Note: this is non-exhaustive and almost certainly subject to some recency bias.

Fiction

The best book I read this year was unquestionably Tolstoy’s Anna Karenina (or “book of books” per Henry Oliver).2 War and Peace is next on the list, but I am pacing myself; I now understand what it means when they say that you never get to read something for the first time twice. In the interim, I have been working my way through many of his short stories: Kreutzer Sonata, Master and Man, and The Death of Ivan Ilyich were among my favourites. I also quite enjoyed “Tolstoy at Home,” a rather wonderful profile on the author from 1891.

Other classics that were particularly delightful reads included Dostoevsky’s White Nights, Albert Camus’ The Stranger (somewhat absurd to confront a character of such a mechanical interiority), and Emily Wilson’s translation of The Odyssey.

I also read some contemporary fiction. Han Kang’s The Vegetarian was by far the most memorable – despite dealing with rather grotesque (?) themes, it is very well paced. This is in contrast to Andrew Miller’s The Land in Winter, which is much slower. However, if you do away with the expectation of a quickly accelerating storyline, it becomes easy to appreciate the finer world-building where the author reveals so much about the anxieties and idiosyncrasies of postwar Britain.

Nonfiction

The first half of this year was spent almost exclusively writing my undergraduate thesis,3 where I looked into the tendency to ascribe aesthetic terms, such as beautiful or elegant, to mathematical discovery. For practical reasons, I focused on turn-of-the-20th-century France as my model case and approached the period with the hypothesis that there must have been some shared philosophical or practical motivators that explained (in part) the use of such descriptors by both scientific practitioners and artists. Some stuff I read during the process that was particularly fun and helpful includes Sue Rose’s The Secret Lives of the Impressionists and Linda Dalrymple Henderson’s The Fourth Dimension and Non-Euclidean Geometry in Modern Art. Related: Ulkar Aghayeva wrote a wonderful essay on what makes an experiment beautiful.4

Another project this year had included an end-of-the-year trip to China (which unfortunately had to be tabled at the last moment). In preparation, Abby ShalekBriski and I started a small discussion group. We read Dan Wang’s Breakneck, which was a very good primer – I particularly enjoyed Chapter 2 where he recounts the pros and cons of the ‘build, baby, build’ mindset through the lens of a bike trip across the country. Chapter 4, on the One-Child policy is perhaps the most informative part of the book and emotionally gut-wrenching to boot (see Hollis Robbin’s review). Patrick McGee’s Apple in China was also a delightful read with important context on how the country came to set up its manufacturing empire (also see the author’s ChinaTalk interview and, tangentially, Karina Bao’s unofficial translation of Morris Chang’s memoir). Lastly, on the China front, Jasmine Sun’s retrospective from her trip there was excellent, as is Afra Wang’s blog (and here is a list of the best China books of 2025).

I have also been doing a fair bit of India reading since moving back. This has included the William Irvine translation of Niccolao Manucci’s Storia do Mogor (Italian spy writes a firsthand account of his experiences in Mughal India) – I was tempted to pick it up after reading this review from the pSmiths. Ramchandra Guha’s Makers of Modern India and India After Gandhi also came highly recommended, and although I haven’t finished either yet, they have been quite enjoyable so far. On the shorter side, Soham Sankaran’s substack post on overcoming India’s technological cowardice is very comprehensive.

The last big grouping of my reading can be filed under the metascience tag. This includes Vannevar Bush’s Pieces of the Action, Jennet Conant’s Tuxedo Park, James Phinney Baxter III’s Scientists Against Time,5 and Brian Potter’s The Origins of Efficiency. I also read my fair share of blog pieces on the general theme of making science better: Karthik Tadepalli on why ideas aren’t getting harder to find, Eric Gilliam on Warren Weaver’s contribution to molecular biology, Stuart Buck on the need for crazy philanthropy, Elizabeth Van Nostrand on the (not so) boring part of Bell Labs, Brian Potter on accidental invention, and Alvin Djajadikerta and Laura Lungu on why science needs outsiders.

Misc

On the built environment: I enjoyed Benedict Springbett’s take that the product of the railways is the timetable. Coby Lefkowitz’s recent article on how New York killed culture is also great (and I learnt more about the city after a two-hour walk with him than in three years of living there), as is Alex Chalmer’s historical look-back on the French success with nuclear energy.

Abhishaike Mahajan writes perhaps the best technical biology blog out there. My favorites over the past year have included: endometriosis is an incredibly interesting disease, why you should work in biology, and cancer has a surprising amount of detail.

Other publications I admire include Karthik Tadepalli’s new literary substack (I don’t know how to define the genre, but it is stuff I can imagine one having fun writing) and Chen Chen Li’s Dendrite, an all-around excellent publication on neuroscience.

Earlier this year, I was contemplating a PhD in the history of science – and although academia has been tabled for now, I am still quite drawn to the prospect of spending my twenties reading beautiful books in beautiful places. Meeting D. Graham Burnett during a fateful visit to Princeton salvaged my sense of what that life could look like. He also wrote a very good take on AI in the humanities, from the perspective of someone deeply integrated within the academy.6

On fertility: Ruxandra Teslo made a compelling argument on how reproductive technology can help women narrow the career/motherhood tradeoff, while Phoebe Arslanagić-Little describes the many factors behind South Korea’s cratering birth rate.

Some final thoughts on books:

I quite enjoy reading about the mechanics of how someone does their job (Smrithi Sunil articulates it better than I do) and particularly liked Henrik Karlsson’s How I read:

“Of the roughly 300 books I start each year, I finish about 50. I skim a lot. Books are not sacred…Reading well is an endurance sport. I sometimes talk to people who want to become serious readers and so pick up Kafka’s The Trial or something like that—it is about as pleasant as running a marathon untrained. They often lose their enthusiasm for reading. By reading within my comfort zone, I gradually build up my stamina and pick up more and more references, words, and patterns of thought, bringing more and more literature into my comfort zone.”

I am a strong believer in abandoning books rather ruthlessly; this has become even more important since I began treating reading as my job (writing, the action, is secondary and makes up a small fraction of the time I spend on each essay). I have, therefore, built up something of an imperfect but functional system that lets me read without losing steam. The crux of it relies on doing away with annotation – introducing a writing supplement prematurely creates needless friction that almost certainly costs me both speed and interest. Instead, I rely on those skinny plastic post-it notes and tack them on the edge of whatever paragraph feels important (a color-coded organization emerges naturally during the process as I become cognizant of the general themes) and move on(!). Then it is light work to go back through the tabs – it’s quite freeing.

