidates — the kind that might, one day, push success rates of drug candidates so high that trials become more of a formality — it also needs rich, dynamic data as input. But remember that such data can only come from trials in people (mice are nice, but most animal results simply do not translate.) This, in turn, creates a feedback loop: better AI models require better trial data, and better trial data requires running trials. The loop is only as fast as its slowest component, the trial itself.

A regulatory structure modeled after Australia’s Clinical Trial Notification (CTN) framework — administered by the Therapeutic Goods Administration — offers a concrete example of the kind of policy push that could speed up these types of trials. There, most early-phase trials proceed after approval by a Human Research Ethics Committee (HREC), with notification rather than pre-approval by the regulator. The regulator retains inspection powers and the authority to halt unsafe studies, but does not duplicate the scientific review already conducted by the clinician-scientists and toxicologists embedded in HRECs. The result is that clinical trial sites can begin giving drugs to patients much sooner (about two times faster than in the United States, according to informal interviews with industry leaders).

In the United States, by contrast, Phase I trials typically require submission of an Investigational New Drug (IND) application to the U.S. Food and Drug Administration before initiation. This dual review — by both an IRB and the federal regulator — creates redundancy that lengthens the feedback loop. A CTN-like model for Phase I trials could preserve safety oversight while shifting scientific and toxicological reviews to accredited, transparently governed IRBs with expanded expertise. The FDA would retain the power to inspect, impose clinical holds, and intervene in high-risk cases, such as for novel gene therapies. But for the majority of small-molecule first-in-human studies, the default could be notification rather than permission.

My criticisms are not meant to imply that AI is irrelevant to trials; that’s certainly not the case. But many of the bottlenecks that determine trial speed and cost are coordination, institutional and regulatory problems, and they cannot be solved by technology alone.

Australia’s system allows early-stage, investigator-led trials to start significantly faster—often twice as fast as in the U.S.—by bypassing the heavy-duty Investigational New Drug (IND) requirements intended for large-scale commercial trials. Australia has operated this way for 30 years with no difference in adverse safety events, yet the U.S. continues to regulate small, bespoke, tightly monitored studies as if they were massive Phase III trials.

It is, frankly, ridiculous that we have left this "Australian advantage" on the table for three decades while our own domestic pipelines have been throttled by administrative inertia. These small-scale, investigator-initiated trials are not just "preliminary" steps; they are the most information-dense experiments we can perform. In Australia, a researcher with a brilliant idea can move from bench to bedside in weeks.

U.S. companies themselves have increasingly started to move their early-stage trials to Australia where possible. So much so that in informal interviews with founders I have heard of the problem of the Australian trial system becoming “too clogged”.

The regulatory burden associated with launching clinical trials in the United States significantly harms American patients—especially those facing life-threatening illnesses such as cancer. Although the U.S. continues to lead the world in biomedical research and academic science, many terminal cancer patients struggle to access cutting-edge, experimental treatments. Instead, they are often forced to wait for full regulatory approval before these innovative therapies become available.

Implementing such changes would not only help current patients, but also future ones. It is perhaps the highest leverage thing we can do from a policy perspective to speed drug discovery. That is because it is precisely faster early-stage trials of this kind that enable something I call “Clinic-in-the-loop”: iterative learning in humans, the most important type of learning in biology.

We know this is crucial because some of the most revolutionary therapeutic breakthrough of the last decade (e.g. CAR-T cell therapy) was the hard-won product of two decades of small-scale, iterative “failed” early-stage trials.

To illustrate why these “information-dense” early-stage trials are the true engines of progress, I will use parts of an essay I wrote a couple of months ago for Asimov Press titled “Clinic-in-the-loop.”

Asimov Press

Clinic-in-the-Loop

For the last several years, I have been trying to understand why biomedical progress, especially in therapeutics, has become less productive despite staggering advances in basic science…

Read more

4 months ago · 94 likes · 9 comments · Ruxandra Teslo and Asimov Press

Clinic-in-the-loop

A couple of months ago, I co-authored an essay with Jack Scannell, arguing that making trials more efficient and informative is essential to breaking Eroom’s Law. Critics of our essay, however, argued that making clinical trials more efficient risks treating biotechnology like a casino.

In their view, making it easier to run clinical trials would risk allowing more potentially harmful drugs to be tested in patients and, instead, biotechnologists should focus on making better drugs that are more likely to gain approval. These critics see Clinical Trial Abundance as accepting the status quo of drug development rather than challenging it.

But this is a misunderstanding.

In fact, Clinical Trial Abundance and better hypotheses for drugs are not merely compatible, but self-reinforcing. Faster testing in the clinic creates a feedback loop: ideas become trials, trials generate rich data (including both successes and failures), these data improve models, and better models inform the next generation of ideas. In this view, the clinic is not an endpoint of discovery but a central component of it.

To understand why clinical abundance is important, we must step outside the prevailing view of clinical testing as a mere “validation step” for scientific ideas. The familiar funnel metaphor of drug discovery, depicting a linear progression from basic science to regulatory approval, reinforces the flawed notion of clinical testing as a passive filter designed to screen pre-existing ideas. While this model is narrowly correct in a regulatory sense, it obscures the clinic’s role as an active engine of discovery.

The reality is that clinical trials rarely just deliver a “yes/no” verdict on a drug’s efficacy. Instead, the history of drug development shows that many successful therapies emerged only after initial versions failed in specific, informative ways. When a trial fails, it provides a unique physiological stress test that reveals exactly where a drug’s design fell short. By collecting data from “failed” trials, we can transform negative results into experimental corrections for the next iteration.

Consider CAR-T cell therapies. Once thought implausible or risky, CAR-T therapies now deliver long-term, treatment-free remissions in cancers where relapse had been almost certain.¹ In pediatric B-cell acute lymphoblastic leukemia (B-ALL) and aggressive B-cell lymphomas, for example, CAR-T has cured patients who, previously, had been given only months to live.

CAR-T therapy works by turning a patient’s own immune cells into living drugs. Doctors collect T cells from the blood, genetically reprogram them to recognize a protein on cancer cells, and reinfuse the modified T-cells into the patient. These engineered cells expand inside the body, move to tumor sites, and destroy malignant cells.

In 2017, the FDA approved Kymriah, the first CAR-T therapy, for children and young adults with relapsed or refractory B-ALL, a cancer of arrested development in which immature blood cells, specifically B cells, multiply out of control while failing to mature into viable immune cells. Relapsed B-ALL is the most severe form of the disease, because it means the cancer has returned after prior therapy. Even with aggressive care, only 10-20 percent of patients with relapsed or refractory B-ALL survived beyond five years.

Against this backdrop, Kymriah received accelerated approval from the FDA based on results from the Phase II ELIANA trial, a global, multicenter study sponsored by Novartis. In ELIANA, 82 percent of treated patients achieved complete remission, and subsequent follow-up analyses revealed that five-year survival rose to approximately 55-60 percent.

ELIANA was not a sudden breakthrough, though. It was, rather, the culmination of nearly two decades of clinical studies. During this period, CAR-T therapies evolved through repeated failure in the clinic, as careful studies of underwhelming results spurred new ideas to correct them. The ELIANA trial was led by investigators at the University of Pennsylvania, a group that had spent years studying CAR-T cells directly in patients well before regulatory approval.

In the mid-2000s, the earliest CAR-T therapies first entered human testing. And they emerged from a fundamental question: is it possible to engineer and redirect the T cell’s innate killing power against malignant cells?

Two well-established biological concepts made this seem plausible. First, T cells are extraordinarily cytotoxic.² However, their natural activation is governed by a “layered permission” system, meaning they cannot recognize targets directly, but must wait for other cells to process and present protein fragments in a precise molecular context. While this evolutionary safeguard keeps us from being attacked by our own immune system, it also provides cancer with many opportunities to evade detection by suppressing these signaling pathways.

To bypass these safeguards, researchers relied on a second insight: the ability of antibodies to bind directly and precisely to proteins on the surface of cells, called antigens. By equipping T cells with a synthetic Chimeric Antigen Receptor (CAR), bioengineers created a functional shortcut that bypassed the need for permission systems. This receptor uses antibody-style recognition to lock onto a cancer cell and is wired directly to CD3ζ, a signaling molecule that triggers the T cell’s internal “kill switch.” The moment the receptor engages its target, it flips the internal switch, activating the cell’s killing program.

In laboratory experiments, these first-generation CAR-Ts were formidable, displaying antigen recognition and potent killing power against tumor cell lines. Yet, this in vitro prowess vanished in patients and did not yield durable clinical responses. Understanding why this happened, though, was not simple. The failure could have been caused by a breakdown in in vivo antigen recognition, poor signaling strength, or other defects that only emerged after the cells were injected into the body.

Progress in understanding why early CAR-T therapies did not live up to their promise came from treating first-in-human trials not simply as therapeutic attempts, but as opportunities to learn. These information-dense studies were conducted throughout the mid- and late-2000s and were relatively small (usually enrolling fewer than ten patients). However, they were designed to be maximally revealing. Researchers used many tools to monitor CAR-T persistence and activity in the body, turning information from a small number of patients into a mechanistic understanding.

One such tool was quantitative PCR (qPCR), a lab method that detects and counts specific DNA sequences, which allowed researchers to measure how many CAR-T cells were in patients’ blood. This showed that CAR-T cells successfully entered the body and were easy to detect after infusion. But the signal quickly faded, suggesting that the cells died off quickly. Other experiments shed light on the problem: CAR-T cells could recognize and eliminate cancer cells in patients — meaning antigen recognition was working — but their functional activity fell over time, suggesting that something in the blood was blocking them.

At this point, the diagnosis of why first-generation CAR-T therapies were failing matched long-standing insights from basic immunology. Decades of research had shown that T cells are not governed by a single on-off switch: signaling through CD3ζ provides only the first activation signal. To keep working, T cells need additional “costimulatory” signals,³ delivered through receptors such as CD28 or 4-1BB. First-generation CARs had been designed to deliver signal one without signal two, which explained their poor performance.

This hypothesis guided the next wave of clinical experiments, which investigated whether adding a costimulatory domain would make CAR-T cells more effective at clearing tumors in vivo.

The field-defining result came from Carl June’s group at the University of Pennsylvania. June and colleagues explored a costimulatory domain called 4-1BB. In a first-in-human study published in 2011, they treated a patient with chronic lymphocytic leukemia using CAR-T cells containing both CD3ζ and 4-1BB. They also administered a dose that was remarkably small by cell therapy standards at that time: just 1.5 × 10⁵ CAR-T cells per kilogram of body weight. (A first-generation CAR-T trial targeting renal cell carcinoma, published in 2013, used a dose more than 100-times higher.)

What followed was unprecedented. The CAR-T cells multiplied more than a thousandfold in patients, at their peak comprising a large fraction of immune cells in the blood. The CAR-T cells also persisted for months. Now, at last, CAR-T cells were a long-lived and self-maintaining immune population. Many cancer patients treated with these second-generation CAR-T therapies have achieved complete remission.

The next step, though, was asking whether the same CAR-T behavior would work in a faster, more aggressive blood cancer, such as B-ALL. In 2013, Carl June’s lab reported striking results in two children with relapsed B-ALL, again showing that the engineered T cells could multiply, persist, and drive cancer into remission.

All of these lessons were built into ELIANA, the study that ultimately supported Kymriah’s approval. Led by Stephan Grupp, who had treated the earliest pediatric patients and worked closely with June, ELIANA translated the early insights into standardized practice. This trial codified chemotherapy given before CAR-T infusion, scaled up cell manufacturing, and measured success using tools like qPCR.

Viewed through this lens, clinical trials are not an alternative to basic science, but rather a mechanism within it that closes a feedback loop. Foundational immunology, antibody engineering, and molecular biology made first-generation CAR-T cells possible in the first place, but early human trials quickly revealed that these designs were incomplete and suggested ways to fix them.

Yet theory alone did not prove this would work; the expansion and persistence observed with 4-1BB–based CAR-T cells came as a genuine surprise even to the therapy’s designers. “It was unexpected,” they reported, “that the very low dose of chimeric antigen receptor T cells that we infused would result in a clinically evident antitumor response.”

This shows why the “casino biotech” critique is flawed. It assumes that experimentation simply reveals a fixed probability of success. But trials can change those probabilities. When clinical testing is understood as part of a continuous feedback system, optimizing trial efficiency is not about accepting failure but about learning fast enough to make success more likely.

The most discovery-rich experiments are often not massive Phase III trials, either, but small, academic, investigator-initiated studies that sit close to the design loop.⁴ These are also the trials most burdened by regulatory, institutional, and manufacturing bottlenecks. And also the trials that adopting an Australia-like model would impact.

Why has this not been done before?

Knowing that Australia has operated this framework safely for 30 years—and that these trials are the highest-leverage instruments in biology—the question becomes: why has this been left on the table? Why is the U.S. biotech engine idling at a red light that doesn’t need to be there?

The answer lies in a fundamental distortion of Washington’s health policy priorities. In D.C., “health policy” is almost exclusively synonymous with demand-side economics: Medicare, drug pricing, and reimbursement. The supply-side—the actual machinery of how we discover and test new biology—is remarkably ignored.

This policy stagnation is a classic “tragedy of the commons” problem. There is no natural, deep-pocketed constituency for early-stage trial reform:

• Big Pharma uses its political capital primarily to protect reimbursement and pricing. For a multi-billion-dollar incumbent, saving a few million dollars or six months on an early-stage trial is a rounding error. Furthermore, their model is often to let others take the early risks and then acquire the survivors once they’ve reached later-stage validation, meaning that they often are not the ones carrying out these trials. By the time a biotech has produced a drug that looks promising, a few millions extra due to early trial costs is nothing.

• Small Biotech would be a more direct beneficiary of these reforms, but these startups rarely have the time, money, or lobbying presence to advocate for systemic change in D.C. They are too busy trying to survive the very bottlenecks we are discussing.

• The Commons. Of course, the primary beneficiaries of these reforms are ultimately The Commons. If more drugs are tested in early-stage trials, all of us get more and better drugs. However, those doing advocacy on behalf of the commons have often had an approach that focuses much more on the distribution of what exists (“demand-side”) than creating abundance (so working at the “supply-side” level).

This is why the emergence of the “Progress” community is so vital. Organizations like and are filling the gap left by this tragedy of the commons. They are doing the “un-incentivized” work of advocating for the structural plumbing of science—the boring, high-leverage regulatory reforms that Big Pharma ignores but that the future of medicine depends on.

We need progress-minded individuals to step into these gaps, supported by institutions that understand that supply-side policy reforms centred around Clinical Trial Abundance are key to the future of medicine. Senator Cassidy’s report is a rare sign that this message is finally reaching the most important echelons and it is an opportunity we cannot afford to waste.

I often say that in drug development, lack of regulatory clarity and consistency can be more damaging than regulation itself. Clear rules, even when stringent, can be understood and navigated. But unclear or constantly shifting expectations can be much harder to efficiently deal with.

This uncertainty has the very pernicious effect of reshaping behavior across the biotech ecosystem in a negative way. Lack of regulatory clarity fosters a culture of defensive decision-making, where companies avoid innovative clinical development strategies and default to more conservative, expensive paths simply because they lack confidence in how the FDA might respond. Over time, this dynamic drives up trial costs, prolongs timelines, and ultimately reduces the number of truly innovative therapies that reach patients.

Unfortunately, a recent FDA decision threatens to make an already existing problem even worse.

Today, the U.S. Food and Drug Administration issued a refusal-to-file letter to Boston-based biotech Moderna regarding its Phase III trial for an mRNA influenza vaccine in adults 50 and older. The letter was signed by Dr. Vinay Prasad, the head of CBER (Centre for Biologics Evaluation and Research)—an unusual move at this stage of the process, suggesting the decision was not routine.

A refusal-to-file (RTF) letter means the FDA has determined that a company’s application is not sufficiently complete or adequate to even begin formal review. It halts the approval process before any full evaluation of safety and efficacy takes place, forcing the sponsor to address the cited deficiencies and resubmit.

The justification offered by the FDA is that Moderna used the wrong comparator in adults over 65. Specifically, the agency argues that because higher-dose or adjuvanted influenza vaccines are recommended for older adults, Moderna’s use of a standard-dose comparator rendered the trial not “adequate and well-controlled.”

If that requirement had been clear from the outset, this would be an unremarkable story about trial design. Companies adapt to clear standards and requirements from regulators all the time. But Moderna’s CEO, Stéphane Bancel, says the FDA had previously signaled that a licensed standard-dose comparator would be acceptable—and has only now changed its position after the trial was completed. If that is accurate, it would mean that roughly $750 million—what the New York Times reports the trial cost—was effectively committed under regulatory assumptions that no longer hold.

Adding to the confusion, entrepreneur Dr. Jing Liang noted on X that publicly available trial data appear to show Moderna conducted analyses involving higher-dose comparators in older adults. If accurate, it would raise further questions about the basis for issuing a refusal-to-file letter in the first place.

No matter which account ultimately proves correct, this episode leaves us with two uncomfortable possibilities. If the FDA changed its regulatory position after years of clinical development, that reinforces a persistent concern about regulatory inconsistency. If, alternatively, Moderna met the agency’s highest level of stated expectations, actually did use a high-dose comparator and still received a refusal-to-file letter, the implications are even more serious, raising questions about transparency and internal coherence at the agency.

In this post, I will focus on the “least bad” scenario—that the FDA revised its stance on a trial design it had previously indicated was acceptable. Even that interpretation carries consequences far beyond a single influenza vaccine. When regulatory expectations shift late in development, companies internalize the lesson and design more defensively, avoid innovative comparators and add layers of redundancy. Over time, this entrenches a culture of safetyism and regulatory aversion. Sponsors begin optimizing not for better science, but for minimizing the chance of procedural surprise. In time, this leads to a ballooning of costs.

The problem with regulatory inconsistency

Even if we assume the “least bad” scenario—that the FDA has indicated a lower-dose comparator would be acceptable and then changed its mind—it still reinforces one of the agency’s most persistent structural problems: regulatory inconsistency and opacity.

Translating science into a clinical product already takes more than a decade and billions of dollars and only about 10% of drug development programs succeed. Most of that time and capital is spent in clinical development—running trials, scaling manufacturing, and interacting with the FDA. Yet over a 10+ year program, a company may have only three to five formal milestone meetings with regulators. Those meetings determine pivotal trial design, endpoints, manufacturing standards, and statistical plans.

Given that formal meetings with regulators are limited and FDA guidance is often broad and caveated, companies therefore make pivotal decisions under real uncertainty. And when the stakes involve hundreds of millions of dollars and years of work, the rational response is caution. This is what Adam Kroetsch meant when he said that “vague regulation breeds safetyism.” Sponsors layer on extra monitoring, avoid novel trial designs, and default to conventional manufacturing processes even when more efficient technologies exist. Each individual decision makes sense. Collectively, they make drug development slower and more expensive.

The slow adoption of risk-based monitoring (RBM) is a clear example of the downsides of regulatory opacity. RBM focuses oversight on the highest-risk elements of a study and could reduce trial costs by up to 30% without compromising safety or data integrity. The FDA has encouraged its use for years. Yet, despite this clear advantage, adoption remains uneven. Surveys show that sponsors hesitate because they worry regulators may not accept RBM in practice—even if official guidance supports it.

The same fear-driven conservatism appears in other areas of trial design. One study found that roughly one-third of data collected in clinical trials may be redundant. Sponsors collect this data due to mistakenly believing this is what regulators might ask for. That redundancy translates directly into higher costs, and longer trial timelines.

It’s trying to solve the problem of regulatory opacity, which has such a negative downstream impact on innovation, that has led me to launch, with OneDaySooner, the openFDA Fund—a centralized repository of past regulatory interactions designed to bring greater transparency to the often opaque FDA decision-making process.

Implications for the future

One might argue that the current decision was motivated by a long-standing skepticism of the current administration towards mRNA vaccines. This wouldn’t make the decision any better, but at least it could give us hope that it’s an isolated incident.

Unfortunately, the problem seems to extend beyond mRNA vaccines. This is not the first late-stage reversal under the current FDA. In November 2025, biotech company uniQure announced that the FDA had reversed its prior position on the company’s gene therapy AMT-130 for Huntington’s disease. Although the agency had previously indicated that Phase 1/2 data—compared against an external natural history control—could support a Biologics License Application (BLA) under the accelerated approval pathway, it now determined that the existing data are insufficient for submission.

The shift came despite promising results showing statistically significant slowing of disease progression and supportive biomarker improvements. CEO Matt Kapusta expressed surprise, noting that the feedback represented a “drastic change” from guidance provided in late 2024.