Lastly, I am very grateful to those who have read and engaged with my work so far. Please do email me (hiya [at] mundane [dot] beauty) if you have feedback, ideas, or reading recommendations for the new year!

With thanks to Abby ShalekBriski and Venkatesh Ranjan for feedback on drafts.

Cover Image: Samuel F.B. Morse, Gallery of the Louvre, 1831–33, Terra Foundation for American Art [Wiki Media Commons].

Yet so much remains the same. Relentless, bad quality construction of public works leaves the roads perpetually dysfunctional.2 Our always overwhelmed judicial system is undergoing what appears to be an unstoppable breakdown.3 And perhaps, most consequentially, the air is poison.

Less than two centuries ago, many deaths fell under the umbrella of miasms: the notion that “bad” air was killing people4 through mysterious vapors rather than infectious viral and bacterial agents. This led to suboptimal methods of dealing with disease and caused an excess of deaths; yet today it might be relevant to re-evaluate the idea (especially in India) because the composition of the air, now a cocktail of pollutants, is increasingly making us sick.

Those living in the country know this and there is a barrage of headlines and social media posts documenting and complaining about the environment year in and year out. Yet, there is little initiative towards making things better and a defeatist attitude has become the status quo: why?

For one, I think it would be painful to the voting population if the government pulled and enforced policy levers on the causal factors of pollution. Quickly curbing our reliance on thermal energy, definitively phasing out old automotives to deal with vehicular emissions, or stringent enforcement of the crop residue burning ban will make things more expensive in the short run – it makes sense why there is a lack of political will to make real change. Therefore, as I contend, the first step to dealing with the problem is to get people on board and doing so requires good data5 – numbers that will help us forecast the human toll of air pollution and demonstrate the very real loss of life that chronic exposure produces.

Subscribe now

Some health threats (those that are causal in nature) are relatively straightforward to quantify – for example we can reasonably trace malaria deaths and calculate lives saved through drug programs, vaccines, and mosquito nets. Air pollution mortality, however, presents a far more complex measurement challenge and accurate assessment requires multiple layers of reliable data: consistent air quality monitoring, particulate concentration values, comprehensive death registration, and detailed cause-of-death attribution. As I have recently discovered, India currently lacks robust systems for each of these components.

As the adage goes, “what gets measured gets managed.” Conversely, what is poorly measured is poorly managed – that is how we get downright hazardous air in India.

This may seem harsh, but for everyone with eyes to see (and a nose to breathe), it is a palpably true statement. One that becomes all the more striking as we barrel towards the colder, dryer months of the year, where large scale crop-burnings, combined with the lack of rain (which tends to “settle” particulate matter), leaves the country with a hazy smog that refuses to lift.6 This is only compounded by India’s continuous reliance on coal: per the PIB as of June 2025, thermal power accounts for 50.52% of India’s total installed energy capacity, with coal alone contributing over 91% of total thermal energy.

All these factors, when taken together, leave people in some of the most populous parts of the world with air that is unsuitable to breathe. However, because the monitoring infrastructure is woefully inadequate we fail to even measure just how bad things are.

Note the difference in scale on the left and right hand y-axis respectively in the graph.

“Upon complying with the direction, India should operate 2312 monitoring stations covering approximately 502 cities across 26 states and five Union Territories (UTs). As of September 15, 2022, India possesses 1266 ambient air quality monitoring stations, which is approximately 54% of the actual target…NGT has directed that these monitoring stations will have to ensure the bare minimum requirement of operation. The degree to which Indian states and cities complied with this directive differs. Only 20 stations have been found to be complying with the direction, and 136 partially complied.”

We need good monitoring so that we can produce the composite standard measure: the Air Quality Index (AQI) which is a numerical scale that translates pollutant concentrations into a single number to communicate health risk. Different countries define the scale in their own ways, but as a general rule of thumb, it usually caps out at 500 and is broadly broken down into the following categories: Good (0-50), Satisfactory (51-100), Moderate (101-200), Poor (201-300), Very Poor (301-400), and Severe (401-500).7

This aggregate number is calculated based on the concentration of several different types of particulate matter including PM2.5 and PM10 (which are measures of diameter in micrometers). PM2.5 is defined as fine inhalable particles that can reach deep within the lower respiratory tract because of their miniscule size. They can be composed of “inorganic elements, water soluble ions, elemental carbon, organic carbon, and [other] organic compounds” that can contribute to a host of health issues.

The World Health Organization updated its air quality guidelines in 2021, recommending that annual mean PM2.5 should not exceed 5 μg/m³ and the 24-hour mean should not exceed 15 μg/m³, while annual mean PM10 should not exceed 15 μg/m³ and 24-hour mean should not exceed 45 μg/m³. India’s standards however are far more lenient (most countries have their own standards which is a separate issue)8 and the limit for 24-hour mean PM2.5 concentrations is set to 60 μg/m³. This means India allows four times what the WHO guideline does over the same time frame.9 Even with such relaxed guidelines however, as the graph below demonstrates, the country still struggles with keeping pollution in check.

The second problem with accounting for pollution related deaths (besides the fact that we have issues with both the pollutant thresholds in use and with state capacity for measuring air quality) is that we are also bad at recording mortality. We don’t actually know with any precision what kills most Indians – and if we aren’t aware of what people are dying from, we will struggle with quantifying the real burden of air pollution.

For context, India runs civil (CRS) and sample (SRS) registration systems. CRS is an account of births and deaths, while the SRS is a more detailed demographic survey that is then regionally extrapolated to give whole population estimates. In theory, these two systems should roughly align, but this seldom holds true. For example, as Nandlal Mishra writes in Data for India:

“The level of suggested under-reporting in death registration by the CRS varies substantially from state to state, with only a fifth of deaths as estimated from the SRS registered in Manipur, for instance, in 2019. Among the major states Bihar reported the lowest level of death registration with every second death not registered with CRS. The level of under-reporting is also higher in rural areas and among women.”

Clearly there are big gaps in the data but even if we decide to take the numbers from the CRS at face value, we are still left with incomplete information. The graphs below from the 2023 national report underscore the problem: not only is the data on death registration incomplete (left), even the information we do have arrives with such latency (right) that we almost certainly lose important details about cause of death.

Besides raw numbers we also need data on what people are dying from: is it a cardiac event or cancer? Was it an accident or the result of a long-term disease? However, per a cross sectional study from 2015, “The registration data were inadequate for a robust estimate of mortality at national level. A medically certified cause of death was only recorded for 16.8% deaths registered…On death reports, the columns for recording information on specific factors – e.g. pregnancy, smoking and alcohol use – were usually left empty. Even if such information was recorded on a death form, there was no space for it on the corresponding death register.”

I flag all of these problems because inaccurate and incomplete data seriously impedes our ability to understand air pollution deaths. Researchers estimate mortality from a given risk factor using what is called the “population-attributable fraction” (PAF), which tries to answer a specific counterfactual question: “suppose every member of a population who was not in the most favorable level of an exposure or some other condition or event with regard to an adverse outcome had been shifted into that level. By what proportion would the entire population’s rate, hazard, risk, prevalence, or caseload have been reduced?” In this discussion we could frame the question as such: if we could reduce everyone’s PM2.5 exposure to some target level (say, the WHO guideline of 5 μg/m³ annually) by what proportion would we reduce the total mortality rate in the population?

Answering this question requires a solid baseline of how many people are dying and from what. Nevertheless, the estimates that researchers have made don’t particularly invoke confidence.

Share

The most widely cited estimates come from the Global Burden of Disease (GBD) study, which attempts to systematically assess mortality and disability from major diseases, injuries, and risk factors worldwide. For 2019, the GBD estimated that air pollution (both ambient and household) contributed to approximately 1.67 million (or 17.8% of all) deaths in India.

However, this figure requires careful interpretation. Unlike deaths from discrete events like car crashes or drowning, air pollution acts as one of multiple straws that can eventually break the threshold for disease. A person might be simultaneously exposed to air pollution, tobacco smoke, and occupational hazards, all contributing to their eventual death from cardiovascular disease or cancer. This means the same death could theoretically be prevented through multiple interventions – cleaner air, smoking cessation, or workplace safety improvements.

A better metric is found in DALYs (Disability-Adjusted Life Years), which capture not just premature death but also years lived with disability caused by air pollution exposure. For India, the burden is a staggering estimate of 62.7 million DALYs annually which shows just how badly persistent exposure to bad air can degrade quality of life for millions. Moreover, if the baseline data was better, we might be seeing worse numbers on the board – excess deaths and suffering, real loss of human life.10

This is a sorry state of affairs and an exceedingly difficult problem to solve, but we must not forget that there are things we can do to move the needle in the right direction. Doing so requires big policy wins such as reducing our dependence on coal for energy or cracking down on agricultural stubble burning, but it also needs people to truly care about the air they breathe – and that begins with better measurement.

With thanks to Abby ShalekBriski, Alexandra Danylysyzn, Grant Mulligan, Pranav Aggarwal, and Venkatesh Ranjan for feedback on drafts.

Cover Image: Camille Pissarro, Morning, An Overcast Day, Rouen, 1896, oil on canvas, The Metropolitan Museum of Art, New York, Public Domain.

Note: In an effort to hold myself to a higher standard, this blog will soon have an errata page. In the meantime, please comment or email me on hiyajain[at]substack[dot]com in case you see an error and I will respond/issue a correction (with attribution) as appropriate. Thank you for reading my work!

I spent a few days mid-October traveling to New York and the Bay Area. It was an all around delightful time and I managed to squeeze in six new books into my suitcase back home, a problem that Lauren Gilbert summarizes rather accurately:

This trip also proved to be very generative and it seemed worthwhile to compile the links, books, and ideas I collected –

Day 1 was spent reacclimatizing with the New York City public transit system (which truly is one of the best parts of the city despite its many woes) before eating El Salvadorian pupusas in the Bronx.1 Over dinner, I caught up with a friend who has just started as an investigative reporter for the City Journal – the mayoral election makes it a particularly interesting time to write about New York and Adam is finding great stories on issues of urbanism (see, for example, his work on Housing Court Hell).

Day 2 and 3: I mostly hung out around the Columbia campus and enjoyed being a student again.

Surprisingly, many of my conversations with academics included the rising threat of AI in higher education – professors have now fully come to terms with the fact that they cannot reliably discern between student vs machine written essays. Work arounds have included assigning significantly lesser reading (no more full books), the addition of one-on-one spoken conversation as a form of assessment, and a lot more closed-note, blue-book exams. I don’t have judgement to pass, but I am going to leave the rather infamous post from the Atlantic here: the elite college students who can’t read books.

Among the student body however, there was a strong anti-AI sentiment (despite the fact that almost everyone relies on ChatGPT for schoolwork) which seemed downstream of anxieties for the future and uncertainty surrounding the job market, climate change, and energy scarcity. Personally, I think much of this milieu can be tracked back to the anti-tech sentiment in mainstream media (see recent example ).2

Besides arguing (in good fun) with friends and faculty alike, I got to pick up my copy of Mark Mazower’s On Antisemitism: A Word in History. It is an etymological treatment of history that demonstrates how two centuries of rapid change for the Jewish community has seen the meaning of the word ‘antisemitism’ evolve to fit the times. (This is perhaps my favorite review of the book).

Day 4 involved a very interesting side-quest in Templeton, CA.

Over the past three months, Aishwarya and I have been following the story of Peter Putnam, a scientist lost to history: he was a contemporary of Einstein’s and a student of John Wheeler, someone who very much should have been part of the scientific canon but isn’t. His work seems quite interesting too, and from what we have been able to put together so far, it is reminiscent of modern day complexity science. Despite his relevance however, he died as an obscure janitor and his multi-million dollar estate was donated to the nature conservancy.

Getting this work to the public eye seems important and we are trying to find, contextualize, and publish Putnam’s ideas for the modern reader. To that end, we made a four hour drive down from San Francisco to meet a documentary filmmaker and former student of his, Barry Spinello. Now in his late 80’s, Barry has held onto important papers and hours of investigative footage on Putnam for decades. He has very graciously handed us these tapes and we are in the process of getting them digitized – hoping to use the audio as a guide for the actual science.3

If you are a biologist or physicist who would like to read, decipher and comment on Putnam’s papers, or are an individual interested in funding this project, please reach out to Aishwarya.4

Share

Day 5-7 were spent in Berkeley at the Progress Conference. This is where I did most of my book acquiring:

• Yuen Yuen Ang, China’s Gilded Age: Following Dan Wang’s Breakneck, which truly is one of the highest impact books of the year, I wanted to do more reading on China. This one came recommended from Michael Hill and talks about how a certain flavor of corruption propelled China further. I am almost certainly going to continue reading about the country and will be using Afra Wang’s China Tech Canon as my guide.

  • Tangentially: Dan Wang also made a quip during his talk on how “made in China” will become a mark of quality. My own perception on this is divided and it might make for a good future post.

• Brian Potter, The Origins of Efficiency: This one is self explanatory, Construction Physics is one of my favorite newsletters.

• Tyler Cowen, Stubborn Attachments: I am a strong believer in judging a book by its cover, and this one does not disappoint on either front.