The current FDA leadership has openly acknowledged growing competitive pressure from China and has put forward a number of thoughtful proposals to modernize the regulatory framework. But reform does not end at changing rules. These rules must also be predictable. Regulatory clarity and consistency are foundational and as important as the quality of the regulations themselves. For the sake of the US biotech ecosystem, I hope that the FDA will reverse course in what seems a trend towards decreased regulatory consistency and clarity.

Over time, I became somewhat obsessed with this problem. I also discovered that most biologists aren’t. That lack of engagement, I think, fuels a kind of magical thinking: the belief that simply doing “more and better science” will reverse the productivity decline. But we have had more and “better” science for decades. Clearly, we need to start thinking about this differently.

On one hand, I think we should be engaging much more seriously with the predictive validity of our preclinical models. Preclinical models—whether cell lines, organoids, animal systems, or computational approaches—are meant to serve as stand-ins for human patients at the earliest stages of research. But we rarely ask, in a systematic way, how well results in these models actually forecast outcomes in human clinical trials. Predictive validity, in this sense, is not about whether a model is biologically elegant or mechanistically informative, but about whether positive or negative findings meaningfully change the probability that an intervention will succeed in the clinic. Taking predictive validity seriously would require treating preclinical models as forecasting tools—explicitly measuring how their outputs correlate with downstream clinical success and failure—and using that evidence to guide model selection, funding priorities, and translational decision-making.

On the other, I believe we should appreciate the importance of in-human testing (and any type of data obtained from whole humans) much more, realize that nothing is going to replace it and focus on improving clinical development itself.

Earlier this year, I learned that Jack Scannell, who coined Eroom’s Law, is similarly frustrated with the endless list of fashionable “solutions” that hand-wave past these fundamental problems. Since he has written at length about how to think about predictive validity, we ended up focusing on the importance of improving clinical trials for a new article in IFP’s new metascience newsletter, .

Macroscience

To Get More Effective Drugs, We Need More Human Trials

Note from Andrew: This week’s piece from Ruxandra Teslo and Jack Scannell helped me understand the barriers to having more and better drugs. Ruxandra is a fellow at Renaissance Philanthropy where she studies how to improve clinical development. Jack is CEO of Etheros Pharmaceuticals Corp…

Read more

5 months ago · 4 likes · 1 comment · Ruxandra Teslo

We think that public debate around improving biopharma productivity often focuses only on the most visible parts of the drug discovery funnel:

Public debates about how to revive productivity in the biopharmaceutical industry tend to be dominated by two camps. Technological optimists usually argue that declining industry outputs relative to investment reflect gaps in biological knowledge, and that advances in basic science will eventually unlock a wave of new therapies. The second camp, which traces its intellectual lineage to libertarian economists, focuses on easing the burden of regulation. In their view, excessive FDA caution has slowed innovation. They propose solutions that largely target regulatory approval: either loosening evidentiary standards or narrowing the FDA’s mandate to focus solely on safety rather than on efficacy.

Both perspectives contain some truth. Yet by focusing on the two visible ends of the drug discovery pipeline, early discovery and final approval, both camps miss the crucial middle: clinical development, where scientific ideas are actually tested in people through clinical trials. This stage is extraordinarily expensive, operationally intricate, and crucially, generates the field’s most consequential evidence. We believe that systematic optimization of this middle stage offers significant untapped leverage and deserves far greater focus.

This piece is also my way of signaling that I’ll be spending much more of my time on this issue—and that a longer list of concrete proposals is coming soon. In the meantime, if you have ideas, especially around regulatory changes or anything that could meaningfully improve clinical trials, I’d love to hear from you.

I think this topic has been relatively ignored compared to its importance, as I argue in the piece:

Clinical Trial Abundance, a framework for scaling and accelerating human trials, stresses the importance of optimizing clinical development. We already have a menu of promising solutions. Increasing regulatory transparency, strengthening clinical trial infrastructure through targeted investment, applying the Australian Phase I model in the US, relaxing excessive Good Manufacturing Practices requirements for early-stage development, and enabling remote and decentralized trials are just a few examples. But many of these ideas remain underdeveloped: the specific policy mechanisms, implementation pathways, and operational models are underspecified and insufficiently advocated for.

For the US, the ideal strategy lies in combining world-class science with highly agile clinical development. Yet clinical development has long been overshadowed by basic research, largely because it is operational, less glamorous, and thus, poorly suited for study within academic frameworks. This persistent asymmetry in attention must be addressed.

We believe one reason clinical development has been systematically underweighted in discussions of how to improve biopharma productivity is cultural. Constructing elegant, internally coherent explanatory narratives is deeply attractive to scientists and academics.

These narratives are intellectually satisfying: they reduce biological complexity to legible mechanisms, assign clear causal roles to specific targets or pathways, and offer the reassuring sense that progress follows from understanding. They are also far easier to communicate, teach, fund, and publish than the messy, probabilistic reality of learning from human trials.

Through multiple waves of technology, from computer-aided drug design, via high-throughput screening, recombinant proteins, and genomics, techno-optimists have overestimated the innovation yield of the hot new thing. Again and again, scientists have placed too much confidence in the power of “biological insights,” or pre-clinical mechanistic foresight. Attention naturally concentrates on the few drugs that succeed, so it is easy to construct post hoc narratives of deliberate design.

Moreover, the biotech ecosystem rewards storytelling. From venture capital pitch decks to internal R&D reviews, a compelling mechanistic narrative makes a program easier to fund and justify.

Yet the empirical record shows that mechanistic foresight provides, at best, rough guidance. Drug discovery is better seen as an iterative design-make-test loop, in which real-world human data repeatedly guide the next cycle of design. Progress may depend less on hitting the best therapeutic hypothesis from the start, and more on generating a broad range of plausible attempts and winnowing them quickly based on clinical feedback. What works survives; what does not is modified or abandoned.

In previous work, we described this dynamic as clinical selection: a process in which the clinic, rather than preclinical mechanistic theory, supplies the decisive information about which interventions genuinely benefit patients. We contrasted this with the familiar “intelligent design” narrative, which imagines a linear march from target identification to rational design to cure.

Many of the most successful drugs did not emerge from deep mechanistic foresight, but from iterative, empirical exploration. The clinic functioned as an evolutionary engine. Anti-TNF drugs failed in their original indication before becoming foundational in autoimmune disease; statins survived only because physicians noticed striking patient responses after the field had largely moved on; and drugs like Avastin and Gleevec accumulated unexpected indications as human studies reshaped both their use and their mechanistic stories over time.

GLP-1 agonists offer a contemporary case in point. The earliest drugs in this class, such as exenatide, were developed for diabetes and aimed primarily at improving glycemic control. Later agents like liraglutide offered better pharmacological characteristics, making weight-loss applications more feasible. Even so, many experts thought that meaningful weight reduction was unattainable, because it required higher doses that caused unacceptable nausea. That side effect was overcome through clinical experimentation: gradual dose escalation markedly improved tolerability, enabling liraglutide’s approval for obesity in 2014.

Once a strong clinical signal existed, investment shifted back to refining the molecules themselves. Through extensive screening and chemical optimization of stability, potency, and half-life, Novo Nordisk developed semaglutide, a more durable agent suitable for weekly dosing. Clinical experimentation in patients without diabetes then delivered another surprise: patients lost far more weight than most experts predicted. At higher doses, semaglutide showed ~12.4% weight loss baseline body weight vs. placebo, a result that had previously been seen as out of reach for pharmaceutical interventions.

On the back of these results, semaglutide became one of the most commercially and clinically successful medicines of the modern era. Ongoing trials continue to reveal additional, unforeseen benefits of GLP-1 agonism, including reductions in cardiovascular events that appear independent of weight loss, as well as improvements in liver disease.

Only in the last couple of years have researchers started to stitch together a more complete mechanistic picture of GLP-1–driven weight loss, integrating evidence from animal studies, neuroimaging, gut–brain signalling, adipose-tissue biology, liver metabolism, and long-duration receptor pharmacology. Crucially, this understanding emerged after, not before, the clinical breakthroughs. And the mechanistic model remains incomplete, while the clinical outcomes are unambiguous, a clear case where human trials revealed the therapeutic potential before mechanistic biology could explain it.

At this point, you might object: “Ruxandra, this sounds like terrible epistemics. You’re a scientist—how can you endorse swapping rigor for folk wisdom? Would you have us return to the Middle Ages and listen to the female intuition of the Village Witch? “

However, unlike the natural sciences, where female intuition is often a misleading guide, the social world is familiar terrain. We’ve been navigating it since childhood, absorbing its particularities long before we had the vocabulary to describe them. So when a paper’s conclusion lands on the other side of the grandma test, a bit of skepticism is wise.

Thanks to the replication crisis in social psychology, people have become far more discerning about published findings and more likely to apply their female intuition. Hence the now-familiar quip: “Findings in social psychology are either trivial or false.” Economics, perhaps because its practitioners apply more statistical rigor and are generally smarter, has been spared some of the skepticism directed at other social sciences. But economists are human like everyone else and crucially their jobs also depend on deriving interesting (and surprising) insights from data.

A recent example of how economics can become enamoured with its own methods, at the expense of revealing the truth, was revealed in the renewed debate over the gender wage gap. Conversations about this topic had long been saturated with layers of unhelpful mythology, largely because the topic is so “politically sensitive”. Depending on who you asked, you’d hear wildly different explanations. Many feminists insisted the gender wage gap was primarily the result of discrimination. Red-pill commentators, on the other hand, would tell you women are just stupid.

Economists, on the other hand, had long held that in developed countries most of the gender wage gap is attributable to motherhood. The career interruptions, reduced hours, and constrained job choices that follow from having children set women back in their careers, often irreversibly. Shedding light on this topic with empirical studies is what earned economist Claudia Goldin the 2023 Nobel Prize in Economics. In this debate they were, in my opinion, the camp closest to the truth.

However, beginning in 2023, a new set of papers using IVF as a causal instrument claimed to overturn this consensus. Many people set aside their female intuition and bought into this. The papers that seemed to contradict the motherhood wage gap findings became a popular topic on twitter and on Substack, with surprisingly few questioning their assumptions and design. There were some notable exceptions, such as Ben Southwood, editor of Works in Progress, and Tyler Cowen, one of the few figures in autism-adjacent circles who consistently recognizes the value of female intuition.

I found the results implausible at the time and raised several objections. I shared them publicly on X and included a long footnote about them in my Works in Progress article, “Fertility on demand”.

In November 2025, a new, impressively designed study, came out which largely settles the debate. It exploits a remarkably clean natural experiment involving women with MRKH, a rare congenital condition in which girls are born without a uterus but otherwise develop normally. Because these women know from adolescence that they will not bear children, the study can cleanly isolate the causal impact of motherhood itself.

At this point, it’s worth stepping back and asking how we ended up entertaining results that failed the grandma test so spectacularly. The answer lies in the methods themselves: what the IVF papers were trying to do, what they actually did, and why the new MRKH study succeeds where they failed. At a more meta level, why we should have always believed the initial studies.

The motherhood wage gap that isn’t

Claudia Goldin is perhaps the name best associated with shedding light on the importance of motherhood for the gender wage gap, a body of work that she won the Nobel Prize in Economics in 2023 for. Goldin demonstrated that the persistent earnings gap between men and women, after accounting for the usual factors, is driven largely by motherhood. Not because mothers suddenly become less capable, and not because employers are uniformly discriminatory, but because modern labor markets reward a very specific kind of time commitment. The highest-paying fields—finance, law, consulting, surgery—do not just pay for skill; they pay for flexibility and availability. They reward the worker who can stay late on short notice, respond to a Sunday-night email, or catch an early-morning flight when needed. These jobs value literal presence: the ability to be responsive and physically present whenever demands arise. Other high-stakes careers like science or intellectual fields rely a lot on accumulated experience. Motherhood often makes this kind of around the clock commitment much harder, and the resulting differences accumulate into a significant wage gap.

One of Goldin’s pivotal studies examined MBA graduates from the University of Chicago to track how the gender earnings gap unfolds across a career. The researchers found that while women initially earned about 95% of what men earned, this ratio fell to 64% by 10–16 years after graduation, almost entirely due to differences in labor supply. Women with children took more time out of the workforce and worked significantly fewer hours, whereas women without children experienced only a small, stable earnings gap relative to men.

Goldin’s work is meticulously documented and grounded in large, detailed datasets. Her conclusions have since been corroborated by numerous studies across countries and cohorts.

But crucially, her claims also pass the female intuition test. If you have spent time around high-powered professionals, you know how much success depends on stamina, availability, and sustained engagement. And if you have spent time around mothers, you know how time intensive it is. I wholly bought into these results and wrote an article, “Fertility on Demand,” for Works in Progress, arguing that one’s thirties, precisely the years when fertility declines, are also the most crucial years for career progression. This creates a real dilemma for women, who find themselves racing against their biological clock during the most critical decade of their careers. I propose expanding women’s fertility window using technology, to give them greater freedom and soften the harsh trade-offs they currently face.

Yet Goldin’s work, and other similar studies, face an important limitation: they can be confounded by selection effects. Women who do not have children may differ systematically from those who do, perhaps in ambition, career orientation, or other unobserved traits. Goldin and her collaborators carefully chose datasets that allowed her to control for many observable characteristics. Yet even with rich data, some differences, especially those tied to preferences or personality, remain impossible to fully control away.

One way to get around selection effects in economics is to rely on a so-called “natural experiment,” a situation in which luck creates something close to random assignment. One natural experiment researchers thought about is IVF success. IVF success is independent of a woman’s ambition. One can consider women who are all actively trying to have a child and exploit the fact that the success or failure of their first IVF attempt is, conditional on medical factors, essentially random. This allows researchers to form two groups of women who are identical in their intentions, ambitions, fertility preferences, and baseline career trajectories—except that one group happens to have a baby and the other does not. The success of the IVF cycle serves as a so-called instrument: a source of exogenous variation that predicts childbirth but is unrelated to unobserved career-relevant traits. By comparing women who succeed at IVF and those who fail, the studies aim to recover the causal effect of having a child on long-run earnings, labor supply, and career progression, free from the selection problems that plagued earlier observational research.

Methodologically, this approach overcomes the core limitation in Goldin’s datasets: she could control for many observable characteristics but could never fully equalize unobservable factors like ambition or personality.

There are three major papers that use this empirical design. While one finds substantial motherhood-related wage penalties, the other two reach a strikingly different conclusion: once IVF successes are compared to IVF failures, most of the so-called “motherhood wage gap” largely disappears. In effect, these studies argue that women who eventually have children do not suffer major long-term earnings losses relative to the counterfactual in which they remained childless. Instead, they suggest that women who become mothers may, on average, have been less oriented toward high-intensity career trajectories to begin with.

To the objection that women’s earnings are roughly equal to men’s in their twenties and open exactly at the time when they have children, which is strong circumstantial evidence for the motherhood gap, those who buy into the IVF studies also have a response. Early-career work is often about fulfilling structured tasks, whereas true differentiation in ambition and grit becomes more visible in one’s thirties. The fact that this period coincides with motherhood is, according to this view, largely incidental.

Yet these conclusions have always seemed dubious to me and I summarized my objections at the time in a long footnote to my “Fertility on Demand” article.

To begin with, the IVF studies do not even agree with each another: one reports substantial motherhood penalties, while the other two conclude that the penalties almost vanish once you compare IVF “successes” with IVF “failures.” Some of these results hinge on mathematical modelling choices, raising questions about their robustness.

But the deeper problem is actually conceptual. These studies treat “women who fail at IVF” as an appropriate equivalent for women who do not have children. That assumption is extremely fragile. Anyone who has worked with, known, or simply listened to women going through IVF understands that treatment failure is not a neutral outcome. It is an emotionally and physically taxing experience, often unfolding over months or years, that can induce depression, financial strain, and marital conflict. IVF failure can meaningfully alter woman’s labor supply, long-term career choices, and capacity to sustain high-intensity professional engagement. You can discover this by talking to women who have failed IVF. Or, ironically, by reading economics papers who prove these statements quantitatively.

In other words, IVF failure is not a clean “no-child” condition. It is its own outcome, with its own psychological, economic, and temporal burdens. This makes it an imperfect and potentially misleading counterfactual. The shock that is supposed to provide exogenous variation may instead introduce new, unobserved differences that bias the estimates.

A second conceptual flaw is that women make career choices in expectation of having children, choosing more flexible and less well-remunerated career paths in advance. Such an observation should, again, be common-sensical to someone who has lived among humans, but there are actual economics papers that attempt to *prove it*. As a consequence, any design that ignores anticipatory sorting, like the IVF papers, treats endogenous career decisions as if they were exogenous, which they are not.

Finally, the people getting overly enthusiastic about the IVF studies often overlook the nuances in the observational work by Goldin and others, which strongly supports the central importance of motherhood in shaping career trajectories. Goldin, for example, follows the same women in high-powered professions before and after having children and shows that female lawyers routinely drop from working more than 50 hours a week to part-time schedules after childbirth. It’s hard to maintain that someone willing to work 50+ hours per week in her twenties was simply “unambitious”. The data instead suggests women have to make hard trade-offs once children arrive.

No empirical design is perfect. But if I must choose between the Goldin-style design and the IVF natural experiment, I would choose Goldin’s. The natural experiment may appear cleaner on paper, but common sense reveals that it introduces more problems than it removes. This is a case where economists may be congratulating themselves for methodological elegance while drifting further from the truth.

Thankfully, in November we finally got a paper that once again demonstrated that most of the gender wage gap is driven by motherhood. This time, in manner that should be accepted even by someone driven entirely by methodology and without any common sense. The paper uses a very clever natural experiment, as highlighted in the abstract below:

The MRKH natural experiment is particularly powerful because it identifies a group of women who are biologically typical in every respect except one: the congenital absence of a uterus. As a result, they are far less likely to have children unless they pursue surrogacy. Women with MRKH experience normal sexual development and typical levels of female hormones, so the condition should not influence personality, ambition, or other traits that might confound analyses. And unlike repeated IVF failure, which introduces the shocks described above, such as depression and physical strain, MRKH creates a genuinely clean “no-child” condition.

It is important to note that the study has its own caveats. One potential concern is that MRKH might have androgenic effects, such as elevated male hormone levels, that could, in principle, influence personality. I examined this possibility extensively, and the evidence consistently shows that women with MRKH have normal ovarian function and typical female hormone profiles. Nonetheless, we cannot completely rule out the possibility that the condition has some other unobserved effect on personality due to biological factors.

Overall, the results make a strong case that motherhood strongly influences the gender wage gap. The fact that women with MRKH look identical in their earnings to the broader female population, until the moment peers begin having children, and then begin outperforming them dramatically in earnings and employment, offers unusually strong causal evidence that motherhood itself drives most of the gender wage gap. There’s a more subtle point that makes these results unusually convincing. As the plot below shows, women actually earn slightly less than men in their 20s, consistent with employers anticipating future childbearing and discriminating based on that. But over the long term, MRKH women earn the same as men, a finding that directly challenges the idea that women’s lower ambition naturally manifests or compounds with age.

The study also shows that the mere expectation of motherhood shapes choices long before any child is born: MRKH women invest more in education, pursue different partners, and develop more progressive gender attitudes because they understand, from puberty onward, that their adult lives will not be constrained by childcare.

An interesting dimension relates to the “Greater Male Variability” (GMV) hypothesis: the idea that men are overrepresented at both the lower and upper tails of many trait distributions. Because the original study was conducted in Sweden, a country with relatively compressed wages, it would be informative to examine MRKH women in the United States, where income dispersion is larger. If GMV holds, one would expect a gender wage gap in the U.S., driven primarily by the highest-earning men.

Yet it is still striking that no such effect appears in Sweden. Even with Sweden’s wage compression, earnings still follow a Pareto-like distribution, which under GMV should still generate a higher male mean. One important caveat is that the sample includes only 152 women, an unavoidable consequence of MRKH’s rarity. Small samples not only increase noise but can also distort the observed distribution if they restrict variance.

Are women born women?

The recent MRKH study has taken even me by surprise. I had always imagined that motherhood would explain a great deal, perhaps the lion’s share, of the disparities we observe in wages between men and women. Yet I also believed that was not the whole story. Surely, I thought, men’s greater appetite for competition and their keen pursuit of status would have produced differences even in the absence of children.