• Donald E. Stokes, Pasteur’s Quadrant: Basic Science and Technological Innovation: This selection is mostly a result of my love for Vannevar Bush’s Pieces of the Action, and it digs deeper into the sort of scientific research that leads to technological progress. [I have written two essays inspired by this concept linked here and here].

• How Farmers Raise their Kids and Other Essays, an anthology of posts by this year’s Roots of Progress Fellows – I had a fantastic time during the program which was spent in conversation with a remarkable set of individuals! (It also resulted in many good essays and I have linked but a few of my favorites below):

  • Abby ShalekBriski, Degrowth and the Death of a Party Grape

  • Benedict Springbett, The product of the railways is the timetable

  • Étienne Fortier-Dubois, The Color of the Future

  • Karthik Tadepalli, The Oracles Gift

  • Lesley Gao, How a $1 Lighter Defied Inflation for 20 Years

  • Nehal Udyavar, The Hunt for Huntington’s

  • Sam Enright, How can science get more replications?

  • Smrithi Sunil, How do we diagnose Alzheimer’s Disease?

There are other books that I came across and wanted to read but for lack of suitcase space, they will be arriving via amazon later this month:5

1. Kenneth M. Pollack, Armies of Sand: The Past, Present, and Future of Arab Military Effectiveness

2. Ramchandra Guha, Makers of Modern India

3. Jennifer Pahlka, Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better

4. William C. Harris & Peter F. Mackey, How to Change the Future: Lessons from Ireland on Revolutionizing Scientific Innovation and Economic Prosperity

Cover image: Asher B. Durand, Progress (The Advance of Civilization), 1853, Virginia Museum of Fine Arts, Richmond.

My dad and I spend many long, lazy afternoons chatting about the mundane – in the still quiet of my childhood home, there are few other things to do. It is in these moments that he doles out little pieces of (always encouraging) advice, as fathers tend to do, even if it’s just to fill the silence.

“The world is your…shell,” he proclaimed sometime back. Naturally, I asked if he meant oyster? Oysters like the sea creatures that are farmed fresh, cracked open on a bed of ice, and served on tall towers of silverware with wedges of lemon. It was a delicacy, I explained, an expensive one that was seldom eaten in our landlocked hometown. One cracks open the mollusk, garnishing the gelatinous creature before gulping it down whole.

He recoiled at the image, now desperately searching for a different analogy, but I had made his point for him. When you are twenty-something, the world does feel conquerable, beckoning you to lift it off its platter and into the palm of your hands. It fits rather conveniently and you are all but set to greedily devour its flesh.

I had felt this way before, spending three exceedingly happy years in New York untroubled by the biting cold of northeast winters. I read incredible books in beautiful libraries, ran into my friends all the time, and worked on esoteric projects – in a sense, I was invincible.

Then, mid-November last year, my favourite person in the world, my dad, got cancer. Incurable cancer. Suddenly, time that seemed infinite collapsed into a matter of years. That night, in my dreams, the oyster grew larger and larger till I was cradling it with both hands, trying desperately to hold on. I was certain that it would soon become crushingly big and, without dad’s steadfast presence, I would be left carrying its weight all alone.

So I did what anyone in my position would do, I packed up my bags and moved back to India.1 There was no plan but to stretch the thread of time, to talk about oysters big and small.

A year on, I wish I could say that we are out of the woods, but treatment for multiple myeloma (MM), as I have learnt, is a marathon. That is to say, this piece was not written with the intention of post-chemo catharsis (in fact, putting this out there is incredibly uncomfortable), but to think through cancer in an undetached manner.

Because it truly is one thing to read about grade-3 neutropenia2 as a potential side effect of the drugs and another to scrub your hands raw with disinfectant lest you bring home a usually harmless variant of the common cold. Owl Posting put it in rather beautifully grim terms:

“At some point you surely have to understand that you have been, thus far, lucky enough to have spent your entire life on the good side of medicine. In a very nice room, one in which every disease, condition, or malady had a very smart clinician on staff to immediately administer the cure. But one day, you’ll be shown glimpses of a far worse room, the bad side of medicine, ushered into an area of healthcare where nobody actually understands what is going on.”

With Myeloma it is a race against time (albeit a rather successful one thus far), to find novel therapeutics that keep you away from the “bad side of medicine.” Over time the malignancy slowly becomes refractory to whatever drug regimen worked previously and when the cancer inevitably relapses, you are left scrambling for something new.

But drug discovery is a relatively slow process. Although I have known this for a while, these days when I scroll through StatNews, there is a different sort of urgency on my mind – I am both in awe of the progress that is to come in but a few short years and saddened thinking about all the people who won’t make it that far, those who are already condemned to a place where medicine has no answers.

It is very true that the mountain of stuff we do know about disease and medicine is but a spot on the horizon when compared to the chasm of ignorance that surrounds it. Moreover, there is an ominous certainty that all of us will fall off the cliff at some point and spend our days grappling in the dark, hoping to make it back to the peak.

With cancer, each year about 10 million individuals inevitably lose their foothold on the mountain of medical miracle. That is a big number, and it’s difficult to truly wrap our heads around such suffering but we must realize that there are real people who make up that figure – people that you and I may even love.

Then, to exhaust the analogy, should we not do everything possible to try and stay on the mountain? Should we not work to test for and identify illness before the rope snaps off, to identify the durability of our gear, to hang on to the good side for as long as possible?

Most patients with multiple myeloma are diagnosed quite late.3 This is because, while the disease eats away at your bones and weakens your immune system by crowding out healthy blood cells, there is no real mortal danger in the short term. Instead, you are hit by run-of-the-mill symptoms, such as aching bones and fatigue. What person over 50 hasn’t complained of being tired?

In fact, it can take over 6 months4 of chronic pain to receive a diagnosis for MM by which time almost 9 percent of individuals are hit by cancer-related fractures. This is preventable suffering.

My dad, thankfully, did not experience this and today he has favorable prognostic outcomes solely because he was diagnosed early. Because we got lucky. Because his position in the world, as a doctor, helped him to identify and verbalize the difference between muscle pain and bone pain. This understanding ultimately allowed him to use the resources at his disposal and slip into the MRI machine on his lunch break to scan for the mild back pain he was experiencing. Had it been anyone else with his presentation, it would certainly have taken months longer. Months where the vulturous latent tuberculosis in his body would have wreaked havoc on his compromised immune system and the conversation here would be significantly different.

Unfortunately, this is a reality for patients with many other types of cancer. Moreover, these are often malignancies where diagnosis delays due to vague symptom presentation have much graver consequences. For example, with ovarian cancer, tumor growth that is “confined to the ovaries (stage I) can be cured in up to 90% of patients, and disease confined to the pelvis (Stage II) is associated with a 5-year survival of 70%. However, disease that has spread beyond the pelvis (stage III-IV) has a long-term survival rate of 20% or less. Only 20% of ovarian cancers are currently diagnosed in stage I-II.”

Pancreatic cancer is even worse: “The small percentage of patients (10–15 %) diagnosed to present with resectable PC, primary surgery followed by adjuvant chemotherapy is the only possible way they may survive long term. However, because the disease presents with minimal symptoms, 40–50 % of PC patients receive a diagnosis of locally advanced and incurable forms. An additional 40 % would present with stage IV metastatic disease at diagnosis.” To put it bleakly, the five-year survival rate for this cancer is ~12%. Those terrible odds are the best we can currently offer because most pancreatic malignancies are caught too late.

These deaths can be prevented – we have known for a while now that early detection drastically increases survival odds across a variety of different cancers. Moreover, we can see this works with the case of breast cancer, which is perhaps the most well-known success story for early detection. For women diagnosed with localized breast cancer, the 5-year relative survival rate is an exceptional 99%. Due to effective screening programs like mammography, approximately two-thirds (66.5%) of all breast cancer cases in the United States are diagnosed at this highly treatable stage. However, for the subset of women whose cancer is not detected until it has reached a distant stage, the 5-year survival rate falls to just 32.9%.

Such examples tell us that perhaps the primary determinant of mortality is not necessarily the organ of origin but how far the cancer has metastasized – treatment becomes exponentially more challenging once a tumor has seeded itself in distant organs like the lungs, liver, or bones. Therefore, any technology that can reliably and systematically detect cancer before this metastatic cascade occurs is as necessary for true cancer control as are effective late-stage therapeutics.

However, as we have seen, diagnosis is difficult especially in cases where symptoms are inconspicuous and non-invasive, inexpensive testing unavailable.5 It is both unfeasible and unpleasant, for example, to have at risk patients undergo annual PET scans to catch early stage tumors. Instead, what we need is ubiquitous blood based testing.

“Most cancers are not screened because their prevalence in the general population is too low to justify screening programmes on an individual cancer basis. However, the number of people that need to be screened to detect cancer goes down when prevalences are combined; that is, when multiple cancers are screened for in a single test. Because all cells in the body have access to the circulatory system (directly or indirectly), blood is an attractive analyte for a multicancer test.”6

There is encouraging progress in this direction with Multi-Cancer Early Detection (MCED) tests which are a promising way to detect cancer’s molecular signature in a simple blood draw, long before symptoms appear. These liquid biopsies work by analyzing cell-free DNA (cfDNA) that tumors shed into the bloodstream, examining the DNA methylation patterns (chemical tags that regulate gene activity and serve as a unique fingerprint for different cancer types). More importantly, this method can simultaneously screen for dozens of cancers, including ovarian and pancreatic cancer that currently lack effective early detection methods.7

GRAIL, in particular, has developed the Galleri test which is the most clinically advanced MCED on the market, and can detect signals from over 50 cancer types. In a paper from 2020 the company writes, “In the validation data set, the multi-cancer early detection test had a false positive rate of 0.7% and an overall test sensitivity (true positive rate) of 54.9%. Sensitivity… for all cancer samples with known tumour stage were: stage I (n = 185), 18%; stage II (n = 166), 43% ; stage III (n = 134), 81%; and stage IV (n = 148), 93%.”8 There is also a larger randomized controlled trial in the works with the NHS, with 140,000 participants, and is expected to report results in 2026.

Yet despite this promise, MCED tests face resistance from segments of the medical community. The concerns are familiar echoes from previous screening debates: What about false positives causing unnecessary anxiety and procedures? Won’t these tests detect indolent cancers that would never have caused harm, leading to overtreatment? Where is the proof they actually reduce mortality? And aren’t they too expensive to be equitable?

These are not unreasonable questions, especially following the case of the PSA [blood] test for prostate cancer, for instance, which led to widespread overdiagnosis of slow-growing tumors that would never have threatened an (older) patient’s life. Critics also argue that stage-shift (diagnosing cancers at earlier stages) is an insufficient endpoint over true mortality benefit. And overall it would be fair to hold the opinion that the technology is yet to finish maturing: here, the results of the NHS-Galleri trial will help accurately pinpoint false-positive rates and real benefits from early detection.

However, in the meantime, I strongly believe we should avoid the kind of extreme rhetoric that has emerged in some quarters. This comment, for example, claims parallels between GRAIL and Theranos, suggesting that investor centered hype is leading us down a similar path of corporate fraud and collapse. They then go so far as to call for Galleri’s complete withdrawal from the market, arguing that until we fully understand its capabilities and potential harms, it should not be available at all.9

Such writing reads as short-sighted and biased, and most importantly takes away agency from individuals. If my dad had been screening for his m-protein levels regularly (blood based test here too) a decade ago, we would have known he had MGUS, the harmless precursor to myeloma. We would have monitored things closely, and when he progressed to Smoldering Multiple Myeloma (SMM), the more indolent version, we would have been ready. There would never have been symptoms, but about half of all SMM cases progress to active disease within 5 years. We would have had time to plan. Time that we did not have last November, when my parents kept his diagnosis from me for two weeks while I applied to PhD programs – two weeks of normal-sounding phone calls so that I could selfishly concentrate on myself, unaware that my world would break apart.

I also bring up SMM because it helps defeat the argument that early diagnosis is pointless without treatment options. For decades, the standard approach for patients who (by some stroke of luck) were diagnosed with the precursor disease was “wait and watch,” regular monitoring so that medical teams could begin treatment if and when symptomatic disease with organ damage, bone fractures, and kidney failure, arose. But in December 2024, the AQUILA trial demonstrated that early intervention with daratumumab (a monoclonal antibody) significantly lowered the risk of progression (to full blown MM) and death compared to active monitoring alone. Meta-analysis shows that early treatment produces a 60% reduced risk for disease progression and a 45% lower risk of death.

Similar success can be replicated across other cancers if we identify high-risk patients early. Moreover, early diagnosis can also help reshape pharmaceutical incentives to make prevention profitable. The larger economic case follows naturally. Treatment costs for late-stage cancer patients are up to 7 times higher than for those diagnosed early and per the WHO, early-stage treatment is 2 to 4 times cheaper than treating advanced disease.10 For colorectal cancer alone, “increasing screening prevalence to 70% among adults aged 50-64 could reduce Medicare spending by $14 billion by 2050.” All of this is not even accounting for the economic opportunity cost of getting and treating cancer for the individuals at the center of it all.