Simone de Beauvoir, writing at the dawn of the last century, declared, “One is not born, but rather becomes, a woman.” Her sentence has since become a banner for the most blank-slatist of feminists. I do not belong to that creed. And yet, the study’s findings compel me to revise my estimates in the direction of social construction. It appears that the mere anticipation, the steady, lifelong knowledge, that one might become a mother shapes a person in ways that not trivial. Expectations laid upon us in youth can cast longer shadows than we imagine.

These reflections return me to Camille Paglia’s observation: “There has been no female Shakespeare because there has been no female Jack the Ripper.” She echoes what is the greater male variability hypothesis and, rather ironically, chose literature as her exemplar—perhaps the field least suited to such a demonstration. For while we may safely concede that history has offered no female Einstein or von Neumann, women have actually reached the highest peaks of writing. The collective uncoscious assigned the role of storyteller to a woman, in the form of Scheherazade, who, as Borges said, “saved herself by her stories”. The very first novelist was a woman. And though my own preference for the greatest writer all times lies with Dostoyevsky, I could not fault anyone who places Middlemarch at the zenith of the novel form. If one prefers experiments of style, the contest between Virginia Woolf and James Joyce for mastery of the twentieth century is, to my mind, a matter of taste. And we should not forget that the best-selling novelist of all time is a woman as well.

It is striking, then, that so many of these women who approached or achieved literary greatness did not have children—and, more curiously still, were often celibate. Jane Austen, the Brontë sisters, Virginia Woolf, Emily Dickinson, George Eliot herself, Harriet Beecher Stowe: none had children. Jane Austen, the Brontë sisters, Emily Dickinson, George Eliot, and Harriet Beecher Stowe all belonged to that company. Virginia Woolf’s marriage, by all accounts, was likely unconsummated; George Eliot’s unconventional appearance and the social judgments it invited shaped the course of her romantic life. Even Mary Shelley produced Frankenstein before she became a mother. Perhaps the raw potential for literary brilliance is equally distributed between the sexes; yet the freedom to cultivate it, uninterrupted, unexhausted, untouched by the practical and emotional claims of motherhood, is not.

Still, I would not go so far as to adopt a full blank-slate view. Motherhood exerts its own gravitational pull, but men and women do appear to diverge in certain native inclinations and appetites. What these findings suggest, however, is that the influence of motherhood and even the expectation of it may be deeper and more pervasive than I had allowed myself to believe.

The latest wave of AI enthusiasm presents itself as the solution to this paradox. I am generally optimistic about the potential of AI to improve drug discovery. But it could just as easily become a net negative if it encourages two kinds of self-deception.

First, we may become so enamored with the sheer volume of new hypotheses it can generate that we forget the limiting factor is not hypothesis quantity but hypothesis quality, or, as Jack Scannell would put it, improving the predictive validity of our models. Generating more mid-PhD-level insights, the biomedical equivalent of “slop”, is not progress. It might be actively detrimental if it leads us to “clog” the clinical development pipeline.

Second, AI advancements may tempt us to overlook the irreplaceable nature of data derived directly from humans, leading us to invest more deeply in model systems that are misaligned with human physiology. We will pat ourselves on the back, happy we have trained a “bigger model” and forget what this was all meant to do in the first place. This has happened before, when, due to more stringent regulations, we switched from fast in-human experimentation (which dominated during the 60s), to the current paradigm dominating drug discovery, so-called “rational design”. More complex mechanistic insight, faster and higher dimensional science coincided with am overall decrease in efficiency of the pharma R&D enterprise, as explained at length in the The Rise and Fall of Modern Medicine.

Becoming captivated by our newest models may cause us to lose sight of a key bottleneck: without regulatory reform that enables easier generation of in-human, ideally interventional data, the evidence needed to truly advance the field will remain limited. I might sound like a broken record by now, but the core objective of the Clinical Trial Abundance project—accelerating and reducing the cost of obtaining human evidence—could be more catalytic than any individual technological breakthrough. Don’t take it from me, take it from Jack Scannell himself.

We do not need more hypotheses, we need better ones

Placing our hopes for future biomedical progress in the vague expectation that “AI will bail us out,” without first rethinking the kinds of data we generate and their impact on predictive validity, is no more rational than assuming that simply training more PhD students will increase pharmaceutical productivity. We have run the experiment of training more scientists and it has not worked out very well!

A common, implicit assumption in the current “AI-for-pharma” boom seems to be that the central bottleneck in drug discovery is the ability to find new drug candidates. The belief is that if AI can propose more molecular structures, explore chemical space faster, or mine literature more efficiently, it will unlock vast therapeutic value. Yet this narrative rests on a misconception, well captured here by Will Manidis.

Pharma is not actually suffering from a shortage of hypotheses or potential drug candidates: they already sit on enormous compound libraries, thousands of unadvanced targets, and a backlog of mechanistic hypotheses derived from decades of molecular biology research. What remains a constraining factor is the ability to test them in humans. Given the high cost of clinical development programs (more than $1 billion on average), pharma must pick their programs judiciously.

We are currently very poor at predicting ex ante which therapeutic hypotheses will succeed in humans. Only about 10% of drug programs ultimately work, meaning we waste billions of dollars pursuing approaches that don’t translate. The core need is to improve our ability to identify hypotheses that are genuinely likely to produce clinical benefit—that is, to increase the predictive validity of our preclinical models.

However, existing AI systems are limited because they are not trained on the kinds of causal, dynamic, multi-scale biological data that would be required to replace or approximate in-human testing. They make use of the same data we have been generating for decades to inform our clearly not very successful strategies of defying Eroom’s Law.

AI has achieved its strongest results in biology in domains where the relationship between inputs and outputs is well-defined and heavily sampled, such as protein structure prediction or de novo antibody design. These areas benefit from large, high-quality datasets and clear optimization objectives, which allow models to learn efficiently. But the availability of these datasets is not accidental: we have good data because these were problems we already knew how to study and measure well. As a result, the applications of AI in these domains, while genuinely useful, mostly accelerate workflows that were already technically feasible. In other words, many current AI-for-biology successes are incremental, not transformational. AlphaFold, an impressive achievement, has not revolutionized drug discovery precisely because, for large classes of proteins, we are quite good at experimentally deriving structures.

This is nicely illustrated by a thought experiment asking what would have happened if Jim Allison (Nobel Prize Laureate for the discovery of immunotherapy) was given modern AI- based generative antibody tools in the 1980s. Likely, this would not have sped up the development of immune checkpoint blockade by more than about a year. The key reason is that the long delay wasn’t due to designing binders: CTLA4 was rapidly licensed and an antibody was patented within months. The slow part was figuring out which immune pathways could be perturbed safely and effectively in humans, and that required years of biological insight, animal models, immunology, and clinical experiments. AI cannot shortcut that at the moment.

However, there are cases where AI applied to protein structure prediction and de novo design directly addresses a genuine bottleneck — although such uses are at the moment the exception rather than the rule. NablaBio is one of my favorite examples of this1.GPCRs are one of the most clinically important target classes, but they have historically been extremely difficult to modulate—especially with biologics—because they are membrane-embedded, structurally dynamic, and adopt multiple signaling states. NablaBio’s platform generates large-scale functional data on GPCR signaling conformations and uses deep generative models to design ligands that selectively bind and stabilize specific receptor states. This enables state-specific modulation of GPCRs, something traditional experimental approaches cannot reliably achieve.

Nonetheless, even in the case of NablaBio, the core challenge it addresses is fundamentally a structural one: identifying and stabilizing specific protein conformations. This is a fine-grained intervention: highly precise, but still operating at the level of individual molecular interactions. What it does not do is resolve the broader biological and clinical context. The ultimate therapeutic outcome still depends on how those molecular perturbations propagate through complex cellular networks, tissues, and the dynamics of the human immune and physiological environment. Without clinical experimentation, we cannot fully predict these emergent effects. The true “Holy Grail” for AI in drug development would be the ability to anticipate such system-level responses with high fidelity.

Unfortunately, this is precisely where AI is the least well positioned to work well, at least at the moment.

We already have cautionary tales about how “more science” has not led to better practical results in medicine. The mid-1940s to 1970s, the so-called Golden Era of drug discovery, were defined by a close coupling of synthetic chemistry and direct human experimentation. Entirely new molecular classes, from antibiotics to antihypertensives, were rapidly trialed in patients. These advances generally did not arise from detailed mechanistic insight, but from iterative, empirical testing. In effect, the clinic itself served as the primary engine of discovery.

Once the obvious chemical scaffolds were exhausted and regulatory demands increased, making the clinical iteration harder, the field moved toward “rational design,” driven by molecular biology, genomics, and high-throughput screening. This approach promised greater precision and efficiency. Yet, despite these scientific advances, our switch to rational design coincided with a decline in pharma productivity. The new models often provided an illusion of progress: their conceptual elegance exceeded their predictive power, diverting researchers toward theoretical frameworks that did not reliably translate into clinical success.

Why? The underlying issue is the dominance of reductionist models. The core challenge in drug development, translating insights from isolated systems into organism-level therapeutic outcomes remains problematic. Across most disease domains, the available data are sparse, heterogeneous, and largely observational. Even the most information-rich resources, such as large-scale multi-omic datasets that combine genomic, transcriptomic, proteomic, metabolomic, and epigenomic layers, offer only static snapshots. They still miss the dynamic feedback loops, nonlinear behaviors, and stochastic processes that define living systems and ultimately determine therapeutic response. That's not say this type of data is useless or that we can’t derive insights from applying AI to it. Just that we must be careful about how we ultimately use it to increase predictive validity.

If we are not careful, AI could amplify the types of mistakes made during the switch to “rational design”. It is easy to produce endless hypotheses from the relatively poorly predictive data we have and fool ourselves with the impression of progress.

A positive example of progress on predictive validity comes from human genetics. Drugs whose mechanisms are supported by genetic variants have roughly a 2–3× higher probability of clinical success. This is because the “experiment” of perturbing a gene has already been run in humans, at physiological scale, over millennia. Genetics is the best example we have of science providing causal, in-human perturbation data without actually running clinical trials.

If AI is to help reverse Eroom’s Law, it must also be paired with a shift toward building and accessing the kinds of data that matter: directly human, mechanistically informative, and tied to real clinical outcomes. This is precisely why I am so keen on making clinical trials faster and more efficient. AI could even help with that (see here for an example of how)!

Making trials cheaper and faster should be pursued not as a replacement to AI, but as its much needed complement. Although I do hope we get catalytic improvements in technology, it is unclear when and if AI will ever replace in-human testing. On the other hand, the benefit from faster and cheaper trials is tangible, clear and will serve as an enabling complement to better AI models. Nothing to lose from pursuing it more aggressively!

AI won’t bail us out of the need for regulatory reform

A common misconception in current debates is that regulatory reform has become less important because AI will (magically?) accelerate drug discovery. The logic of such arguments goes something along the lines of: only 10% of the drugs we design succeed in trials. Clearly, the problem is that we are not good enough at designing drugs. Relaxing approval would just allow non-efficacious drugs on the market.

This view implicitly assumes that regulation mainly affects the final approval stage2. One reason for this misunderstanding is that regulatory reform is often associated with libertarian critiques that focus on lowering approval thresholds3.

In reality, the true bottleneck is earlier: the clinical trial pipeline, where we test hypotheses in humans. These trials take years, cost hundreds of millions to over a billion dollars, and are tightly regulated at every stage. Because this process occurs inside companies and out of public view, it attracts far less attention than the endpoint of approval. But when critics like Jack Scannell call for deregulation, they are largely referring to these phases. This is also where China has reduced barriers and is a large driver of why its biotech sector is accelerating. And it is precisely because of the low success rate of each individual attempt that we want to test more ideas, earlier!

AI can generate vast numbers of candidate mechanisms, targets, and molecules, but without efficient, affordable, and scalable ways to validate them in real patients, we risk producing a flood of low-quality or ungrounded ideas. This is the “slop trap”: more predictions, but no meaningful feedback loop. If we want AI to genuinely improve biomedical progress, we need to reduce the friction of running early and exploratory human studies. Making it easier to test more ideas faster would allow us to identify what actually works, discard what doesn’t, and use that empirical data to iteratively refine AI models.

The simple answer lies in the fact that everything surrounding this topic lies in the world of whispers. You talk to someone working as a practitioner about which regulations are too burdensome and they tell you x and y. You then ask: “ok, could you provide more examples of this? Ideally, would you be willing to give an official quote, so I do not sound like a lunatic that proposes ideas that can kill people? You know, if someone who is a Professor in this field says the risk-benefit profile of regulation X is suboptimal, people are more willing to trust them than they would a random graduate student.” And then… silence.

Nobody wants to actually do that: put their names behind an idea that might have (gasp!) trade-offs, because, usually, it would involve getting their name associated with said idea and risk being further slowed down in their research by whatever admin facility is trying to already squash research under overly burdensome compliance requirements. Their colleagues might look down upon them for arguing against patient safety: they would be committing a breach of medical ethics, which must always be concerned with potential downside and never with potential upside.

You then suggest maybe they can provide you with some useful alternative, in the form of an official source where you can read more about this: “Has *anyone* at all argued about this specific regulatory process, written about this, quantified the risk-benefit profile of it?”. And it usually turns out that no, nobody has. The next step is asking for some direct proof in the form of some receipts. This is also very hard to get: those operating in the world of medicine are again very reluctant to share anything that can be traced back to them. After endless reassurances that no, you are not going to reveal their identity, you will, if you are lucky, get some concrete evidence about why X regulation is bad and how. Maybe some chain of emails about a particularly annoying IRB. Again, I do not blame these friends. In fact, I am very grateful to the few who are willing to share even such information, as they are very rare.

To give a concrete example, a friend of mine who is an academic immuno-oncologist wrote to me that while some of the proposed policies are great, to truly make clinical development faster and more efficient we need more radical ideas. This is something that I felt too, so naturally, I asked him what the more radical ideas are.

He immediately suggested that in his field of cell therapy, small, bespoke phase I trials are encumbered by onerous requirements for full-GMP standards of each reagent. While such standards make sense for larger, industrial scale trials, they are not necessary for the kind of academic trials that he carries out, as careful testing of the final product has to be done anyway in a bespoke fashion, something that is not feasible when manufacturing at large scale.

These requirements apparently decrease the amount of cell therapies that can be tested in small numbers of patients by a factor of around 10x and lead to minimal if not zero decreases in patient safety. We could be testing 10x “cures that work in mice” than we currently are…. That is a big deal. And we could be doing it in people who are suffering from terminal cancer and who would gladly accept a therapy that has 0.0001% more risk associated with it versus nothing. But we don’t.

And since Phase I trials are the funnel to everything else downstream, it means we are severely limiting our capacity to test potentially transforming therapies at a critical step. It is important to see, at least in principle, if an idea works in humans and then, if it does, scale it up with full GMP in larger trials — this is what Phase I is for. But if we only test 1/10th of these ideas, we are just giving ourselves less chances.

This is infuriating to me!

I will quote my friend exactly:

“As an example, testing a new CAR T cell therapy approach in cancer patients requires $2-3 million between procurement of GMP grade DNA ($100,000+ for GMP, $5,000 for research grade with extra testing) and viral vector ($1,000,000+ for GMP, <$50,000 for research grade), and required engineering runs, meaning multiple millions of dollars are needed to test a new idea in 5 patients. The initial proof of principle that transferring T cells into patients could treat disease (in this case a viral infection in immune compromised patients) was done on a standard NIH grant due to less stringent regulation at the time of cell therapies.

With higher regulatory barriers, this proof of principle would never have happened. The expense of early phase clinical trials means that in the absence of venture capital support, most academic oncologists working in cellular therapy are effectively forced to work in less relevant mouse models, as those are the only experiments that are feasible with realistic funding from NIH, DOD, or foundation sources. While foundation and NIH sources of funding exist for trials, they are not nearly on the scale that is necessary. As a scientist developing cell therapies and a doctor who treats patients with these therapies, I believe that 2nd and 3rd generation T cell therapies will transform cancer treatment. There are 100s of ideas for how to make these therapies work better from mouse and in vitro models, but these models are poorly predictive of efficacy in patients, and the path forward is to facilitate the rapid testing of multiple approaches in patients, to get more shots on goal. The patients themselves very much want this also.”

I am very grateful to my friend, as this is a lot of evidence. But still not enough to write a policy memo or convince anyone who does not already agree with you that this should be done.

He also explained to me that while the FDA allows for some flexibility regarding GMP requirements for Phase I trials, in practice someone (?) at his university ends up imposing much stricter rules because… they can. I posted about this on linkedin and was connected to an ex-FDA person who told me that yes, their regulations are actually more flexible than what my friend is forced to implement (albeit still not flexible enough according to my friend). When I asked why they are implemented in the most maximalist and safetyist possible way by local compliance officers at universities, the ex-FDA person shrugged. I then talked to a manufacturing consultant about this and she told me that some manufacturing consultants (but not her) impose overly strict rules that are beyond what the FDA itself requires. “Can you suggest what the FDA could do to change this? Can you tell me how you would change this?”, I then asked. And this is the point at which I got ghosted.

In a desperate move, I said: “F*** this, we just need to release some full regulatory filings”. Since regulatory filings are not open to the public at the moment, this could at least create a common, shared language and we can begin to quantify how these trade-offs and costs stack against each other, how the FDA requirements are implemented in practice and so on, instead of talking about everything behind closed doors. This is how the “Biotech’s lost archive” piece around releasing full CTDs was born. But there is a broader point to be made here.

Small c cancel culture slows down innovation

“This quest may be attempted by the weak with as much hope as the strong. Yet such is oft the course of deeds that move the wheels of the world: small hands do them because they must, while the eyes of the great are elsewhere.”

(Lord of the Rings)

Tolkien’s choice of hobbits as the creatures that ended up destroying The Ring wasn’t accidental. He wanted the agents of history to be the least obvious ones. They are chosen because they are willing to do a quiet, unglamorous job for the right reason, even when their place is at home, chilling out with their families, not in Mordor. Sam and Frodo are not supposed to be Heroes, Aragorn is. But it is the hobbits who ultimately save the world.

In the last year I have come to appreciate “hobbitian courage”. Much has been made in the last couple of years about cancel culture, a real phenomenon whereby people lose their jobs because they express an unpopular opinion about some widely accepted orthodoxy. When people talk about cancel culture, however, they usually refer to people taking unpopular stances in “big”, visible debates. I do not wish to undermine the courage that these people must exhibit in order to do what they do. However, the returns are also outsized: many of those who criticized some crazy idea from the last years, ended up as popular media figures.

But a more prevalent force, and what I have been banging my head against, trying to figure out how to propose ideas to speed up drug development is a more mundane form of cancel culture: let’s call it small-c cancel culture. It’s not the public, spectacular shaming we see on social media. Rather, it’s the steady, private pressure to never risk disapproval, to never put your name to an idea that might be messy or controversial. The quieter, hobbitian form of courage that clinical development reform (or any other hard systems problem) requires is humble: a researcher agreeing to let you cite them, an administrator willing to deviate from an inherited checklist, a policymaker ready to question a default. Small-c cancel culture starves exactly those acts, not through a single blow but through a thousand unspoken signals that say don’t stick your neck out.

I have floated a similar idea in an essay from last year titled: “Intellectual courage as the scarcest resource”. There, I distinguished between two types of courage: the low potential upside courage versus high potential courage. Telling a random biofoid that might leak your name why full GMP is bad and giving her receipts of it is low upside, hobbitian courage. But it’s much needed.

“It seems to me that there are two types of courage: high potential upside courage and low potential upside courage. Going to war or taking part in any form of physical confrontation, at least for the nobility, was a high potential upside endeavour: one would be showered in glory if they succeeded. Yes, there was a lot of risk involved, but also, much to gain. And what was to be gained was tangible and widely reinforced by society: status, something we all crave. It was not some subjective thing dependent on deep conviction. I think we still have that type of courage in our society. It’s rare, indeed, but not impossible to find. And while it does not manifest itself in people going to war, it shows up in those who take risks to pursue ideas with low probability of success but huge upside in terms of status. If you follow me, you probably recognize that I am talking about the world of start-up founders. Quitting your cushy job to pursue the life of a founder requires a lot of courage, with a lot of comfort to lose, but also with the promise of very tangible rewards for success. Our start-up founders are the warriors and kings of old.

What we do lack though is courage that has low potential upside: that is, not much to be gained from it. And I am increasingly convinced this type of courage has always been even more rare than the high potential upside type. It requires one to not only have high risk tolerance, but also ignore one’s self-interest — two traits that are already rare to begin with, and even rarer together. It requires one to be a sort of Joan of Arc of intellectual life.

Being intellectually honest about a topic your academic (or journalistic, or any other type of intellectual) colleagues disagree with falls into the low potential upside courage bracket. There is stuff to be lost but relatively little to be gained. What’s worse, you will most likely not even be awarded the dignity of being openly cancelled: most likely, your career will become a bit shittier with each open disagreement you have, a dreary slog you cannot even wear as a badge of honour. You’ll become that which most ambitious people fear the most: a no-name. And you won’t even be able to tell where this comes from, to point to a culprit. If you are even a tiny bit ambitious the calculus is clear: shut up and agree. You need to be a bit mad to do it”

I recently wrote an article about how the universal newborn vaccination policy for Hepatitis B did not originate, as many of its critics claim, out of a desire to not offend liberal sensibilities and that if Hepatitis B became more widespread due to low vaccination, it could end up affecting people beyond those they call “degenerates”. Now, to be clear, I do not think the children (!) of these so-called “degenerates” deserve to get chronic liver issues because of their parents’ actions (what happened to the whole “luxury beliefs”/Christian morality thing?), but since this was framed explicitly as a matter of responsibility of said “degenerate” parents, I felt like I had to clarify that even assuming one does not care about these children, the disease could end up affecting a larger proportion of the population IF it starts being more common in these groups of people.

The main purpose of the post was not even to argue this is the best policy that could be envisioned in an ideal world: I acknowledged that it might not be “the platonic ideal of a policy”, but given that we know it has reduced rates of Hepatitis B induced liver damage and cancer massively and has no known downside after decades of it being implemented, my principle is: “If it ain’t broken, don’t fix it”. It’s also important to remind everyone that here we are debating whether to vaccinate a newborn or a few months old baby (unless you actually want to advocate for no vaccination whatsoever). Overall, to assume that this difference in a few months might have any negative impact, in the absence of any evidence, seems weird to me.

The other notable thing here is that the US did try a “selective vaccination” policy before 1991, whereby only the newborns of mothers at risk and mothers who tested positive for hepatitis B were vaccinated. For some reason, this did not work very well (maybe miscommunications between hospitals?), and this is why in 1991 CDC recommended universal newborn vaccination. I am not sure exactly why this happened, because I do not fully understand the logistics of US hospitals.

If someone wants to investigate this further and can come up with a better policy that works within the US healthcare system and its logistics and models the Hepatitis B spread arising from it in a credible way, they should do that, instead of screaming about degeneracy on X. I personally think this is a low EV area to focus on, as I cannot possibly imagine how vaccinating your kid at a few months old versus as a newborn could possibly make a difference to the supposed risk from this vaccine, but if someone is passionate about the topic, they should propose a better framework, taking into account all the trade-offs and real life practicalities of how the US healthcare system works. Something to also consider for this hypothetical new policy proposal are the changing demographic patterns versus pre-1991 US, since I keep getting messages like this (and no, I do not think anti-vaxx sentiment is a good reason to be against high-skill immigration from China). On the pro side, I guess we might have better tech to test mothers/data infrastructure to communicate information. Again, things to be researched.

Anyway, because I have been accused of misrepresenting how transmissible Hepatitis B can be, and apparently that’s how the edifice of Science gets destroyed, I have edited my original article to make it clear that spread via saliva in toddlers is not a common occurrence. New section included here in a block quote. You can also see the entire debate here.

What I learned from this experience is that certain institutional failures, which I have written about before, have created a low trust environment. However, they have given rise to nihilists that endlessly criticize the system and who will not build something better: they prefer being mad at degenerates & trolling on twitter. If they cared about public health, they’d discuss the new Florida vaccination rules that will mean that mandatory vaccination for polio and measles (!!) is removed or write the better policy memo I discussed, taking into consideration all the existing real-world constraints and all the trade-offs. Overall, I have been very critical of how public health officials handled Covid in the past, but if the only alternative is being overridden by plagues and crank medicine, then I will pray at the altar of Dr. Fauci.

I will discuss this more at length in a future post, but since I have other things to do (including writing and researching biotech policy) and I do not want to have another sentence taken out of context and used improperly, I cannot do it now. This underscores a deeper problem: people with full-time jobs who know about this stuff do not have the time to endlessly argue against the barrage of bad information for free, only to be “owned” by those with endless time on their hands to take any possible thing out of context. And it’s not just this. Some influencer with a gazillion followers told me I belong in prison because of the tweet in which I explained this is much more transmissible than HIV. She has 810k followers!!!! With this following she can probably live off posting anti-vaxx nonsense all day long. Not sure what the solution to this is: paying people to research this full-time and refute these points?

You might think refuting this stuff is not consequential, but given that now everything related to vaccines is taken to be some conspiracy, and the spread of ideas like “if I vaccinate my dog they might get autism”, it does seem quite consequential to me.

The new part of the article:

Edit: It was claimed that an initial version of this post was misleading people into thinking that Hepatitis B is more transmissible than it is. You can see the back and forth here. I have amended my article to make sure that no misunderstanding might take place. All in all, the premise on X was that this was as hard to transmit as HIV, which is not true. Hepatitis B is a virus that transmits via bodily fluids and is much more likely to affect “at risk” populations that engage in risky behaviour. Nonetheless, it is considered one of the most contagious viruses in its class and, as I show below, can affect ~ large parts of non at risk people.

“What is your body count?” — asks the doctor attending to your birth, looming over you. You are stressed, in pain, and can barely think: but this is an important part of defeating wokeness via “whoreness score” pre-screening for Hepatitis B. Now, some might really, really want to embed incel internet discourse into medical practice and instantiate such a reality (would they get the whoreness score qualified as a biomarker under the Biomarker Qualification Program?)

Fine. But even then, I want to explain to these people that this might be bad for THEM, since uncontained Hepatitis B can be bad in the long-run for everyone, not just high-risk populations.

While hepatitis B infections may be concentrated in certain high-risk populations NOW (or those high in degeneracy score), due to widespread vaccination, the virus may NOT stay confined to those groups in the long run. In the end, hepatitis B, given its high rate of infectiousness, might not be something that can be kept neatly contained within a few groups we label as “high risk” if it becomes prevalent enough in the “high risk” groups. And if your “high risk” population is large enough, which in the U.S. it might very well be, it might become an actual widespread problem. Viruses don’t respect categories. They move in quiet, ordinary ways: through families, through partners, through the accidents and medical encounters that touch all of our lives.

Before the universal vaccination recommendations in 1991, a study from 1986 found that 1.5% of white heterosexual students with less than 3 recent sexual partners were Hepatitis B infected. Given that college attendance rates were even lower in 1986 than now, this is a quite highly selected group. In a larger cohort, including those from lower socioeconomic backgrounds, 30% of those with Hepatitis B infection had no known risk factor. In Communist China, a country hardly known for its degeneracy, Hepatitis B used to be endemic, with 10% of the population being infected. This led to the government mandating vaccination at birth. It is hard to estimate exactly how Hepatitis B prevalence would look like in the US today if mothers were tested at birth before vaccinating: changing demographics, transmissibility of the virus, healthcare system logistics and so on, would have to be taken into consideration to propose a better policy. But none of the critics of the current system, which works, is doing this exercise.

A Chinese follower of mine, Dr Dale Wen, who is hardly woke (look at her tweets!), summarized her experience in a tweet copied below:

“And she is totally wrong scientifically. Hepatitis B is highly highly transmissible, at least before the vaccine. In 1980s, about 10% of Chinese were carriers. Even though drug use and prostitution were almost unexistent under the communists. My father died from Hepatitis B at age 53. He got it by eating with my uncle, a known patient. When he got diagnosed, my whole family got tested and everyone of us tested positive for antibodies. Meaning we all got exposed but fortunately our immune system fought it off successfully . But my father was not so lucky. There are also academic papers showing toddlers transmit hepatitis B to each other, by sharing saliva. It is the vaccine that changed the tragic situation in China. Today only 3 % Chinese are carriers, despite the fact there are much more drug use and prostitution in China today than in 1980s. Hepatitis B is only transmissable through sex and drug use is a lie pushed by patient activists, because they don't want to be discriminated. Fortunately it has largely become true because most people are vaccinated. If most people are no longer vaccinated, it will no longer be true, as the case in 1980s China.”

She also provided some links to papers showing how easily transmissible hepatitis B can be: one study found a case of transmission between toddlers biting each other in daycare, and another one found horizontal transmission within a family with everyone getting hepatitis B. Because I have been accused of misrepresenting these studies, I want to make it clear these are not the most common ways of spreading the virus: risk increases massively for those who engage in high levels of sexual activity with many partners and so on. But it’s much more contagious than HIV and we probably do not fully know all the ways in which it can spread.

As an anecdotal story, I know of a family in Romania (where baseline rates were higher in the 90s) who was upper middle class and where everyone ended up with Hepatitis B. One of the parents died in his 50s due to cirrhosis.

“I am a great negative. I protest. I have the right to say that I do not accept anything.”

(Kirillov - The Possessed)

In Dostoyevsky’s The Possessed, a group of young radicals embrace destruction as an end in itself: “We shall burn everything, we shall destroy everything, we shall let the cry resound: From the ashes we shall build!” But there is no building that follows, only smoke and ruin. In a recent debate with Curtis Yarvin, I argued that his ideology has seeded the same type of nihilism: destroying The Cathedral is not enough. Among rising rates of mental health issues in the youth and a general sense of ennui and lack of purpose that I can sense in my generation, a constructive path forward that avoids devolving into Third Worldism must be carved: a reinvigoration of classical liberalism, I hope.

My points were perfectly illustrated yesterday, when I found out from some of the luminaries of our age that widespread newborn vaccination against Hepatitis B happens because of wokeness. That sounded weird to me, since as a Romanian newborn, I was also vaccinated against Hepatitis B, regardless of my mom’s Hepatitis B status. The policy had been in place since 1995, at a time when Romania was definitely not “woke” (homosexuality, for example, was still illegal up until 2001 in Romania). Could there be another explanation for this state of affairs other than a refusal from health agencies to construct a “whoreness/degeneracy score” and apply it to women before vaccinating their newborns? Or, as Curtis Yarvin put it in a now deleted tweet, a refusal to do some sort of racial screening of mothers?

This situation shows how the illiberal right tends to be innately reactive to whatever happens, as opposed to constructive. They have no agency, they are merely victims (so Nietzschean!): if the “wokes” do something, then the normal response is to become super anti that thing and produce nothing of value instead. Irrationality is defended, as those trapped in a permanent Victim state cannot be expected to use their mental faculties to judge situations on a case by case basis. Vaccination is also a particularly interesting issue because it’s in some ways a crowning achievement of the Enlightenment’s emphasis on science, rationality, and the application of knowledge to improve human welfare. I guess those advocating for a RETVRN to pre-liberal values really mean it and strongly so.

When Edward Jenner introduced the first smallpox vaccine in 1796, he faced widespread resistance. His method involved inoculating people with material from cowpox lesions, and detractors spread rumours that those who received the vaccine would literally turn into cows or grow bovine features. The reluctance to get vaccinated from people of the day is caricatured in drawings like the one below.

Now, getting to the actual debate: we do know that in many other European countries testing for Hepatitis B of the mother is done before vaccinating newborns. However, this does not require any form of racial or “whoreness” screening: you just test every mother and then only vaccinate the babies of those who test positive. There is little connection to wokeness in this approach. But one might still wonder why in the United States all infants are vaccinated.

Having studied medical regulation and written about it, my bet was on some sort of random idiosyncrasy in the system. Medical regulations and systems are often slow and contradictory, not because of woke conspiracies, but rather because bureaucratic processes tend to accrete inconsistencies over time. This is actually a big problem for biotech, especially when it comes to regulatory inconsistency/opacity, which can cause hundreds of millions of dollars lost in revenue, for example if a drug approval is delayed due to manufacturing issues.

I have written about potential fixes to these inconsistency issues (for example, in my piece on how to efficaciously achieve FDA transparency). I wish people spent as much time digging out inconsistent FDA regulations/criteria (you can start with this white paper here) that make the approval of new therapies slow and onerous, as they spend spotting supposed ideological plots in the vaccination system. Oh, how I wish armies of people dedicated themselves to analyzing these much more important inconsistencies to own the libs! But this would, again, require some Nietzchean vitalism from those most obsessed with Nietzsche, some actual desire to constructively and positively impact the healthcare system.

I also reasoned that there is probably no way of really knowing what the ideal schedule for vaccinating against Hepatitis B is and since the system works as it is we might just as well stick to it. But in this case, it turns out there are actually potentially good reasons why universal vaccination for hepatitis B happens.

The core issue is that perinatal hepatitis B infection is uniquely dangerous: up to 90% of infected newborns go on to get chronic infection, which carries high lifetime risks of cirrhosis, liver failure, and hepatocellular carcinoma. Preventing that single moment of transmission at birth can avert decades of downstream suffering and cost. So not vaccinating has a very high potential downside risk and pretty surely no upside potential.

The United States initially tried risk-based approaches, but experience showed that relying on maternal testing and selective prophylaxis left too many babies unprotected. In 1991, the Advisory Committee on Immunization Practices (ACIP) issued its first recommendation for universal hepatitis B vaccination of all infants. In 2005, ACIP went further, recommending that every medically stable newborn receive a hepatitis B vaccine dose within 24 hours of birth, regardless of maternal status. Basically, it was concluded that it was just easier to recommend universal vaccination for everyone. That this is not the result of some refusal to condemn “degeneracy” is clearly showcased by the fact that the 1991 recommendations were supported by both parties with George W Bush Sr for example being very pro-immunization. I am pretty sure Republicans are not a party that has historically been afraid to judge “degeneracy”; on the other hand, pretty much everyone whom I consider a decent human being would agree that it’s better for children to not get liver cancer, regardless of what their parents did. Senator Bill Cassidy himself, a Republican and liver doctor, and someone who has taken socially conservative stances throughout his career, finally declared himself against this entire anti-vaxx non-sense:

This safety-net logic is especially important in the fragmented American health system. Prenatal care, hospital delivery, pediatrics, and public health reporting are often handled by entirely different providers, with no single national electronic health record or centralized tracking system. A woman might be screened for hepatitis B in one clinic, deliver at a different hospital, and bring her newborn to yet another pediatrician. Each handoff is an opportunity for data to be lost or delayed. In countries with unified national health services, selective strategies can sometimes work more reliably. But even in such countries, this is not entirely efficacious: in 2020, among the 25 EU countries with universal antenatal screening policies, only 77% had achieved ≥90% coverage among pregnant women.

Some argue you could simply test mothers right before birth and only vaccinate the babies who need it. In fact, in the U.S. this is recommended if a mother’s hepatitis B status is unknown at delivery: hospitals can run “STAT” tests with results in a few hours. But in practice, this doesn’t fully solve the problem. Testing at the moment of labor is vulnerable to delays in turnaround time, limited lab availability in smaller hospitals or rural clinics, and false negatives if a mother has only recently been infected. Birth itself is also a chaotic, high-pressure moment, where adding another test-result-communication-action chain increases the chance of error. There is also the chance of false negatives. And if results come back too late, you miss the critical 12-hour window in the vaccine is maximally effective.

That is why countries from Romania to the United States adopted a universal birth-dose strategy. It acts as a safety net: every newborn is protected immediately, regardless of maternal records or testing logistics. Now, I am not entirely sure if this is the platonic ideal of a vaccination strategy, but it works pretty damn well and in any case, it does not happen due to wokeness.

Ok but what if I deliberately want to get a whoreness score assigned to every woman at birth?

Edit: It was claimed that an initial version of this post was misleading people into thinking that Hepatitis B is more transmissible than it is. You can see the back and forth here. I have amended my article to make sure that no misunderstanding might take place. All in all, the premise on X was that this was as hard to transmit as HIV, which is not true. Hepatitis B is a virus that transmits via bodily fluids and is much more likely to affect “at risk” populations that engage in risky behaviour. Nonetheless, it is considered one of the most contagious viruses in its class and, as I show below, can affect ~ large parts of non at risk people.

“What is your body count?” — asks the doctor attending to your birth, looming over you. You are stressed, in pain, and can barely think: but this is an important part of defeating wokeness via “whoreness score” pre-screening for Hepatitis B. Now, some might really, really want to embed incel internet discourse into medical practice and instantiate such a reality (would they get the whoreness score qualified as a biomarker under the Biomarker Qualification Program?)

Fine. But even then, I want to explain to these people that this might be bad for THEM, since uncontained Hepatitis B can be bad in the long-run for everyone, not just high-risk populations.

While hepatitis B infections may be concentrated in certain high-risk populations NOW (or those high in degeneracy score), due to widespread vaccination, the virus may NOT stay confined to those groups in the long run. In the end, hepatitis B, given its high rate of infectiousness, might not be something that can be kept neatly contained within a few groups we label as “high risk” if it becomes prevalent enough in the “high risk” groups. And if your “high risk” population is large enough, which in the U.S. it might very well be, it might become an actual widespread problem. Viruses don’t respect categories. They move in quiet, ordinary ways: through families, through partners, through the accidents and medical encounters that touch all of our lives.

Before the universal vaccination recommendations in 1991, a study from 1986 found that 1.5% of white heterosexual students with less than 3 recent sexual partners were Hepatitis B infected. Given that college attendance rates were even lower in 1986 than now, this is a quite highly selected group. In a larger cohort, including those from lower socioeconomic backgrounds, 30% of those with Hepatitis B infection had no known risk factor. In Communist China, a country hardly known for its degeneracy, Hepatitis B used to be endemic, with 10% of the population being infected. This led to the government mandating vaccination at birth. It is hard to estimate exactly how Hepatitis B prevalence would look like in the US today if mothers were tested at birth before vaccinating: changing demographics, transmissibility of the virus, healthcare system logistics and so on, would have to be taken into consideration to propose a better policy. But none of the critics of the current system, which works, is doing this exercise.

A Chinese follower of mine, Dr Dale Wen, who is hardly woke (look at her tweets!), summarized her experience in a tweet copied below.

And she is totally wrong scientifically. Hepatitis B is highly highly transmissible, at least before the vaccine. In 1980s, about 10% of Chinese were carriers. Even though drug use and prostitution were almost unexistent under the communists. My father died from Hepatitis B at age 53. He got it by eating with my uncle, a known patient. When he got diagnosed, my whole family got tested and everyone of us tested positive for antibodies. Meaning we all got exposed but fortunately our immune system fought it off successfully . But my father was not so lucky. There are also academic papers showing toddlers transmit hepatitis B to each other, by sharing saliva. It is the vaccine that changed the tragic situation in China. Today only 3 % Chinese are carriers, despite the fact there are much more drug use and prostitution in China today than in 1980s. Hepatitis B is only transmissable through sex and drug use is a lie pushed by patient activists, because they don't want to be discriminated. Fortunately it has largely become true because most people are vaccinated. If most people are no longer vaccinated, it will no longer be true, as the case in 1980s China.

She also provided some links to papers showing how easily transmissible hepatitis B can be: one study found a case of transmission between toddlers biting each other in daycare, and another one found horizontal transmission within a family with everyone getting hepatitis B. Because I have been accused of misrepresenting these studies, I want to make it clear these are not the most common ways of spreading the virus: risk increases massively for those who engage in high levels of sexual activity with many partners and so on. But it’s much more contagious than HIV and we probably do not fully know all the ways in which it can spread.

As an anecdotal story, I know of a family in Romania (where baseline rates were higher in the 90s) who was upper middle class and where everyone ended up with Hepatitis B. One of the parents died in his 50s due to cirrhosis.

Overall, if something ain’t broke, don’t fix it. If someone has a concrete policy proposal, that takes into account all the caveats of changing what we know has massively reduced liver cancer in children and protects everyone from possible endemicity, they should write up a detailed memo, instead of screaming about degeneracy on X or justifying broad anti-vaxx sentiments.

I believed I might be infertile because, at sixteen, I’d been diagnosed with the unaptly named “Polycystic Ovary Syndrome” (PCOS). No one had ever explicitly said I was infertile, but the implication hung in the air, and I accepted it as fact. Still, at the time, I told myself it didn’t matter — who wanted children anyway?