In every respect, I believe that knowing is always better than not knowing. There are many who will share this opinion with me and for them, electing to take something like the Galleri test might still be worth it despite its perceived short-term drawbacks. Patients can make informed decisions about imperfect information – they should have the choice to do so, it would be deeply paternalistic to take that option away from them.

Early detection is not a luxury. It is the difference between cure and palliation, between manageable treatment and devastating (co)morbidity, between lives saved and lives lost. It is the difference between having a decade of good years with my dad and not.

Living in a world with cheap, non-invasive, ubiquitous diagnostics would be one where everyone had the opportunity my dad did and not through privilege or luck. This is the true purpose of technological progress: to give people the crucial ability to make informed decisions about their body, to try and live a good life despite peaking at the bad side of medicine.

With much gratitude to Emma McAleavy, Mike Riggs, Abby ShalekBriski, and Samarth Jajoo for their measured feedback and unbounded support.

Note: I will be in the Bay Area for the Roots of Progress event later this week (16-19th Oct). If you will also be around then, free to reach out and we can get coffee!

Edited for typos.

This is a cross-post. It was first published online here with the Good Science Project on October 10th, 2025.

Did penicillin help win WWII? Perhaps not independently, but for the estimated 100,000 Allied soldiers treated between D-Day and German surrender, it must have felt like a miracle. It still remains one, and today penicillin and its derivatives are amongst the most widely prescribed antibiotics globally. However, when we talk about penicillin the wonder drug, we often sidestep the fact that bringing it to market took an incredible amount of effort, and without the immense scientific coordination required to scale it, penicillin may have arrived much too late.

In other words, the story of penicillin is more than a tale of scientific luck; it’s a reminder that breakthroughs depend on how we fund and organize science. In the 1940s, penicillin succeeded not because a lone scientist made a brilliant discovery, but because government, industry, and academia worked together with urgency and coordination. Today, we have few equivalents. The United States spends roughly $50 billion a year through the NIH, but our system often rewards incremental proposals over ambitious ones and buries investigators in administrative overhead. The lesson of penicillin is that scientific miracles may often be accidental in their origins, but the ultimate impact depends on larger systems designed around implementation.

If wartime America could marshal its industrial and scientific capacity to mass-produce a fragile mold, then peacetime America should be able to modernize its research institutions. That means creating space for risk-taking inside the NIH, strengthening programs like ARPA-H that back high-risk, high-reward ideas, and building new pathways – public, private, and philanthropic – that can translate discoveries into real-world impact.

The [penicillin] mold is as temperamental as an opera singer, the yields are low, the isolation murder, the purification invites disaster. Think of the risks!”

Such was the conversation in the Pfizer board room in the early 1940s as the company, a market leader in fermentation, grappled with the longstanding problem of penicillin production. Yet, amidst the turmoil of World War II, they decided to take the gamble anyway. “On March 1, 1944 Pfizer’s penicillin plant opened. It contained fourteen 7,500-gallon tanks; soon the company was producing five times more penicillin than originally estimated…most of the penicillin that went ashore with Allied forces on D-Day came from this plant.”

Yet industrial machinery is not what comes to mind when we think of penicillin, instead the familiar narrative in the annals of science history goes something like this: In the fall of 1928, Alexander Fleming was tidying up his laboratory at St. Mary’s Hospital in London when he noticed something peculiar about one of the petri dishes. A spot of greenish mold had contaminated a culture of staphylococci and had caused the once-thriving bacterial colonies to dissolve into “faint shadows” of their former selves. Species of mold that naturally produce penicillin are rare1 – what are the chances that one would grow in a tiny culture, and in the lab of a man who was actively on the lookout for substances that could destroy disease-causing bacteria without harming patients? Fleming would later go on to write that he “thought it should not be neglected” in what may be one of the greatest acts of foresight.

Much like the apple that just happened to fall on Newton’s watch, the simple story we tell about penicillin says that a zero-to-one breakthrough arrived in an act of fate. But attributing this world-changing discovery to serendipity alone ignores the incredibly difficult work that followed during the tense years of WWII, without which Fleming’s lucky break might not have mattered at all.

For over ten years, the antibiotic, one of our first effective defenses against the largest cause of human mortality, languished in the lab. While we romanticize the discoveries of lone scientists, history also shows that a single researcher cannot scale their work from the bench to the factory unassisted.

Prologue: Fleming

Before the so-called “golden age of antibiotics,” doctors typically carried morphine to ease pain, digitalis to strengthen failing hearts, and iodine or phenol as topical antiseptics. These were useful remedies, but none of them could strike at the root cause of infection.2 Thus, the great killers of the age remained as lethal in the 1920s as they had been in the 1890s: “Staphylococci, streptococci, gonococci, even pneumococci and meningococci, were practically as gay [happy/thriving] as they had been in the [eighteen] nineties.” Septicemia, or blood poisoning, for example, killed four out of five patients who contracted it, and pneumonia claimed roughly 30 percent of its victims. In return, the only prescription medical textbooks could offer was “good nursing care and general supportive treatment.”

It was in this milieu that Fleming had come across penicillin – could this be what Paul Ehrlich had called the “magic bullet,” a drug that killed a specific organism while leaving the host’s tissue intact?3 Prepared by his earlier work in the study of antimicrobials, Fleming did most things right. He isolated the mold, tested it against various bacteria, proved it nontoxic in animals, and even predicted its potential as an antiseptic. But then he stopped.4 The chemical challenges of producing penicillin in quantity – instability, fastidious growth requirements, and extraction difficulties – proved insurmountable, especially in 1929, given the limited technology and resources available.5

Soon after, however, in 1932, a new class of sulfa drugs was discovered. Derived from a particular type of dye, these sulfonimides were successful in treating a small group of infectious diseases, reducing mortality for some types of pneumonia and meningitis by two-thirds. They certainly were not a cure-all, being ineffective in cases of syphilis and against a majority of gram-negative bacteria, staphylococci, and certain streptococci. Moreover, they came with a slew of side effects, including fevers, low red and white blood cell counts, blocked kidneys, and hampered coordination. Despite these drawbacks, sulfonimides were some of the only effective treatments available in the 1930s. With limited options, when WWII broke out, “a package of sulfadiazine tablets and of sulfanilamide powder was added to each [American] soldier’s first-aid packet and he was instructed, on becoming a casualty, to swallow the tablets and sprinkle the powder on his wound.”

Florey and Chain in Oxford

As the war loomed, finding better drugs became imperative. In 1939, Howard Florey’s team at Oxford picked up where Fleming left off, demonstrating penicillin’s remarkable potential in animal trials. The results were stunning: penicillin appeared more potent than the sulfonamides or any other bacterial agent, it affected organisms that were otherwise immune to sulfa drugs, and overall it seemed far less toxic. But as Florey knew, the true challenge lay in treating “men, not mice.”