Around the time I turned 25, as career oriented as I was, the idea of never being able to have children started to bother me. I did not want children then, but it seemed like something I would want later. Infertility has been a harrowing spectre for women since biblical times. Hannah wept her eyes out and Rachel let her sister sleep with her husband in exchange for mandrakes that would make her fertile. I did not have a sister, nor a husband for that matter, and I could probably get mandrakes if I wanted to, but that seemed like it would not help. So I went to a doctor in the UK — a private one at that. She was pretty much useless and looked at me incredulousy when I explained that no, I was not trying to conceive, but I wanted to prepare for the future and understand what impact PCOS might have on my fertility. She did not seem to get why I was paying her 300£ for such an unserious matter and said I should come back when I am actually trying for a baby, because nothing is guaranteed. I asked her about odds — but she refused to discuss them: the concept of probabilities seemed to elude her. The interaction was so annoying that at that point I would have let my sister sleep with my husband, had I had either a sister or a husband.

So then I turned to the next best thing, which is “doing research on my own”. After all, I am a biologist, so I reasoned that would help. I went into a deep, obsessive dive into what it actually meant to have PCOS and then into fertility itself. I came away with 4 main conclusions:

1. Fertility science was fucked: underfunded, understudied, messy, confounded.

2. Women are gravely misinformed about their own fertility and the existing possibilities for preserving it at a mass scale.

3. PCOS was not really that big of a deal.

4. Egg aging on the other hand was a much bigger deal than PCOS. But egg freezing works and it’s something I should probably do.

This is actually how this blog was born: some of my first pieces were on the efficacy of egg freezing and the causes of female fertility decline. I have also written a longer article for on the impact that having children has on women’s career and life choices and the fact that being able to extend the female fertility window would be a really big deal for expanding women’s freedom. The piece struck a chord with many. At the core of it might be the following plot, which shows something fundamental about female fertility decline: it is mainly determined by the egg and not the uterus. When older women use frozen and thawed young donor eggs, they have ~ the same success rates as younger women. This is why egg freezing works in the first place: you are becoming your own “donor” for your future, older self.

This is the first in a series of posts documenting my egg freezing journey. First and foremost, I want to demystify the process for women and raise awareness around fertility. But this is also a personal topic, one that gets at the core at what it means to be a woman and a biologist — or, as someone has called me, a “biofoid”. So this is also intended as a personal story, a diary of some sorts of what dealing with one’s fertility means for a woman and how harrowing it can feel to navigate the space of IVF, even when one is equipped with biological knowledge. For a more streamlined version of “How to do it” for the practical minded, I will be writing an article for on egg freezing.

I believe it’s natural that the first piece should start from the beginning: why I thought I was infertile, how I discovered more about PCOS and the fact that no, it does not make one infertile, as well as the very first step in the egg freezing journey itself: the transvaginal ultrasound to get my antral follicle count (AFC)1, an approximation of how many eggs I could hope to get from the process, which in turn is highly correlated with further success.

The misconceptions and contradictions around PCOS are key here and should, I think, raise the alarm bells for people. “What the hell are we doing here?” was my reaction to reading a lot of fertility-related literature. “Why do we not know this basic fact?” was a constant reaction as I was reading on various topics. Maybe some of that sentiment and annoyance can be transmitted by writing.

The monster within my ovaries

Polycystic ovary syndrome… The name itself brings to mind something ominous. My ovaries had cysts... Some monstrous outgrowths were emerging from within my ovary, canibalizing it, crowding it, killing it from inside, like Saturn eating its own son. I was not able to produce kids, my future kids were cysts, I was barren, sterile, an abomination of nature.

At least that was the imagery I had in my mind, when at 16, I was told I had PolyCYSTIC ovary syndrome. But maybe it was also sort of fine, because at the time I was reading The Second Sex by Simone de Beauvoir: “One is not born a woman, but becomes one”. Maybe the abominations growing inside my ovaries would retard this journey to becoming a full woman, the second, lesser sex; maybe they would spare me from this unwanted fate and I could forever remain a semi-androgynous creature. I was not going to be the first sex, but somewhere between first and second seemed better than second. Maybe…

I got my first period at 14, which is apparently normal by historical standards, but was very late compared to others in my cohort. And the periods themselves took place every 3 or 4 months. Sometimes I would wait for half a year with no “bleeding”. As a result, I might have been one of the few females out there who was happy to get her period: it made me feel less like an anomaly. If I was the second sex, at least I’d better be good at it. But I was not.

I was underperforming at being the already underperforming second sex. On all fronts. No boy in my high school would hit on me. I later pieced things together and I realized it was in large part because I was taking part in these things called “olympiads”, which were very popular in Romania and I was very good (and mouthy) in school — so boys were intimated. Maybe I should have made more of a conversation I had overheard among two boys sitting behind me in class, along the lines of:

Boy 1: “Ruxi is actually very hot”

Boy 2: “Shut up, Ruxi is a genius (genius in Romanian is used liberally), she’d never look at a dumbass like you”

But no, this did not wake me up. At the time, all the movies and “received wisdom” I was perceiving (despite my mom’s best efforts to the contrary) was that my worth was solely determined by my looks. Since no boy would hit on me, my looks must have sucked. Ergo: I would never be loved. No fairytale for me. It might sound silly, but it was as simple as that. The life of a teenage girl is not easy: the surge in puberty hormones in women seems to have the effect of increasing our neuroticism. Perceiving oneself as ugly is enough to strangle whatever glimmer of self-esteem might manage to find its way amidst this unrelenting biochemical attack.

For about 2 years, my mom thought my period situation would stabilize itself. That did not happen, so I started doing investigations. I could not do transvaginal ultrasounds, because the mere thought of it scared me, so I only did echocardiograms of my lower abdomen. The doctor did not observe any monstrous cysts so she said: “No, Ruxandra does not have PCOS”.

Then I went to an endocrinologist. My dehydroepiandrosterone (DHEA) levels, a steroid hormone produced by the adrenal gland, were sky-high. My testosterone and estrogen levels were high too, as a downstream effect of that. I had PCOS, but somehow the person who did my scan missed my cysts. The endocrinologist gave me two options: get on the contraceptive pill (Yasmin) or take androgen receptor inhibitors. I jumped at the idea of the contraceptive pill. It had estrogen so I reasoned maybe it would make my breasts bigger and then maybe, possibly, a boy would deem me worthy of being hit on.

Boys did not start hitting on me, but I did start getting regular “fake periods”, as it happens when you are on the pill, which made me feel a bit less abnormal. But the name of my condition did not. PolyCYSTIC ovary syndrome… You might have trouble having kids…CYSTS…. Infertile…CYSTS… Words spinning in my head. Depending on the day, I was either happy to have been spared the fate of being fully the second sex or devastated that I was nothing clear, but rather an in-between creature, my ability to give life squashed from within by my ovarian cysts. In fairytales, the ugliness of the soul often spills on the outside: that’s why you have the ugly witches. The same was happening to me, why I must have repelled the boys. The ugly witches also had warts and a wart is kind of like a cyst, isn’t it?

My ovaries do not have cysts

Women are born with all the eggs they will ever have (fun fact: you come from an egg that was once part of your grandmother as she was pregnant with your mother). The eggs, or oocytes, don’t just sit around by themselves in the ovary: they are usually protected by surrounding cells, forming so called follicles. Follicles themselves go through various maturation stages: primordial, primary, secondary, antral and so on. A young woman might have hundreds of thousands of dormant primordial follicles and their number decreases with age, until it hits ~ 1000 at menopause. Primordial follicles form one’s ovarian reserve. Yet, this is not the pool follicles that are “available” for ovulation each month. That pool is the much smaller group of so-called “antral follicles”. Their number is called AFC or antral follicle count. An antral follicle is an ovarian follicle that contains a developing oocyte (egg) and has formed a fluid-filled cavity (the antrum), making it capable of being recruited for further growth toward ovulation.

This pool of antral follicles is also important for egg freezing and IVF, because that’s also where you are drawing from when you are doing ovarian stimulation and retrieving eggs. Given the high attrition rate at each step of the IVF journey, the higher the AFC, the better (shown below is an approximate plot for a 35 yr old that has a ~ high AFC for her age: 14). In general, a woman under 30 should probably freeze around 15-20 mature eggs to be almost sure of a live birth.

So more than 10 years after my PCOS diagnosis, I started my egg freezing journey like all women who do: with a transvaginal ultrasound to get my antral follicle count or AFC. My worries about boys had diminished, but a recent event had reminded me, in full force, that I was still a member of the second sex. And catalyzed my decision to freeze my eggs, which I had been postponing. And although my scientific explorations had diminished my PCOS related fears by a lot, I still had it in the back of my head that something anomalous was taking place in my ovaries.

So I entered the clinic where I would have my transvaginal ultrasound paralyzed by fear. Pastel colours, immaculate and sterile: the environment of the clinic felt designed to quell the fears of the women entering its doors. But my fears were anything but quelled. I knew the statistics, especially for older women2. How many hopes must have been crushed in this nice, sterile, comfortable, reassuring environment!

I was informed by the receptionist that a nurse would come soon to attend to me. A smiling, very nice, friendly nurse, I found out soon enough. I ambulated myself to the consultation room. After answering a few questions from the nurse, I started to undress myself, on autopilot. The nurse quickly reacted: she asked me if I wanted some privacy and turned a curtain between me and her while I was undressing. I thought it was funny: she was going to insert a device deep inside my vagina. What privacy was there in all this process? A woman’s fertility journey is a continuous and unrelenting invasion of her privacy, one might even say an attack on it, whether the process happens in vivo or in vitro. Even the implantation of the embryo is an invasion of the uterus, with the future fetus employing mechanisms not too dissimilar from that of a cancer to attach to its mother. But I nodded along: “yes, I want my seconds of privacy to undress myself before you insert that phallus-like object (the ultrasound head) deep enough into my vagina to get an image of my possibly degenerated eggs.”

And I did gain my seconds of privacy before the phallus-like object was introduced in my vagina, gently I must say, by the nurse.

She started with the uterus, which looked normal. I wanted to see my eggs though and I wanted to see them quicker. However, the nurse was calmly and politely taking notes and commenting on the health of my uterus. “I WANNA SEE MY EGGS MY UTERUS DOES NOT MATTER I KNOW IT IS OK”, I was screaming internally, annoyed. But the nurse was nice, so I smiled instead. She finally made a sharper turn of the ultrasound head, a turn that felt more invasive and uncomfortable than the already uncomfortable situation I was in. She pressed the ultrasound head firmly to one of the sides of my vagina. Finally, she was looking at my ovaries; and hence my eggs. And then she gasped.

“Oh my god, do you have many eggs!”. “Is that good?” I asked, momentarily forgetting everything I had read about all this. “Yes, yes it is”. She then started to count. Right ovary: 26. Left ovary: 24. There might have been more of them, but at some point I could tell she was bored and gave up counting. I had at least 50 eggs in my ovaries in total. In scientific terms, that was my antral follicle count (AFC). This was good. Phenomenal even — the average for my age is something around 15. I could not quite believe it though, so I asked again: “no abnormality?”. The nurse shook her head. A consultation with the doctor afterwards confirmed what the nurse said.

The so-called “cysts” in my ovary were no cysts at all. They were antral follicles. On the ultrasound above, they are the black holes in the centre of the image. And the opposite of anomalous growths: they are seeds of life. And, like many women with PCOS, I had many of them, although 50 is high even for PCOS. That was good. It felt good.

I exited the clinic and started to cry. Even now, as I write this, I look at the big round black holes that are my follicles with a vague feeling of affection, as if I was looking at the potential of my future children. They are black, gaping little holes to others, maybe even a bit gross; but to me, they are round and beautiful and plentiful — little black holes full of potential. Oh, how annoying will I be as a mom, showing pictures of my newborn!

That I unnecessarily struggled with anxiety around my PCOS and thought of myself as somehow abnormal, when in reality I had more eggs than I knew what to do with, is a testament to how antiquated, as well as poorly communicated and understood women’s health is.

Why PCOS is not that bad, but age is

The term polycistic comes from an almost century old paper (published in 1935), where authors observed a specific “string of pearls” like appearance of ovaries in women with amenorhea (the absence of menstruation). They called this a polycystic ovary. And because medicine does not like to update itself, the name stuck.

Of course, these are not actual the monstrous outgrowths the name “cyst” implies, but rather immature follicles, or the structures that contain eggs and the surrounding protective cells. Women with PCOS often have more immature antral follicles3, which leads to the above-mentioned “strings of pearls appearance”. And more follicles is actually what you want during IVF.

Why PCOS can be “good” for IVF

In natural cycles, women with PCOS are often subfertile because follicle development “stalls”: several follicles start to grow, but hormonal imbalance prevents one from fully maturing and ovulating.

In IVF, however, ovulation is no longer left to chance. Controlled hormone stimulation can push many of those arrested follicles to mature simultaneously. Instead of being a liability, the follicle surplus in PCOS can actually become an advantage: more eggs can often be retrieved, more embryos can be created and embryologists have a better chance of selecting chromosomally healthy embryos4. In fact, with ovarian stimulation, women with PCOS often do not even have to do IVF.

Not only do women with PCOS not experience worse IVF outcomes, multiple studies even suggest that women with PCOS are somewhat “protected” against the age-related decline in fertility, precisely because they start with a larger pool of eggs. For example, one retrospective study on 1540 older women (>40) undergoing IVF and comparing PCOS with otherwise healthy patients, found that the PCOS-women had a live birth chance per cycle of 26% compared to 15.2% in the normal population.

PCOS is not one disease, but many

That PCOS is called “polycystic syndrome” when no cysts are there is just the beginning of the issue with how we classify this disease. Today, the Rotterdam criteria are used to diagnose PCOS, with 2 out of the 3 symptoms below having to be present:

1. Oligo- or anovulation: irregular or absent menstrual cycles.

2. Clinical and/or biochemical hyperandrogenism: symptoms (hirsutism, acne, androgenic alopecia) or elevated testosterone/androgens on labs.

3. “Polycystic” ovaries on ultrasound: ≥20 follicles per ovary (2–9 mm in diameter) or increased ovarian volume (≥10 cm³).

But this classification obscures more complications. In reality, PCOS is a heterogeneous syndrome with distinct biological subtypes:

1. Insulin-resistant PCOS

   • Often associated with higher BMI and metabolic syndrome.

   • Linked to worse pregnancy outcomes (gestational diabetes, preeclampsia).

   • Many of the fertility benefits of weight loss (e.g., the reports of women getting pregnant on Ozempic) are mostly seen in this group.

2. Hyperandrogenic but non–insulin-resistant PCOS

   • Characterized by high male hormone levels without systemic insulin issues.

   • Can be further subdivided into:

     • Adrenal PCOS (high DHEA from adrenal glands).

     • Ovarian PCOS (the ovary itself overproduces androgens).

Despite these clear biological differences, research rarely stratifies women by subtype. Instead, most research treats “PCOS” as one homogenous condition. This is almost certainly obscuring important differences in outcomes — not just in fertility, but in pregnancy complications, long-term metabolic health, and even how ovarian ageing unfolds. But more on this in another post, on how I ended up managing my PCOS.

Why eggs age (and why this matters)

For a long time, I thought PCOS was the monster inside me that would stop me from ever having children. But after years digging through scientific papers, I came to another realization: the real monster was not PCOS at all. The real monster was time.

Unlike sperm, which men make fresh every few weeks, women are born with all the eggs they will ever have. I was made from an egg that was once part of my grandmother’s body. That’s an awe-inspiring thought — but also a frightening one. Because this comes with a price.

For decades, these eggs just sit there, waiting. But “waiting” is not neutral. This statis means that important proteins are slowly degraded, which has several negative knock-on effects. Perhaps the most important one is an increase in aneuploidy rates, or the proportion of eggs with an abnormal number of chromosomes. This percentage slowly increases with time, accelerating after a woman turns 35 and then even more vertiginously increasing after 40.

At the heart of this is a degradation of the cell division machinery, which makes sure that chromosomes are correctly partitioned to the right cell during meiosis, the cell division that eggs undergo. This happens for several reasons:

• The “glue” that holds chromosomes together weakens. Cohesin, a protein that literally holds chromosomes together until the right time in cell division, degrades with time. This means chromosomes can drift apart too early, causing eggs with the wrong number of chromosomes. Other proteins that “protect” cohesin, like SGO2, also degrade with age.

• The cell divison machinery falters. The spindle and kinetochores, which are supposed to help pull chromosomes neatly apart during cell division, also become less functional with age: the egg becomes like a machine with missing screws. As a result chromosomes lag, get missattached, or go to the wrong pole during cell devision.

• The DNA frays. Damage accumulates over decades, and the repair systems that used to be vigilant become sluggish. Epigenetic “memory” also fades, like a book whose words are smudged by too many readings.

An increase in DNA damage is also part of why the total number of eggs or ovarian reserve decreases with age: in the presence of DNA damage, primordial follicles “commit suicide”, or in scientific terms, apoptosis. Most of a woman’s primordial follicles or ovarian reserve is thus lost not through repeated ovulation, as many think, but rather due to this kamikaze-like action the eggs undertake.

But there is a way to fight all this: unlike 20 years ago, a woman can freeze her eggs and prevent this age-related decay. If done at the right clinic and enough eggs are collected, this has a high probability of success. We have the tools to defeat the monster. But the journey is not easy: from being able to afford egg freezing itself, to choosing a clinic, to dealing with the side-effects of ovarian stimulation, the burdens are non-negligible. It is with the hope that I can demystify at least some part of this process for young women that I have started this series.

But my reaction to this is that of trying to find a solution, a way to reduce, if not completely eliminate, these trade-offs: investing in technology that would advance female health and increase the fertility span to allow women more choice over their reproduction. After all, across developed countries, family and children are consistently found to score at the top for sources of meaning among adults of both genders. This suggests to me that what we should aim for is to make it easier for women to combine their career ambitions with achieving their family goals, since the latter seem to matter a great deal when it comes to living a meaningful life (with career often coming as a close second).

Having mentioned all this (and sorry to my frequent readers, who by now, know all this stuff), I consider pieces like ’s “Don’t have children” to be catastrophizing and moving us away from a positive vision for the future. For this reason, I wrote a long thread on twitter, in which I debunk some of the claims in her article, including those related to the danger of pregnancy in the US. Since then she has doubled down on previous claims, with another article:

More to Hate

Pregnancy is (Currently, in the US) Too Big of a Gamble

Phew. I’ve rarely had such a big reaction to something I’ve written. In my previous post, I argued for a feminist anti-natalism in the US currently, on the grounds that the package deal of pregnancy and early childrearing, especially with a male partner, is currently an intolerably bad deal for many if not most people contemplating it. There were some caveats: some women, and other people who can get pregnant, will want a child so badly—and may need to have one soon, for a variety of reasons—that their desires will rightly silence the considerations I adduce in favor of refraining for the time being. That seems to me entirely proper: it’s your life, and your body, and I believe that, ultimately, your autonomous desires should trump rational argumentation. For that reason, I was careful to specify…

Read more

a year ago · 28 likes · 3 comments · Kate Manne

She specifically references my thread by saying:

I was accused on X of catastrophizing vis-à-vis the risks of pregnancy: the chance of dying are, after all, on par with that of being struck by lightning. But I don’t go out and stand around during thunderstorms either. It is perfectly reasonable when contemplating pregnancy, much as with any elevated-risk activity, to insist that conditions are tolerably safe and oriented to protecting you properly before going through with it. More: it is fair to want pregnancy in particular to be as safe as it can be, given the ostensible moral good of producing children for our society.

Yet I maintain her initial article was catastrophizing! If I were to be very pedantic, I would point out that the chances of dying in pregnancy are 10x less than that of being struck by lightning. But, more importantly, the absolute risk of dying due to pregnancy is small: 1 in 100,000. Despite this very small risk, she seems to at least in part base her argument on the idea one should not have children specifically because in US death rates from motherhood are higher than in other developed countries. In her article, Kate says:

Pregnancy is not just far more likely to result in preventable death than it should (comparing the US, again, to other developed nations). It is also far more likely to result in long-term as well as acute sickness, suffering, and a subsequent sense of precarity. It is hard to get decisive figures on the matter, but an estimated 8% of pregnancies involve complications “that could result in harm to the mother or baby if they are not treated.” And treatment is often woefully lacking. I heard from reader after reader who catalogued not death but near-death, as well as forms of pain, suffering, and long-term ill health that were the result of two things: (a) pregnancy, and (b) inadequate health care during and after the process. This is both very bad in and of itself, and it is also an injustice that one may quite fairly balk at. Moreover, it can result in a sense of vulnerability, helplessness, and trauma even if the very worst outcomes do not materialize. The sense that you could have died, and that nobody would have saved you, is bad over and above any lasting health consequence

But this argument, even if we accepted that US maternal deaths are higher than in other countries, does not really make sense to me. Firstly, life is not about minimizing risk. We need to balance this against the positives! Secondly, whether mortality rates from pregnancy are slightly higher in your country than another one, when very small in the absolute, should not detract you from doing something as life-changing as having a child. It’s the equivalent of making choices based on comparing yourself to others instead of some internal sense of what is, on aggregate, good for you.