Thus began the human trials. By the spring of 1941, Florey’s team had treated exactly five patients, all of whom had severe infections that did not respond to sulfonamides. These were all cases where mortality was almost certain, yet three of the patients survived. Moreover, the two deaths that did occur were particularly frustrating: one patient had recovered, only to relapse and die when the available penicillin was exhausted; the other passed away from an unrelated complication. These individuals had also been “grossly undertreated by present standards.” The drug was not only difficult to produce, but it was also filtered out quickly after injection – remaining in the body for mere hours before being excreted in the urine, which Florey likened to “trying to keep a bath full with the plug out.”

Here, Florey faced the same problem that had stymied Fleming: he simply couldn’t manufacture enough penicillin for it to really be useful.

The challenge was fundamentally industrial. At Oxford, Florey and Ernst Chain had developed a laborious process that required 300 glass flasks to produce a single gram of penicillin powder. It was a “dry stable preparation a thousand times as strong as the crude preparation used by Fleming in 1929, [but Florey and Chain] obtained it only with enormous effort and in relatively small amounts.”

Most of these production problems can be traced back to the mold itself, which was extremely volatile. It would grow readily enough from spores into the long tangled green threads that could produce penicillin, “but the production is irregular, the product exceedingly unstable” – easily destroyed by heat, a pH imbalance, contaminating bacteria, or even by other molds. There was another worrying development. By growing staphylococci in the presence of insufficient concentrations of penicillin, the bacteria began showing signs of resistance to the drug. This ensured that any treatment would need to begin with “large and decisive doses” rather than small, incremental ones, which made the manufacturing issue all the more urgent.

By the summer of 1941, progress was still slow, and Britain was under siege. Resources were exceedingly scarce, and it became clear that a trip across the Atlantic was imperative. Protecting the technology and scaling up production required personnel and facilities that Florey hoped the American industrial muscle could provide. The stakes of this trip were so high that a fellow scientist on the mission, Norman Heatley, even suggested applying a thin layer of the mold to their coats to ensure it made it to the United States.6

Finding success in Illinois

In America, Florey and Heatley found themselves in Peoria, Illinois, hosted by the United States Department of Agriculture (USDA). The Northern Regional Research Laboratory (NRRL) located there had done work with fermentation and seemed like a reasonable place to begin work on increasing production. It is here that the story diverges: Heatley stayed behind to start problem-solving while Florey ran around the country to draw interest from pharmaceutical companies and the government alike.

In Peoria, scientists began work and saw first-hand the scope of the issue: “Put mold and medium in a flask and, ten days later, assays would show a good yield of penicillin. Put larger amounts of the same mold, the same medium, in larger vessels and assays would show no penicillin; or penicillin would be there on the fourth day and, suddenly, on the fifth day it would be gone; or it would be present in the final filtrate and vanish during one of the stages of extraction.” It soon became apparent that scaling production would need better techniques that increased yield and a new strain of mold that produced penicillin reliably.

Subscribe now

A series of happy coincidences quickly moved them forward. For one, Heatley did not have access to exactly the same media that he had used to culture the mold in Britain; instead his newfound colleague, A. J. Moyer, suggested that they use corn steep liquor. Then, as now, middle America was corn country, and corn steep liquor abundantly available as a waste product from cornstarch production. Introducing the liquor into the growing culture immediately increased the yield by tenfold. In fact, R. D. Coghill, another scientist at NRRL, noted, “One of the least understood miracles connected with [penicillin] is that Florey and Heatley were directed to our laboratory in Peoria—the only laboratory where the corn-steep liquor magic would have been discovered.”

The second miracle was finding a more efficient strain of the mold. While this discovery is sometimes attributed to luck, it actually resulted from coordination and determination. “The NRRL conducted an intensive search by collecting naturally occurring molds from both local and global locations. These collection activities involved everything from searching local markets for moldy fruits and vegetables to contracting the Army Transport Corps to collect and deliver soil samples ‘from wherever Army planes set down.’” The strain eventually used was found in a rotting cantaloupe in Peoria, Illinois, by Mary Hunt, a lab assistant at the NRRL. This new discovery again doubled production.

Branching Out

With these breakthroughs, researchers moved on to perfecting deep-tank fermentation – a process that enables the mold to grow evenly within large tanks, rather than just on the surface of laboratory glassware. This, however, was “essentially [a] problem of chemical engineering when it came to finding solutions on an industrial scale” and required a different set of expertise.7

It was part of a set of deeply technical challenges that required coordinated effort. Leaving Heatley in Peoria, Florey spent the winter of 1941 meeting with pharmaceutical companies and the U.S. government – convincing them that penicillin was a worthwhile endeavor. In D.C., Alfred Newton Richards recognized the production challenge and brought it under the wings of the Committee for Medical Research (CMR), a sub-division of the incredibly efficient wartime Office of Scientific Research and Development (OSRD).8

“Meetings were arranged by CMR in October and December [1941] between representatives of the National Research Council Division of Chemistry, the Department of Agriculture, and the pharmaceutical firms of Merck & Company, Chas. Pfizer & Company, E. R. Squibb and Sons, and the Lederle Laboratories. The Peoria Laboratory, which had started work immediately after Florey’s visit, and the pharmaceutical houses agreed to conduct research aimed at increasing the production of penicillin.”

This cooperation between the private and public sectors was rooted in an interesting structure, one that was built around an information-sharing agreement. Each firm agreed to conduct independent production research and present its findings to the CMR, which served as a clearinghouse for these findings. The CMR would then share these results with others in the consortium while providing technical support to individual companies via the NRRL.9

With the coalition in place, a technique for deep-tank fermentation was perfected, “[spores were] transferred into an 8000-gallon vat, containing culture media the acidity of which has been carefully regulated. There, the mycelia grow for forty-eight hours, constantly stirred by a huge ‘egg beater,’ constantly oxygenated by streams of sterile air, their temperature regulated by refrigerating coils in the walls of the vat. At the end of this period, the yellow fluid is filtered off and the concentration process started…[resulting in a] solution of the order of 50,000 units per cubic centimeter, 250 times that of the original solution [in Florey’s Oxford lab].”

Over the course of a year and a half, the laboratory procedures for consistently producing high yields of penicillin were in place. This allowed enough of the drug to be available for early studies at Bushnell General Hospital in Utah where over 200 wounded soldiers with bone infections (cases where sulfa drugs had completely failed) were successfully treated with penicillin. In-fact, “the results of the Bushnell military penicillin testing were so encouraging that the US Army now regarded penicillin as vital to the war effort.”