Leaving aside these moral considerations, however, I do not think the argument is correct factually, mostly because it’s not even clear US has higher maternal mortality rates than other countries. As this excellent article from points out (I recommend reading the entire piece), the recent apparent rise in maternal mortality in the US is down mostly to changes in measurement:

To follow the ICD-10 definition and make sure that maternal deaths weren’t going uncounted, the United States added the “pregnancy checkbox” to death certificates, starting in 2003. The US used an automated system to code deaths as maternal deaths if the checkbox was ticked for women between the ages of 10 and 54, for deaths caused by medical conditions, regardless of other information on the death certificate.⁸

(…)

The researchers also estimated what the maternal mortality ratio from 2003 to 2017 would have looked like under two hypothetical scenarios: (a) if all states adopted the checkbox simultaneously, or (b) if none of them did. In both scenarios, they estimate that there would have been no change in maternal mortality ratios between 2003 and 2017. In other words, the rise in maternal mortality is largely explained by the staggered adoption of the checkbox.¹³ The researchers also noted that the impact of the change in measurement was greatest among older women and non-Hispanic black women.¹⁴ The checkbox increased the ability to detect pregnancy-related deaths that would have been missed otherwise, but in some cases, it also resulted in overcounting deaths from other causes.

Saloni also brings compelling evidence that other developed countries are underreporting maternal deaths:

While the United States has used the checkbox to automatically code deaths as maternal if it is ticked¹⁹, this practice is not followed in several other countries.²⁰ There is strong evidence that maternal mortality, as defined in the ICD, is underreported in national statistics in many countries.²¹ One reason is that some countries do not use data from the checkbox to identify potential maternal deaths, or do not routinely conduct additional investigations to identify unreported maternal deaths.²⁰ Some countries have implemented systems separate from their vital registries to investigate potential maternal deaths further.²² These systems include “enhanced surveillance”, which involves an additional system for more detailed monitoring, and “confidential inquiries”, which are private investigations into individual cases. These investigations have been conducted infrequently, and the maternal deaths identified through these systems are not necessarily counted in vital registries for national statistics and given to the WHO.²³ Research finds that the number of maternal deaths from vital registries tend to be lower than equivalent definitions from these other surveillance systems.²⁴ In low- and middle-income countries — where death certificates and vital registries are often lacking — other sources of data are used to determine maternal deaths, including hospital records, and verbal autopsies.

Her recommendation and conclusion?

To help identify missed deaths, the United States introduced a “pregnancy checkbox” on death certificates, and deaths of women with this box ticked would be coded as maternal deaths in most age groups.While this helped identify maternal deaths that would have been missed, it also led to some misclassification and false positives from women who had not been pregnant or had died from other incidental causes. Because of this, the US changed its coding system in 2018 to disregard the checkbox for deaths of patients under 10 or over 45 years old. Researchers have also recommended that additional quality-assurance measures are used to verify potential maternal deaths before they are compiled in US national statistics. In other high-income countries, there is strong evidence that maternal mortality is underreported in national statistics (…) By improving data collection and surveillance of maternal deaths further, the world can have a better understanding of where and why mothers are dying, mobilize resources and policies to save lives, and reduce maternal mortality further.

Other arguments Kate brings up are related to complications related to pregnancy — yet again, it’s not clear these are more frequent in the US than in other places. Otherwise, yes, we should make female health more of a priority, which is exactly what I am advocating for.

As for the abortion related bans that she mentions, I have repeatedly mentioned my stance: as someone coming from an ex-communist country where contraceptives of all kinds were literally banned, I have a strong, visceral repulsion to all forms of reproductive control and I am strongly pro-choice. However, I am not sure about the relevance of these bans to this specific discussion. Presumably Kate wrote the article arguing against having kids for women who are deciding whether to have them or not, which means they are presumably in control of their fertility, not for women who are in the unfortunate and tragic situation of being forced to bear a child due to lack of access to proper medical care and lack of options. Furthermore, it seems strange to me that one would make a life decision as important as whether to have children or not based on whether Trump is president or not (or any other political event).

Pregnancy and childbearing is not for everyone and women should be generally appreciated much more for what they are doing when they take on the task of bringing children into the world — here I suspect me and Kate agree. I also agree with her that nobody should feel “obligated” to give birth — it’s entirely a personal choice. But other than that, I still think that many of the arguments in these posts are based either on poor data or on stuff that’s not that relevant, like who happens to be President at a given point in time. Unless that affects you directly, e.g. via a large decrease in net worth because the economy is dwindling, I do not think important life choices should be influenced by whether your preferred candidate is in power or not.

And to end all this this, I would add that the right response is that which I always emphasize in my articles. If you want to help women, get involved in initiatives to study fertility and women’s health, raise funds for such research, start companies around it, raise awareness of current fertility preservation strategies (e.g. egg freezing and so on). The header image of this post is a Gustav Klimt painting of a pregnant woman called “Hope”. There is a literal reading here: pregnancy is the essence of hope, the way in which life regenerates itself. And a metaphorical one: we should approach the challenges that women face with hope for a better future, without forgetting what a unique moment we are living in: for the first time in the existence of humanity, we are armed with the tools of science to achieve this.

The simplest definition of autoimmune diseases is as a class of pathologies wherein the immune system starts attacking the self; something that, in healthy individuals should not happen, due to the body “training” immune cells in something called “self-tolerance”. Behind this very broad definition though, hides a labyrinthine complexity. Autoimmune conditions differ wildly in phenotypic manifestation: ranging from relatively “mild” psoriasis to conditions that can have a ~ large impact on life expectancy like type I diabetes (approx 10-12 years). They are heterogenous in terms of the organs affected: some autoimmune conditions are relatively “localised” (e.g. inflammatory bowel disease (IBD)), while others are systemic1 (e.g. systemic lupus erythematous) and they can involve a variety of immune cells (B cell, T cells, monocytes etc).

Historically, autoimmune diseases have been treated with non specific therapies like corticosteroids: these are drugs that broadly suppress immune function; because of this, they have limited efficacy and come with a host of unwanted side-effects, including a high susceptibility to infections.

In recent years, however, biotech and pharmaceutical companies have become increasingly interested in more targeted therapies against autoimmune diseases: molecules that are directed towards specific immune cell populations (e.g. T cells or B cells). In many ways, this can be seen as a “drug repurposing” effort, after the success of a host of modalities in treating blood malignancies2. After all, in a blood cancer you are trying to destroy an immune cell population gone rogue — not too dissimilar from what one’s aim is in autoimmune conditions! Trying an already tested approach in another type of disease that shares a key commonality seems like the next logical step. And that’s exactly what’s happening at the moment.

After antibodies had their moment, currently the “hottest” modality that is being repurposed from cancer to autoimmune diseases is perhaps CAR-T cells (Figure 1). These have already been transformative in blood cancers (e.g. Brexucabtagene autoleucel or Tecartus achieves 63% complete remission rate in mantle cell lymphoma3 ). The excitement related to CAR-T in autoimmune disease was spurred by a study on just five patients with treatment resistant systemic lupus erythematosus (SLE), a debilitating autoimmune condition, published in Nature Medicine in 2022. What this study lacked in numbers of patients, it more than made up in terms of results: the CD194-targeted CAR-T cells, which basically attacked most of the existing B cell population in the treated patients, led to drug-free remission. This was true even after the B cell compartment was repopulated with new ones. In context, these are very promising results: not only do they point towards a potential “cure”, but the fact that repopulation of B cells does not lead to disease relapse, suggests that patients do not have to suffer the consequences of long term immune suppression, as seen with corticosteroids: they just seem to develop new, better functioning B cells post this very harsh treatment. Of course, such small studies have to be treated with caution. However, subsequent investigations (also in small numbers of patients) showed similarly promising results in other autoimmune conditions, including systemic sclerosis and idiopathic inflammatory myositis. These very optimistic results have spurred larger scale clinical trials (Figure 2).

Autoimmune diseases are in some sense piggybacking on years of research done for oncology indications. And they will continue to benefit, in the future, from all the advancements that are currently being made in achieving more specific targeting of blood cancers.

One problem with many CAR-T cell therapies at the moment is that they are broadly attacking all cells expressing a specific antigen — this is quite a “blunt tool”. For example, CAR-T cells targeting the CD19 receptor can cause severe side effects like neurotoxicity or cytokine release syndrome. For patients with relapsing cancer, which face short survival timelines, such risks might pass cost-benefit analyses; but for autoimmune conditions, however life impairing, these side-effects can be real detractors. Fortunately, the oncology field has also given rise to companies like Outpace Bio or SentiBio, which are pursuing more sophisticated and precise strategies: using protein engineering techniques, they are aiming to create CAR-T cells that only get activated when specific combinations of antigens are present on the surface of a cell using logic gates5 . This would allow the targeting of ever smaller and more relevant cell subpopulations. In the meantime, techniques like single-cell RNA sequencing (scRNAseq) will allow more precise identification of relevant immune cell subsets, although determining causality will still remain a challenge6. OutpaceBio is also pursuing a strategy that would allow precise cytokine7 release only at the desired site, a way to prevent the potentially fatal side effect of cytokine release syndrome.

Another problem with current CAR-T therapies is that they require lymphodepletion, usually involving chemotherapy to remove existing immune cells, in order to “make place” for the CAR-Ts, which would otherwise be “outcompeted” by the patient’s existing T cells. This is of course very taxing on the body and there is evidence chemotherapy increases the risks of further developing malignancies down the line. Yet this is another problem that companies are working on solving — for example Capstan Therapeutics has raised a 175$ million Series B from leading investors like RA Capital to pursue in vivo CAR-T engineering. That means targeting the T cells of a patient with mRNA to express the required CAR in vivo, which would mean no lymphodepletion is needed8 (Figure 3).

Notably, Capstan is specifically focusing on autoimmune diseases9, which suggests that we are quite deep into “engineering CAR-Ts for the specific purpose of autoimmune indications” territory. Of course, all this has to be taken with a grain of salt. There is no incentive for start-ups like Capstan or Outpace to disclose the issues they are facing and we are yet to see their promised technologies work. And Capstan’s value proposition is so “big if true”, that them not getting acquired for a formidable sum in the following years will look bearish to me. What we can observe though is clear interest and work in the space, which seems to have convinced some top biotech investors10.

And CAR-Ts are just one of the ways in which autoimmune diseases are benefitting from advances in cancer treatment. Other modalities from oncology that show promise include bispecific antibodies, or antibodies that bind two targets at the same time. Overall, one can say the field of autoimmune disease has greatly benefitted from advances that were largely driven by the pursuit of therapies for blood malignancies. I believe that we might see something similar in the treatment of aging related disease.

My prediction is that at some point, the continuous development of ever more specific and efficacious molecular tools for depletion of certain immune subsets, driven first by oncology and then increasingly by autoimmune indications, will meet a better understanding of the ways in which the immune system is implicated in systemic age-related dysfunction. The growing importance of autoimmune indications is particulary important, as these demand higher standards of safety and specificity than cancer, which will incentivize companies to optimize on these dimensions. These will be highly beneficial for aging-related diseases like atherosclerosis, which already benefits from existing treatments, hence specificity and safety are needed. Furthermore, if we ever aim to do this preventatively (deplete pathogenic immune cell populations in the absence of overt disease), these two properties become even more important. So to summarize, the way I see the direction of positive spillovers is:

Blood cancer therapies → autoimmune therapies → aging therapies focused on immunosenescence

But, before getting into more details, let’s establish some premises. Firstly, that immune aging is very important for systemic aging.

Immune aging is very important

I have long believed that out of all types of aging, immune aging is the “Queen” of all. This is because, as blood circulating entities, immune cells interact with virtually every tissue, so immune senescence reverberates throughout the body. Perhaps the earliest hints that point towards the importance of immune aging to the decline in function across the entire body come from classic heterochronic parabiosis studies —in which young and old mice share a circulatory system. These revealed that “young” blood leads to younger phenotypes across multiple organs in older animals, including the brain (here and here); conversely, “old” blood accelerates aging features in young animals. Although the precise pro‑aging factors remain elusive, aged pro‑inflammatory immune cells are leading suspects.

Why might they be the prime suspects?

The first obvious answer is that they are circulating blood cells. The second answer refers to a phenomenon called “inflammaging”. We know that as the immune system ages, its performance declines, which translates for example in worse responses to vaccination and infection in elderly individuals. Paradoxically, it also drifts into a state of chronic, low‑grade activation known as “inflammaging”. Even in the absence of overt illness, older adults typically carry higher blood concentrations of inflammatory proteins—interleukins (IL‑1, IL‑6, IL‑8, IL‑13, IL‑18), type I interferons  (IFN‑α, IFN‑β), tumor necrosis factor (TNF) and its receptors—as well as C‑reactive protein, transforming growth factor‑β, and serum amyloid A. Another hallmark of dysregulated immunity is the rise in antinuclear autoantibodies11 (ANA) with age — these are self-reactive antibodies commonly seen in autoimmune diseases. Perhaps one could describe all this, at a very high level, as an increase in entropy in the immune system with age: somehow worse at doing what it’s supposed to do and bad at doing what it is supposed to do.

More direct links between immune cells specifically (and not just blood generically) and aging come from studies in mice. For example, deleting the DNA‑repair gene Ercc1 specifically in hematopoietic stem cells, which are the precursors to all other immune cells, drives their senescence and triggers premature aging in multiple organs, whereas transplanting youthful immune cells dampens these effects. Similarly, inducing T‑cell–specific loss of a mitochondrial‑stabilizing protein in mice induces systemic frailty—neurologic, metabolic, muscular, and cardiac—through chronic TNF‑α‑mediated inflammation; blocking this inflammation reverses many deficits .

These are very interesting studies (from 2021 and 2020 respectively) that are certainly indicative of an important role of immune aging in systemic dysfunction, but they have a severe limitation: to “age” the immune system in these mouse models, the authors resort to very “extreme interventions” like deleting an important DNA repair gene or a mitochondrial protein. Yet a recent paper from 2024 does not suffer from such limitations: the authors show that in otherwise normal, old mice aging leads to emergency myelopoiesis (or overproduction of cells of the myeloid lineage, like monocytes) and an enhanced production of interleukin-1α (IL‑1α ) —a pro-inflammatory cytokine, from progenitor myeloid cells. These IL‑1α producing myeloid cells accumulate in lung tumours and stimulate its progression through suppression of other, positive anti-cancer immune responses. Blocking IL‑1α production using an antibody (a common therapeutic strategy in cancers), also slowed the growth of lung, colon and pancreatic tumours.

But this is about cancer, not systemic aging! Well, it’s worth mentioning aging is one of the most important risk factors for cancer development, with those over 65 bearing more than half of the cancer burden in the US. And non–small-cell lung cancer (NSCLC) is one of the most associated with aging cancers out there — which is precisely why the authors of the study focused on lung cancer. To further support this study, another recent paper in mice showed that depletion of myeloid-biased hematopoietic stem cells led to a more phenotypically youthful immune system.

Ok, ok but this is in mice! We have cured cancer in mice so many times. How are these findings relevant?

Well, such in-depth studies are mostly possible in mice. But besides that, there are two interesting findings in this study that point to its relevance to humans. Firstly, in mice, this overproduction of IL‑1α is driven by an age related decrease in the activity of a protein called DNA methyltransferase 3 (Dnmt3a). In humans, we observe somatic mutations that impair DNMT3A function and the frequency of cells carrying such mutations increases with age. This is all part of a phenomenon known as clonal hematopoiesis of indeterminate potential (CHIP), which I have already written about. CHIP has been shown to be associated with an increased risk of age-associated diseases, including, most prominently, cardiovascular disease. Secondly, the authors mined human existing publicly available scRNAseq human data and found expression signatures of the proposed pathogenic IL‑1α secreting cells among the monocytes that accumulate in human lung cancers. Thirdly, myeloid bias (a switch to production of cells of myeloid lineage) with aging is also seen in humans.

And, if studies in mice do not convince you, there is accumulating evidence (from humans) regarding the importance of so-called “chronic inflammation” in a host of age-related diseases including cardiovascular disease, dementia etc.

On long covid and autoimmunity

A few months ago I wrote a thread in which I explained why findings about post SARS-Cov2 vaccination syndrome are not what people made them seem to be. Writing this thread led me down a rabbit hole of reading about exposure to SARS-Cov2 antigens12 and the potential mechanisms behind long COVID — the disease whose very existence has been questioned so vehemently by many. One common “meme” on the internet (and in broader society) is that the higher prevalence of this disease in young women is just proof that it’s somewhat of a “fake disease” — just a psychosomatic manifestation of the higher anxiety that is prevalent among this demographic. The fact that “long COVID” has somewhat diffuse, hard to pin down or characterise via “hard measures” symptoms, makes the conversation even more difficult. Commonly cited manifestations of long COVID include: fatigue, shortness of breath, joint pain, insomnia etc. The nature of these symptoms means it’s a condition that’s easy to dismiss: after all, we cannot measure “fatigue” on a Computer Tomography (CT) scan like we do with cancer. Furthermore, they can be abused by bad actors and used to justify leaves of absence. But, in the majority of cases, as many people who have met actual long COVID sufferers know, the idea that this is just in young women’s imagination is nonsense.

There is something else that young women are predisposed to, besides anxiety: that is autoimmune diseases. And the group of Dr. Akiko Iwasaki, a very well-known immunologist, who is behind the post vaccination syndrome study I mentioned above, found something very interesting: individuals with long COVID, particularly those with neurological symptoms, often produce diverse and functional autoantibodies (AABs)— which are specific of autoimmune diseases, but also, as mentioned in the previous section, accumulate in the blood of healthy elderly people. Using a human proteome array, the researchers identified that these AABs frequently target proteins in the nervous system and correlate with symptoms such as brain fog, headache, and dizziness. Now, Iwasaki is looking into other post-acute viral syndromes with similar symptoms to long covid, including post-treatment Lyme disease and myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS).

Purified IgG13 from Long Covid patients reacted with both human and mouse neural tissues, including the pons, sciatic nerve, and meninges. When these IgG samples were transferred to mice, they induced symptoms similar to those reported by the patients, such as pain sensitivity, muscle weakness, and coordination issues. Mice injected with long covid patient IgG showed biological changes, including small fiber neuropathy, supporting a pathogenic role for these antibodies. The study suggests that autoimmunity may underlie a subset of Long Covid cases and that targeting AABs could be a therapeutic strategy. It could also explain why young women are both more likely to have long covid and autoimmune conditions: and underlying predisposition could underpin both.

Ok… but what does that have to do with aging? Well, it’s related to two concepts I mentioned in the second section of this essay: firstly, the broad phenomenon of “inflammaging” or chronic inflammation, which is thought to be a hallmark of immune aging and the accumulation of autoantibodies in the blood of “healthy” elderly people (two phenomenona that I think are related).

Viral exposure, autoimmunity and age related diseases

This is where things become very speculative.

Like in many other age-related pathologies, the role of inflammation or to use a more precise term, “improper activation of immune cells” in cardiovascular disease, most notably atherosclerosis, is being increasingly recognised, including by top leaders in the field, like Prof Peter Libby. I have copied exactly his explanation from one of his latest reviews on the topic in Figure 5. Interestingly, the two most successful and researched treatments for atherosclerosis (statins and various ways of inhibiting/downregulating PCSK9 — a protein involved in lipid metabolism) involve cholesterol control, with relatively little attempts to precisely control immune cell populations.

Against this backdrop of an increasing recognition of the role of inflammation in atherosclerosis (which is, of course, another age-related disease), my interest was piqued by a recent study analysing CD8+ T cell14 populations in human coronary plaques15.

The study found that CD8+ T cells within plaques are clonally expanded and often specific to viral antigens, including influenza, (Cytomegalovirus) CMV, (Epstein-Barr Virus) EBV, and conserved coronavirus epitopes. Epitopes are the specific parts of antigens that are recognized and bound by an antibody or a T cell receptor. Clonal expansion of T cells refers to the process by which a single T cell that recognizes a specific antigen rapidly multiplies into many identical cells to mount an effective immune response — overall the findings above suggest that these T cells had been stimulated to expand by the above mentioned viral infections.

These virus-specific T cells were most enriched in fibroatheromas, a stage associated with plaque instability and rupture. Notably, many of these T cells remained in the plaque despite the absence of active infection, raising questions about their persistent activation. Through sequence analysis, the researchers discovered that viral epitopes recognized by these T cells shared amino acid and nucleotide similarities with self-proteins expressed in vascular smooth muscle cells (SMCs), endothelial cells, and cardiomyocytes. As explained in the caption for Figure 5, the interaction between pro-inflammatory T cells and SMCs has been postulated to contribute to plaque progression. These findings suggest a potential mechanism of molecular mimicry, where immune responses initially targeting viruses cross-react with self-tissues, in autoimmune-like fashion. In vitro experiments confirmed that T-cell receptors (TCRs) from the plaques responded not only to viral peptides but also to self-antigens like TSPAN17 and Zip9. Such cross-reactivity provides a possible explanation for viral-triggered cardiovascular complications, even in previously infected or vaccinated individuals.

The findings also align with observations that autoimmune conditions elevate cardiovascular risk. These data imply that a subset of virus-specific T cells may misidentify vascular proteins as foreign, contributing to chronic inflammation and plaque progression. Overall, while not conclusive, the study supports a model in which viral exposure may initiate or exacerbate autoimmunity within atherosclerotic plaques.

Ok… so what am I suggesting here? I have presented only two studies so far — they both suggest a connection between viral exposure and triggering of a low-grade autoimmune-like response. In the first case, associated with long COVID. In the second case, with an age-related disease. But let’s recap some of the earlier stuff I mentioned: the phenomenon of “inflammaging” or low-grade aberrant immune activation with age, as well as the clues regarding its importance in driving systemic aging. I believe there is a possibility that accumulated exposure to viral infections with age could be responsible for “inflammaging”. In other words, that, in the long run, we are all subclinical sufferers of long COVID (or long EBV, long CMV etc). For some, this flares up very blatantly. For most, accumulated exposure to various viral infections merely leads to a progressive increase in these autoimmune-like, pro-inflammatory cell populations, which in sufficient numbers, end up impacting a whole range of age related diseases.

There are other pieces of evidence that antigen exposure produces a pro-inflammatory potentially self-directed response: for example, individuals with HIV, even those on antiretrovirals, are more likely to suffer from cardiovascular diseases. And let’s not forget other recent results that took the scientific world by surprise, like the recent link between EBV infection and multiple sclerosis (a full-blown autoimmune condition).

This is of course, all highly speculative. However, you do not need to believe my theory that links viral exposure, autoimmunity and aging to think that treatments for autoimmune diseases might lead to “positive spillovers” into aging-related pathologies. Instead you need to buy a much narrower claim: that there are subpopulations of aberrant immune cells that are causal in age-related decline and that depleting those would lead to amelioration of a host of these age-related phenotypes. This is a much narrower claim, which at this point I think has pretty solid evidence.

The good news is that we have been developing ever more precise tools to target blood malignancies. This has led to a “positive spillover” towards targeting autoimmune diseases. Biotech and pharmaceutical companies will be incentivised to further engineer these tools for autoimmune conditions specifically, a market that is expected to grow to 193$ billion by 2031 by IQVIA (from $154 in 2021). And given that autoimmune therapies have to pass higher safety thresholds than cancer ones, this is a good sign for age-related diseases: it means strategies that avoid side effects will be heavily pursued. As mentioned in the first section, we are already seeing this with Capstan, whose main selling point is avoiding lymphodepletion, precisely because this is too harsh of a side-effect for most autoimmune disease suffering patients (as opposed to those affected by cancer).

Although the first part of this essay was mainly focused on CAR-T therapies, it might be that for anti-aging immunosenescence targeting therapeutics, another modality is more suitable. These are bispecific T cell engagers (BiTEs): antibodies that can bind two antigens at once and direct T cells to kill a desired cell type — so far these therapeutics have shown promise in autoimmune disease, though there is less data than for CAR-Ts. Overall, BiTEs have shown less

I believe that when we understand the subpopulations of pro-inflammatory immune cells that drive age-related diseases better, these tools will be repurposed yet again, in the same way we saw a repurposing from oncology to autoimmunity. At some point, we might even deplete certain immune cell population preventatively and not just as a treatment. For this, we would need incredibly precise and safe modalities: the safety requirements for preventative anti-aging medicine will be higher than for any type of disease.

Although the first part of this essay was mainly focused on CAR-T therapies, it might be that for anti-aging immunosenescence targeting therapeutics, another modality is more suitable. These are bispecific T cell engagers (BiTEs): antibodies that can bind two antigens at once and direct T cells to kill a desired cell type — so far these therapeutics have shown promise in autoimmune disease, though there is less data than for CAR-Ts. Overall, CAR-Ts have shown more striking results in B cell malignancies in terms of pure efficacy: for example, a meta-analysis of success rates in of third‑line therapies in relapsed/refractory multiple myeloma (MM) found pooled complete remission rates of 54% for CAR T‑cell therapies versus 35% for bispecific antibodies (P < 0.01). There are many reasons for why CAR-Ts might be more efficacious, but, briefly TCR complex activation is usually not sufficient for target killing. CAR T cells provide co-stimulatory signals that BiTEs do not.

However, they also have higher risk profiles and are harder and more expensive to manufacture than BiTEs. Given the increased importance of safety/convenience in anti-aging therapeutics compared to cancers, unless issues like lymphodepletion are solved for CAR-Ts, BiTEs might actually be more promising long term for depletion of aging related immune cell subsets. This is particularly true for 2nd and 3rd generation BiTEs. What do I mean by a 2nd or 3rd generation BiTE?

For example, one could engineer BiTEs that include domains which increase their efficacy. For example, checkpoint inhibitory T cell engagers (CiTEs) comprise a BiTE core with an added immunomodulatory protein, which serves to reverse cancer-cell immune-dampening strategies (by incorporation of an anti‑PD‑1 Fv), improving efficacy. Although this is relevant to cancer, the chemical strategy (a version of click-chemistry) used to generate these CiTEs is what’s important here. One could imagine that for aging applications, one could use similar chemical methods to attach another molecule that improves efficacy (e.g. a cytokine like IL-2).

Besides this, modified BiTEs have another advantage for the future: we are getting better and better at AI modelling of antibodies, which means more complex BiTEs could benefit from that (although this discussion would warrant an entire new post).

8

For the biology nerds out there, the “secret sauce” of Capstan is that they make LNPs (the vehicles that carry the mRNA) cell type specific. In this case, T cell specific. In this case, they do so by attaching a CD8 antibody to the LNP, such that the mRNA is delivered only to CD8 expressing cells (so CD8+ T cells)

I apologize to Gideon Lewis-Kraus for this mistake and edited my article to reflect the fact that he was merely describing objections to Scott, not necessarily taking a side or another. To be clear, that does not mean I do not think Scott has been unfairly maligned and that the “Great Man Theory of Platforming” is unfortunately very prevalent. It’s just that the specific article I was quoting was merely describing the stances of different sides, as opposed to taking a position.

The previous version is in the footnote1, but the new one reads as follows (I am only including the relevant parts of the article):

One of the main accusations against Scott comes from those who blame him for “platforming” neoreactionaries like Yarvin. How did Scott do that? Did he run events with Curtis as the headline speaker, heaping praise on him? No. To many, Scott seems to be guilty of not completely refusing to engage with neoreactionaries, something that stains Scott by association, despite him arguing against them.

To better understand this debate, I recommend an article from 2020 in the New Yorker, written by the author Gideon Lewis-Kraus. The piece discusses Scott’s “doxxing” by the NYT⁴, but also does a pretty good job at summarizing the interests of the rationalist community for a broad audience and brings up both the arguments pro and against Scott and the rationalists’ involvement in the spreading of neo-reactionary thinking. The paragraph below is particularly relevant, with the sentences written in bold and italic illustrating the main points of criticism against Scott that are often brought up by those who consider him guilty of “platforming” Yarvin:

“Many rationalist exchanges involve lively if donnish arguments about abstruse thought experiments; the most famous, and funniest, example, from LessWrong, led inexorably to the conclusion that anyone who read the post and did not immediately set to work to create a superintelligent A.I. would one day be subject to its torture. Others reflect a near-pathological commitment to the reinvention of the wheel, using the language of game theory to explain, with mathematical rigor, some fact of social life that anyone trained in the humanities would likely accept as a given. A minority address issues that are contentious and at times offensive. These conversations, about race and genetic or biological differences between the sexes, have rightfully drawn criticism from outsiders. Rationalists usually point out that these debates represent a tiny fraction of the community’s total activity, and that they are overrepresented in the comments section of S.S.C. by a small but loud and persistent cohort—one that includes, for example, Steve Sailer, a peddler of “scientific racism.” (…) Alexander has long fretted over the likelihood that the presence of these fringe figures could tarnish the reputation of the blog and its community. In late 2013, he published “The Anti-Reactionary FAQ,” a thirty-thousand-word post now regarded as one of his first major contributions to the rationalist canon. The post describes the world view of a group, centered around a figure called Curtis Yarvin, also known as Mencius Moldbug, whose “neoreactionary” views—including an open desire for the restoration of feudalism and racial hierarchy—contributed to the intellectual normalization of what became known as the alt-right. Alexander could have banned neoreactionaries from his comments section, but, on the basis of the view that vile ideas should be countenanced and refuted rather than left to accrue the status of forbidden knowledge, he took their arguments seriously and at almost comical length—even at the risk that he might lend them legitimacy. Ultimately, he circumscribed or curtailed certain “culture war” threads. Still, the rationalists’ general willingness to pursue orderly exchanges on objectionable topics, often with monstrous people, remains not only a point of pride but a constitutive part of the subculture’s self-understanding.”

Who knows? maybe without this “platforming” that Scott facilitated, we would have never had Curtis Yarvin. Of course, this is completely backwards. Elon Musk is much more of a Great Man of History than Scott Alexander is a Great Platformer of neoreactionaries, in the sense that the latter role probably had much less causal impact on Yarvin’s rise⁵.

There is a meta-discussion to be had here: in this mainstream vs fringe debate, it is easy to get carried away and ascribe to individuals general views that one associates vaguely to the faction to which they belong. I believe it is generally true that Scott has been unfairly maligned by the mainstream. However, I let this impression cloud my judgement when it came to discussing a specific article written by an actual individual, which was, as mentioned before, much more descriptive than normative. This is exactly what I have been criticizing the mainstream for doing when discussing people like Scott — doing a vibe-based assessment based on vague feelings and assumptions instead of unbiasedly engaging with the actual material! And now I am guilty of it!

Such behaviour is responsible for the polarization that I abhor and I am genuinely sorry for taking part in it, albeit in a small way. I believe this is what happened here and it was a mistake to take part in this type of fallacy. I think it is inevitable that, as a writer, one would much such mistakes. Here is a post whose whole purpose is to acknowledge one such mistake and sincerely apologize for it. And to add to this: the SSC community is one that normalized this practice of admitting epistemic failures of this kind— so I guess hurray to Scott & co for that.

Anyway, to add to all of this, it seems the “Great Man Theory of Platforming” has somewhat been broken. Curtis Yarvin will be debating very respectable Dr. Danielle Allen, a professor of political philosophy at Harvard. And she does not seem to be just *any* professor — her CV is remarkable in a very legible to the mainstream way. As an aside, twitter user AlicefromQueens has an interesting thread about how she might have been passed over as President of Harvard over Claudine Gay.

Edit 2: In the original version of the article, I actually described the article by Gideon Lewis-Kraus as much more normative than it actually was. I edited that part to reflect he was merely describing both the pro and anti side re Scott Alexander.

The average reader of The New York Times must have been a bit puzzled as to why the magazine chose to interview a 51 year old guy called Curtis Yarvin for its 18 January 2025 issue: to most, he must have looked as coming out of nowhere, an obscure intellectual. To those “online” enough, he is anything but. Curtis, also known as Mencius Moldbug, has long been at the heart of the so-called “Dark Enlightenment”, an ideology that I have witnessed growing in influence in the last years of being “online”, only for this to culminate in what one could consider real political influence. Don’t take it from me, a very online person, who might be biased to assign too much importance to internet writers; take it from an article in another mainstream outlet, Time Magazine. The following excerpt not only highlights Yarvin’s growing influence, but also sketches, in broad terms, what the “Dark Enlightenment” is all about:

Largely ignored by academic philosophers, the “Dark Enlightenment” movement and Yarvin have curried favor and influence with tech executives in recent years. (…) Not unlike the Futurists, Yarvin advocates for replacing democracy with a kind of techno-feudal state—for the government to be run like a corporation, with the president as its “CEO.” This new system is elitist—“humans fit into dominance-submission structures” Yarvin wrote in 2008; and it’s authoritarian—“If Americans want to change their government, they’re going to have to get over their dictator phobia,” he said in 2012 (…) What’s even more alarming is that Yarvin’s outsize influence on tech executives has now made its way to Washington. The signs are everywhere: Yarvin was a feted guest at Trump’s so-called “Coronation Ball” in January 2025. Vice President J.D. Vance, a protegee of Thiel’s, spoke admiringly of the blogger’s influence on his thinking when interviewed on a podcast in July 2024. And while Andreessen’s role in the Trump White House is unofficial, The Washington Post reported in January 2025 that the executive “has been quietly and successfully recruiting candidates for positions across Trump’s Washington.”

Even the journalist running the NYT interview, David Marchese, seems a bit perplexed by this growing influence. He preambles the interview with:

I’ve been aware of Yarvin, who mostly makes his living on Substack, for years and mostly interested in his work as a prime example of antidemocratic sentiment in particular corners of the internet. Until recently, those ideas felt fringe.

In another post from last year, I argued against Ross Douthat, whose argument was that the Internet was enabling a “Monoculture”. On the contrary, I said: it’s an amazingly efficient and fast incubator of new ideologies, and ideas born on the internet shall come to dominate the 21st century much more than those born in traditional institutions (note the term “ideology” — I am not including here STEM stuff). In that post, I was mostly citing examples of how rationalism had become embedded into tech culture and influenced the development of arguably the most powerful technology we have now, AI, itself. Curtis’ neoreactionary movement is just another very important example that proves my point.

If the mainstream media has been ignoring Curtis for a long time, other, at the time obscure internet writers, were not. Most notably, Scott Alexander wrote one of the first comprehensive take-downs of neoreactionary thought back in 2013: the Anti-Reactionary FAQ. In fact, back then he spent quite a lot of time arguing against proponents of this ideology.

It was for this reason that I strongly disagreed with another, more recent article in The New York Times, this time by Michelle Goldberg, titled “The vibe-shifts against the right”. This piece brings the names of 3 purportedly ex right-wingers who have changed their tune recently, in response to the Trump election. Their names? , and Scott Alexander. The article uses Scott’s recent post, “The Twilight of the edgelords”, a meditation on whether centrists like him led to the rise of populism to clump him together with Alex and Richard, who were bona fide right-wingers1, and more recently disavowed some of their past beliefs.

A few days ago, I wrote a response to the aforementioned Scott Alexander post, arguing that no, intellectuals2 like him are not responsible for the rise of populism and his doubts are a result of excessive humbleness and self-awareness. If individuals promoting ideologies, as opposed to the pure force of social media, played any role in the shifting winds of culture, it is the mainstream that is to be blamed for becoming out-of-touch, not people like Scott. And precisely because of this recent NYT article, I am doubling down and defending Scott Alexander from his own self-accusations, voiced in the original post through the character Adraste.

That’s because I think Michelle Goldberg’s opinion piece makes a fundamental mistake by bundling Scott with Alex Kaschuta, a mistake that is representative of what I mean by “out of touch” when I refer to many mainstream intellectuals. I appreciate Alex for changing her stance so vehemently — I am sure she has faced a lot of backlash from her former coalition members for openly going against them. So this is not some sort of indictment of her.

But Scott and Kaschuta could not be more different in their intellectual evolution: while Scott was writing anti-reactionary FAQs back in 2013, Alex was inviting their figureheads on her podcast in an affirmatory manner: she was part of the neoreactionary movement, not arguing against them. The differences between her past ideology and Scott’s relatively constant one are miles apart. To put them together, just because they are both vaguely “anti-mainstream” shows a lack of curiosity regarding ideas. And here I am going contradict Ross Douthat yet again: it is not The Internet that enforces a Monoculture, it is the mainstream. And it’s precisely because of this Monoculture that the arguments against Trump coming from the mainstream felt somewhat lacklustre, whereas Scott, as much as one might disagree with him, has always been at the very least interesting.

So yes, I am doubling down on defending Scott (and by extension people like him3). Because it is exactly this blindness and lack of curiosity that has been the problem for a while in terms of intellectual elite thinking, not the “centrist edgelords”, as Scott refers to himself and related individuals.

It’s not like I agree with Scott on everything — there are a lot of points where I think he’s wrong. I could write an entire article about them. To begin with, I think he has started a movement, whether he likes to accept it or not, which has shot itself in the foot through bad aesthetics (broadly construed), despite espousing a lot of good ideas. I suspect it is these bad aesthetics that have led to so many vicious and often unwarranted attacks on rationalists. And because rationalism somewhat overlaps with an area I am interested in, progress studies, I am worried some ideas have been tainted through these bad aesthetics. I also disagree with his stance on autism: in my opinion, high functioning individuals with what one might call Asperger’s can still be considered to be severely impaired and depriving them of any name to rally around is not a good idea. I am much more on the side of in this regard. He has written about the topic in his book “The Age of the Infovore”, has a paper addressing autism from an economics perspective, interviewed important figures related to “the movement”, like Michelle Dawson — I recommend people check out this part of his work, that I think has gone unfairly less noticed. I also publicly disagreed with Scott on Lumina, a year ago. Anyway, the list of disagreements with SA could go on.

But the man is simply NOT responsible, in any way, of causing populism. On the contrary, he has created antibodies against illiberal thought in other intellectuals, as I explain in the following section.

The Great Man Theory of platforming

“The Great Man Theory” of history remains unpopular among the academic and journalistic elite. That is not surprising: it carries with it an implicit admission of inequality: some individuals are truly, monumentally more influential than others and it’s not just chance or faceless “systems” that leads to this: it’s something about that person’s character, whether innate or acquired. You see this a lot when people like Elon Musk are discussed: it’s not him who created the hard tech empire he has, it’s his employees, or luck or his father’s emerald mines. But while intellectuals certainly know things business/start-up people don’t, the reverse is also true. And almost everyone competent in the start-up world that I have talked to, sees Elon for what he is: a one in several billion talent at being a hard tech company builder. He has done it over and over and over and it cannot be down to luck alone. That does not mean one has to agree with his moral, politics or fatherhood skills, and indeed many of the people who think that he is a sort of Kwisatz Haderach of tech, do not regard him as a personal model.

In terms of where I stand on “The Great Man Theory of History”, I am somewhere in between: I think both impersonal forces, historical currents that are beyond anyone’s control *and* extraordinary individuals shape the future. After all, had Elon been born in the Middle Ages, even in a privileged family, his technical skills would not have mattered at all and indeed, he might have gotten his head chopped off for speaking against the King, given the personality he has. So luck did play a role: the luck to be born in the right era, to be able to immigrate to the US and so on. The luck to be in an environment where a disagreeable personality did not run the risk of literally getting you killed (as it used to happen in many authoritarian regimes). To achieve the success he has, Elon probably committed many acts of “treason”. Overall, it is the institutions built post Enlightenment, the liberal norms of the US and indeed the pool of talent that this society has created, that he could draw upon, which allowed him to create what he has. And I also believe that Great Men (and Women) should always strive to recognize this luck and strengthen, as opposed to weaken, the societies that allowed them to achieve such success — one might even call this a form of noblesse oblige.

And while when considering the “Great Man Theory of History”, many elite journalists/academics tend to assign “power” only to broad historical trends, systems or groups of individuals ("society”), they to do the exact opposite when it comes to something else, namely so-called “platforming”. They seem to almost believe in a “Great Man Theory of Platforming”, where hosting an event where you invite someone with “bad ideas”, or even choosing to politely engage in debate with them can alter the course of history, at least that of ideas, which are broadly regarded as upstream of all else, in very, very dangerous ways. That such “bad ideas” would spread anyway due to impersonal cultural forces and the decentralized nature of the internet seems, in this case, to be downplayed.

One of the main accusations against Scott comes from those who blame him for “platforming” neoreactionaries like Yarvin. How did Scott do that? Did he run events with Curtis as the headline speaker, heaping praise on him? No. To many, Scott seems to be guilty of not completely refusing to engage with neoreactionaries, something that stains Scott by association, despite him arguing against them.

To better understand this debate, I recommend an article from 2020 in the New Yorker, written by the author Gideon Lewis-Kraus. The piece discusses Scott’s “doxxing” by the NYT4, but also does a pretty good job at summarizing the interests of the rationalist community for a broad audience and brings up both the arguments pro and against Scott and the rationalists’ involvement in the spreading of neo-reactionary thinking. The paragraph below is particularly relevant, with the sentences written in bold and italic illustrating the main points of criticism against Scott that are often brought up by those who consider him guilty of “platforming” Yarvin:

Many rationalist exchanges involve lively if donnish arguments about abstruse thought experiments; the most famous, and funniest, example, from LessWrong, led inexorably to the conclusion that anyone who read the post and did not immediately set to work to create a superintelligent A.I. would one day be subject to its torture. Others reflect a near-pathological commitment to the reinvention of the wheel, using the language of game theory to explain, with mathematical rigor, some fact of social life that anyone trained in the humanities would likely accept as a given. A minority address issues that are contentious and at times offensive. These conversations, about race and genetic or biological differences between the sexes, have rightfully drawn criticism from outsiders. Rationalists usually point out that these debates represent a tiny fraction of the community’s total activity, and that they are overrepresented in the comments section of S.S.C. by a small but loud and persistent cohort—one that includes, for example, Steve Sailer, a peddler of “scientific racism.” (…) Alexander has long fretted over the likelihood that the presence of these fringe figures could tarnish the reputation of the blog and its community. In late 2013, he published “The Anti-Reactionary FAQ,” a thirty-thousand-word post now regarded as one of his first major contributions to the rationalist canon. The post describes the world view of a group, centered around a figure called Curtis Yarvin, also known as Mencius Moldbug, whose “neoreactionary” views—including an open desire for the restoration of feudalism and racial hierarchy—contributed to the intellectual normalization of what became known as the alt-right. Alexander could have banned neoreactionaries from his comments section, but, on the basis of the view that vile ideas should be countenanced and refuted rather than left to accrue the status of forbidden knowledge, he took their arguments seriously and at almost comical length—even at the risk that he might lend them legitimacy. Ultimately, he circumscribed or curtailed certain “culture war” threads. Still, the rationalists’ general willingness to pursue orderly exchanges on objectionable topics, often with monstrous people, remains not only a point of pride but a constitutive part of the subculture’s self-understanding.

Who knows? maybe without this “platforming” that Scott facilitated, we would have never had Curtis Yarvin. Of course, this is completely backwards. Elon Musk is much more of a Great Man of History than Scott Alexander is a Great Platformer of neoreactionaries, in the sense that the latter role probably had much less causal impact on Yarvin’s rise5. Musk’s tech empire would not have happened without Musk. However, Yarvin’s ideas would have penetrated anyway, sooner or later. After all, he was completely deplatformed, ignored, cancelled etc by most mainstream publications and academia, yet his ideas did penetrate to the highest level of US political thought in the end. This shows there was appetite for such ideas, an appetite driven by deeper forces, as well as technology (the internet). Yet, despite this nearly universal deplatforming, somehow, we are led to believe that, another at the time unknown, anonymous blogger (Scott) engaging in a dialogue that was mostly critical of Yarvin’s ideas, might have dramatically changed culture in very, very bad ways. When I say “we are led to believe”, I do not just mean through the Gideon-Kraus article — so many critiques of Scott that I have read ultimately mention his engagement with neoreactionaries as some sort of mortal sin.

My interpretation is that Scott had his pulse on where the culture was heading and which were the important battles way before most of the mainstream did. And if he changed the history of culture with respect to neoreactionaries in any way, it’s against, not pro them. Scott’s points 2013 defense of liberalism and The Enlightenment reverberates now across many younger intellectual’s writings defending these principles. I, myself, was influenced by his piece and ended up reading on the history of ideas related to this topic much more. Around one year and a half ago I got acquainted to the writings of Joel Mokyr, Deirdre McCloskey and others who defend the moral framework of liberalism as the basis of the current relative prosperity and good life that we are all enjoying — an awakening I wrote about in one of my first piece that really took off: “Ideas matter: how I stopped being a Culture Incel”.

And such engagement with the history of liberalism from younger writers is good. For the battle lines now are not between Grey Tribe and Blue Tribe or Red Tribe or whatever (which are often mentioned in the Gideon Lewis- Kraus article several times as the important factors and are stand-ins for rationalists, Democrats and Republicans, respectively). The battle lines of culture, at their very core, are between illiberalism and liberalism. Liberalism has grown stale, almost impotent, like an old car that still does its job but has sort of lost its shine. It has become associated to what Deirdre McCloskey would call “prudence and prudence only“, when she refers to the excesses of bourgeois morality (which she sees as necessary/ a precursor to liberal thought). Or, using another term of hers, “idiotically strategic”6. By contrast, neoreactionary thought offers the appeal of the grandiosity of the old, the prudence-stripped courage of the ancient hero — an aesthetic that for good reason appeals to young men, when the alternative is a sort of neutered form of “idiotically strategic” prudence. And when I say for good reason, I mean that in an explanatory fashion, not in order to suggest they are correct.

This has happened for many reasons, that go beyond the scope of this essay. But liberalism7 is, I am convinced, the best way forward and the question is not how to get rid of it, but how to re-inject it with vitalism. I do not believe that Scott, rationalism and associated movements like Effective Altruism (EA) were good at injecting this vitalism. If anything, EA makes liberalism feel more on the side of “prudence and prudence only”. But one man cannot do everything. If one does not see that Scott was clearly on the side of liberalism, identified its opponents from very early on and chose to do something about it, with the means that he had at his disposal (writing), then one does not understand culture at all.

Edit 2: many ppl have pointed out that actual edgelords are bad. I agree! I only refer to these people as edgelords because Scott, through his character Adraste, does so. I certainly don't consider Scott or Nate or myself actual edgelords.

Edit 3: There is an important question as to whom SA, through the voice of Adraste, refers to when he thinks of “centrist edgelords”. He clearly includes Scott himself there, but Brett Weinstein would probably consider himself a centrist too, and there is an ocean of difference between Scott and Brett.

Scott Alexander has a recent post that I actually recommend everyone reads, called “Twilight of the edgelords”. It’s written as a dialogue between two fictional characters and it’s basically a debate of the piece’s subtitle: “Should edgy heterodox centrists accept some of the blame for Trump?”. Its conclusion is somewhat left to the reader (which is why this piece is titled “a response to Scott Alexander” as opposed to “contra Scott Alexander”).

The argument debated here is whether what one might call “centrist edgelords”1, while motivated by good intentions, have been partially responsible for unleashing a catastrophic wave of populism madness that will tear the US apart. They did so by criticizing what one might call “liberal elites” in a rational way, hoping that we would reach a sort of middle-ground, rational consensus. Instead, we got disaster. Although Scott never explicitly states a definitive conclusion, my hunch is he is siding more with Adraste, whose viewpoint is effectively encapsulated by the following paragraph2:

We wanted to be able to hold a job without reciting DEI shibboleths or filling in multiple-choice exams about how white people cause earthquakes. Instead we got a thousand scientific studies cancelled because they used the string “trans-” in a sentence on transmembrane proteins.

We wanted to be able to prevent biological men with testosterone-boosted muscles from competing in women’s sports leagues. And somehow that morphed into a world where whenever someone with more subtlety than Attila the Hun tries to stake out a position on anything, roving bands of dead-eyed reply guys interject “HOW MANY GENDERS ARE THERE? BET YOU CAN’T ANSWER THAT HAR HAR HAR!”

Firstly, it feels like Scott has been reading a lot of , whom I agree with on a lot of points, but who I also believe is catastrophizing the situation (something I told him personally). I totally agree that a wave of unimaginable stupidity, racism, anti-semitism, hate against transgender people, misogyny and all the other stuff Scott talks about has been unleashed on X. It is very bad that so many people feel this way. But I have reason to believe, and this will become more evident towards the end of this post, that these sentiments will not become dominant among so-called “elites”. My prediction is that in the long term, they will remain confined to the moral gutters of society.

But, even assuming things do become bad, I do not think “edgelords” like Scott are to blame for this. It’s clear to me that if anyone is to actually be assigned indirect responsibility for future negative consequences, those would be previously miscalibrated “establishment elites”, who should have listened to the likes of Scott Alexander, and other “edgelords” sooner. Ezra Klein and Derek Thompson, who are no Trump supporters, have said as much in thier latest book, Abundance:

California’s problems are often distinct but not in their structure. The same dynamics are present in other blue states and cities. In this era of rising right-wing populism, there is pressure among liberals to focus only on the sins of the MAGA right. But this misses the contribution that liberal governance made to the rise of Trumpism.

My specialty is not politics/the kind of policies Derek and Ezra discuss in their book, but I have been writing about elite culture, especially as reflected in academia a lot over the past year or so. My overall opinion is that the elite consensus pre 2024 had gone wrong in important ways and that this is what is to be blamed for populism. I will go more into this in the next section, but first I want to make some very technical points about Trump getting elected that go against Scott’s theory. The first one is that most of the shift to Republicans has happened among politically disengaged voters. I do not mean to downplay the influence that Scott has, but I highly doubt these people have even heard of his name or other “edgelords”. These are normal people who were feeling something was going wrong.

It is true that the shift of tech billionaires towards the right probably had an actual electoral impact, and that can be partially blamed on “edgelords”. However, it was not edgelords who led to the extreme wing of the liberal elite (e.g. Taylor Lorenz) alienating tech through constant attacks — an extreme self-own from the liberal coalition. It is worth noting, however, that despite tech billionaires donating more to the right, Democrats still amassed more money in total donations. So, overall, they had resources *and* the main establishment institutions on their side. To me, it clearly looks like something had gone wrong among classical establishment elites, or what calls “The Village”. To blame the “edgelords” instead seems incorrect and not in line with the data we have.

Elites have a higher responsibility than most people

What I described above are to some extent technicalities that avoid the meat of the subject. Before getting into it, however, I want to sketch a bit what my views on elites are. Firstly, I am not a populist or an anti-elitist. I think domain experts and strong institutions that foster such experts and enable them to be in charge of key parts of running a good society are very important. I also take the view that some would consider cynical, which was articulated very well by , that democracy is in many ways a form of regulating intra-elite competition and not quite “direct power of the people”.

I simply believe elites have a higher responsibility than the rest of people and I think the health of a society depends in large part on their behaviour. I guess in some sense, I have a vision of old-school honour, where people who have any sort of influence or hold power also have more responsibility, because their decisions are more consequential. The choice of the header image for this post is that of Sansa Stark — whom I think George RR Martin intended to turn into a character that would embody such an elite3: combining the moral backbone and deep sense of honour of the Starks with the pragmatism she gained through her experiences in King’s Landing. This is why, although I am more disgusted by groyper twitter anons, I focus more on discussing and criticizing the behaviour of people who actually run institutions.

This broad view is reflected in my writings: on one hand, I have repeatedly criticized academics when they got things wrong, for example their reaction to the plagiarism scandal in which the former Harvard president Dr Claudine Gay was involved. I have warned that people like Dr. Holden Thorp, who was the Chief Editor of Science, one of the most prestigious science publications, were doing a lot of harm by using their positions to make explicitly political/policy related statements. This is back when it was highly unpopular to do so, especially from within academia. My warnings came true in a way that was worse than I expected (recent science cuts). While this makes me feel good about my predictive validity, which is ultimately all that a take-haver has to defend their honor, it does not make me happy about the state of science.

At the same time, I also counteracted populist and over-simplistic critiques of elites that aim to paint them as these cartoonish villains, through pieces like: “Shut up about luxury beliefs” or “Elites are mostly lazily well-intentioned”.

Such articles aimed to reflect that I believe when elites go wrong, it’s often due to ignorance/social-conformism, and the right way to approach this is to slowly goad them towards better views (e.g. by starting a new movement or writing edgelord blogs, which is what Scott himself does4), instead of taking a “burn it all” attitude that throws the baby out with the bathwater.

Having established this framework, I want to explain, point-by-point, why the “edgelords” were right to criticize existing elites:

1) It is axiomatically true that what Richard Hanania would call “Elite Human Capital” (EHC) will generally be better at governance than the masses.

2) It is also true that you will always have a normal distribution of EHCness in the population, which can be shifted to some extent via culture and good institutions (mostly influenced by the top EHC people, so here we have feedback loops).

3) At the same time, it's clear you can have better or worse elites and that the outcome of a civilization depends on their quality. They are the rate limiting step to things happening/how things happen! Assuming a constant EHC distribution across countries, some will be better than others, and I think that is largely due to how well these elite run institutions are designed and how well their purpose is executed.

4) Following from 3, having better elites should be a priority. So I do not think criticizing elites should be avoided, simply because there is a risk of populism taking over. This would suggest that we can never hope to advance in our thinking or better our institutions: we just have to take them for granted as they are, no matter how flawed, because the alternative is chaos and rule by uninformed people. Permanent stasis. And stasis actually equates to decay. So, if we take Adraste’s thesis seriously, we would end up with a progressively decaying elite that we never criticize for fear that something worse might take over.

5) It is also clear to me that elites and many institutions have become slightly lower quality in the last decade or so, due to a multitude of reasons. Western elites and institutions are still better than Eastern European ones and probably most other elites (I know this from personal experience), but that does not mean we should just allow them to become progressively worse.

6) Misinformation experts, blatantly political activists at the helm of scientific publications like Holden Thorpe and the like, did not "predict" Trump or populism. Rather, they caused these phenomena to some extent, through their smugness and by being fundamentally wrong. They gave ammunition to the braindead anti-wokes that Scott describes (the “HOW MANY GENDERS ARE THERE? BET YOU CAN’T ANSWER THAT HAR HAR HAR!” people).

TLDR: Honest intellectuals and others who can influence culture indirectly (e.g. via philanthropy) cannot simply sit still and allow elites to become progressively worse for fear that any criticism might lead to populism. In some way I see this as a form of extreme cowardice and defeatism. The spirit of the West is Faustian, which means it are seeking continuous betterment. Breaking some things is part of it. Of course, it should be followed by rebuilding. And rebuilding we shall do! In the next section I explain that what has happened recently in the culture has created a unique opportunity for a better elite, one that aims to build, both figuratively and literally.

The future

Now, the question is: what does the future hold? There are two possible narratives that can be envisioned starting from here:

1) a pessimistic one: maybe a slow influencing of elites to better views, or the middle-ground that I and other “edgelords” propose has been made impossible by social media. Maybe in the past elites policed each other behind more closed doors, so that only people who could understand the nuances of "Elites are mostly lazily well intentioned" had a say in the conversation. Social media may have become the equivalent of elites airing their dirty laundry in public and grifters taking advantage of it. So before someone like Scott can actually make elites better by convincing them, 100 populist writers will take advantage of the situation, spout some misinformation and move us irremediably towards populism.

Another force contributing to a pessimistic view is that it is highly possible intellectual elites today are lower quality, via expansion of higher education on one hand, and flight of very capable people to the private sector on the other. However, this latter problem is solvable by raising the pay/status of high-caliber intellectuals through concerted effort by individuals who understand this and can support it.

2) the optimistic one: populism serves as a "shock" to the elite system, without a catastrophic blow to democracy itself. My opinion is that now the Overton window has been expanded such that views like those held by so-called “edgelords” have become "normal" and "moderate". Ezra Klein and many Democratic figureheads have changed their tune in response to the Trump election. I suspect they were always more moderate than they seemed, but they have been emboldened in actually spelling out things they would not have said pre-Trump. They have become explicitly critical of the more extreme members of their coalition and it feels like individuals like Jeet Heer or Taylor Lorenz have become sidelined and proven wrong.

It might be the optimism of youth, but I personally believe much more in option 2. I think books like “Abundance”, which takes a pragmatic view to progressive issues and aims to convince liberals that virtue signalling and cancellation campaigns are poor substitutes for engaging with real problems like housing and climate change represent signals in this direction. This fits nicely with the emergence of the “progress studies” movement. I see a way towards a coalition of true liberalism that is focused both on building a better culture and dismantling the institutional barriers to creating material progress at the same time. In any case, there is an element of self-fulfilling prophecy to all of this: if “edgelords” meekly accept defeat, we might as well get 1. The future is ours for the making, and we might as well start working towards the second option.

I would also like to add I do not know if *this specific internship* is perfectly designed, but the good thing about it is that it widens the Overton window of what is currently possible. More companies could offer schemes that would allow access to what one might consider “high-end, prestigious jobs” to exceptional high school students without college degrees. And the companies in question do not have to be Palantir — they can be NVIDIA or Microsoft or Facebook. The ideal scenario would be one where different companies would try variations on the theme: for example, for those who got a job offer, one could imagine the company paying for some necessary training in parallel to doing the job (e.g. it’s probably hard to learn advanced Math entirely “on the go”). The point is that such experimentation is welcome and relatively low cost — it’s not like an internship at 18 will set back anyone forever and it’s probably going to be at the very least an interesting experience.

Ok… but how is this feminist?

I have been spending the past month going on about how we need to invest more in science that would extend women’s fertility window and how this should be a central tenet of modern feminism. This would allow more women to achieve two things that are currently at odds to some extent: success in “greedy careers” and motherhood. I have written about this originally at Works in Progress and then had follow-up articles on the topic here and here. The core of the argument, which many who have read my pieces already know, is that one’s 30s are at the moment decisive in a lot of prestigious careers that are demanding of one’s time, including business, law, medicine or science. At the same time, this is also the period when women’s fertility starts to decline, putting them in a race against time when it comes to achieving their goals.

Allowing women freedom to choose to have kids later with a high degree of certainty is one way of going around this. But there is another way, which could be done in parallel, and would benefit both women *and* men. That is working from the other end and wasting less of young people’s time, such that by their mid-30s, women (and men) are more advanced in their careers. It’s something I mentioned in a previous article:

Anyway, why do one’s 30s matter so much? Some of it might be societal, or the way we set up our institutions. Indeed, Dr. Arpit Gupta made the clever observation that we should do the following things as a society:

1. early graduation from high school

2. 3 rather than 4 year BA degrees

3. major in law or medicine up front; rather than in professional school

4. social shift to promote the young

A worrying trend in society has been that of credentialism and what many call “elite overproduction”. This has naturally been accompanied by a general increase in the age at which young people can expect to get promoted and reach various career milestones. For example, the age of getting a first R01 grant (which signals independence in a biomedical science career) in the NIH has been steadily increasing for the last decades:

Apart from a bunch of other negative social impacts, including on the science itself (which I discuss here), this also affects women in particular, because timelines to promotion clash even more starkly with their fertility windows! This is something almost nobody talks about, but as I also highlight in a previous post:

I wish I would have graduated much earlier from high school. I wish I would have started doing research instead of just going to uni courses much earlier. And so on. In fact, if I were to complain about how society “has wronged me as a woman”, it would be that it has treated my limited “fertility time” with extreme disregard. At each step of the way I was encouraged to “be patient”, do more training, told that “things will figure themselves out”, even when I wanted and could have speedrun through things. So I fully support Arpit Gupta’s proposals (which would be good for men too, not just women)!

Internships schemes like those proposed by Palantir play into Dr Gupta’s agenda. To be clear: they would not work for everyone. I am talking here about very talented high school graduates. But these are already the kind of graduates most likely to choose greedy careers anyway, so my argument is consistent. Imagine a precocious girl who gets to work on actual science at the age of 18 instead of 24! This would give her a 6 year advance that could prove quite valuable later on, when she decides to start a family.

This isn’t an outlandish proposition, and we already have real-world examples to prove it. Take Laura Deming and my friend, Lada Nuzhna, for instance—both left college thanks to Thiel Fellowships and have gone on to become prominent figures in the aging field. Laura Deming is a founder of one of the first Venture Capital firms that focuses on aging, The Longevity Fund, has done a lot in general for the field and now has her own company focused on cryopreservation! Meanwhile, Lada spearheaded and led Impetus Grants, distributing $34 million to aging-related research projects with an impressively fast turnaround—decisions in under three weeks—while prioritizing innovative ideas over credentials. In other words, funding high-risk high-reward projects and taking bets on relatively unknown researchers. Some of the projects funded have already demonstrated remarkable success. Lada herself has also started a stealth-mode company in the aging space.

This is not for everyone — and not all women will benefit from these. But I think it is worth taking young people’s time more seriously as a society and allowing those who are very talented to exploit their talents in unconventional ways! This would benefit everyone and women in particular.

As a side-note, this is not at all in my interest to say. I graduated with a degree from Oxford and also have just submitted a PhD thesis: it is in my direct interest to defend credentialism, not criticize it.