Therefore, in early 1943, it was of little surprise that an attempt was made to seriously ramp up manufacturing. The War Production Board (WPB) was called upon and threw the full weight of American industrial mobilization behind the production of penicillin. In May of that year, the WPB granted AA-1 priorities (the highest wartime designation) to selected commercial firms and over the course of the next year, with $20 million in funding, 21 plants had been erected and operationalized. Production scaled at a rate that defied comprehension: “The monthly production of penicillin, which had approximated 60 million units in pilot plants in May 1943, became 117,527 million units in June 1944. In June 1945, it was 646,818 million units” (it is worth noting here that about a million units were needed to treat the average patient).10

To achieve this scale in just about two years, the WPB orchestrated an high level of cooperation. They arranged monthly meetings where penicillin producers could exchange information on production methods and secured a waiver from the Department of Justice to exempt these meetings from antitrust laws. “On July 16, 1943, the WPB [also] issued allocation order M-338, formally partitioning all penicillin supplies among the Army, Navy, Public Health Service, and the CMR for clinical testing. [By] May 1944, when the amounts available for civilian use became greater than could be handled by CMR, the WPB established a Civilian Penicillin Distribution Unit in Chicago to allocate supplies to 1,000 selected hospitals.”

Share

The Aftermath

The clinical results were, as expected, nothing short of revolutionary. Diseases that had been virtual death sentences suddenly became treatable. While staphylococcal septicemia, a serious bloodstream infection, once killed four out of five patients, now four out of five patients survived. Subacute bacterial endocarditis, an infection of the heart lining and valves, once fatal in 100% of cases, became curable with sufficient doses. Even syphilis, which had required months of painful arsenic and bismuth treatments, could be controlled with a shorter penicillin regimen. Perhaps most importantly, penicillin proved to be that rarest of medicines that Ehrlich conceived of: a “magic bullet” that was both extraordinarily potent and remarkably nontoxic.

Thus, in just about four years (1941-1945), with a sense of urgency and with immense cooperation, the problem that had stumped Fleming and nearly defeated Florey was decisively conquered through purpose-driven industrial might.

If we continue moving further downstream, there is certainly more that could be written about the impact of penicillin. Even looking at a single target, pneumonia in American children, the numbers are staggering: the “rates of mortality from childhood pneumonia in the United States declined by 97 percent in the 58-year period from 1939 through 1996…The number of deaths ascribed to pneumonia dropped from 24,637 in 1939 to 800 in 1996, an absolute difference of 23,837 deaths, despite the substantial increase in the population during this period.”

It is unsurprising, then, that organizations went on to study the drug’s many applications while others were enlisted to synthesize chemical variants that could be remixed into second-generation antibiotics – all pushed forward by the same efficient central apparatus. Market incentives also drove costs down dramatically, making penicillin accessible at scale and saving countless lives.11 In every sense of the word, the drug is remarkable.

From Vannevar Bush, Director of the Office of Scientific Research and Development (OSRD), in Pieces of the Action:

“If I were to expand upon any single war program I would pick the efforts of wartime medicine, for here the tale has not yet been told so that it is well understood…For the first time medical science was adequately supported and encouraged… It put penicillin at our service, as could have been done ten years before had there been ample effort, and thus introduced the wide range of antibiotics. All this was important; it raised the possibility of healthy life on the planet even as war brought distress. But its importance was more than this. The war effort taught us the power of adequately supported research for our comfort, our security, our prosperity.”

Replicating success

There are many levers in the story of penicillin that we cannot reliably replicate – who knows when the next figurative speck of mold will fly through a window? What we can do is work to incentivize sustained effort. To signal to scientists that taking time to rigorously work on their most ambitious ideas is a good thing.

This is particularly important because, as some have argued, it seems we are making fewer consequential discoveries in the present day – as if we are slowly running out of the “low-hanging fruit” that pushed science forward in the past.12

For example, Matt Clancy of Open Philanthropy writes:

“We should expect science to get harder because of the ‘burden of knowledge’...Whenever new knowledge is discovered, it opens the way for new discoveries, and it may also displace or make obsolete some older knowledge. But if new knowledge does not entirely displace old knowledge, then it may be [that] you need steadily more knowledge to make new discoveries. Unfortunately, I think we have quite good evidence this seems to be the case as a general (though probably not absolute) rule. All else equal, if you need more and more knowledge to make a discovery of a given size, then you can probably expect discoveries of a given size to require more time or manpower to bring about.“

This seems plausibly true. After all, it was objectively easier to discover the earlier elements on the periodic table than the unstable, radioactive combinations we create today.

Yet, the burden of knowledge is not the only culprit. As Ben Southwood of Works in Progress argues, “there is good reason to believe that insofar as we haven’t seen big progress in energy subfields, this is itself down to problems with how we pick fruit, not because the lower fruit have already all been plucked.”

Regardless of which explanation carries more weight, both ways of looking at the perceived difficulty of modern-day science converge on the same prescription: invest. If new interventions are just plain harder to come across, then we should spend more money on ladders, and account for the additional effort of climbing up and down. Alternatively, if the problem isn’t with the ideas themselves but the urgency and willingness with which we work towards them, then we should, again, incentivize effort.

What could this look like in practice? Medicine offers clear examples of where more deliberate effort would pay dividends. As Matt Wedel, a professor of anatomy, points out: if surgeons collectively performed ten thousand well-documented autopsies, “that wouldn’t be enough to declare the science of human anatomy a completed project, but we’d know a heck of a lot more than we do now.”13

Similarly, discovering effective vaccines against high-prevalence but difficult to solve diseases like tuberculosis requires coordinated will. So does making diagnostics cheaper and easier – this includes things like blood-based screening for hard-to-catch cancers like ovarian and pancreatic, or bringing down costs for regular MRI/PET scans for people at risk for Alzheimer’s. These are worthwhile causes even if they require unusually large amounts of resources.14

The story of penicillin production reminds us that hard problems are solveable, but they require hard work. Fleming’s discovery was certainly critical, but what transformed medicine was the decision to marshal resources, coordinate expertise, and provide the right encouragement (in the form of investment) to scale up a solution.

To exhaust the metaphor: There is still plenty of fruit on the tree. We just need to do the work and learn to pick it before it rots.

Note: Unless indicated otherwise via nearby link or footnotes, all quotes from James Phinney Baxter III, Scientists Against Time, page 337-359.

With thanks to Mike Riggs, Abby ShalekBriski, Alexandra Danylyszyn, Elizabeth Van Nostrand, Etienne Fortier-Dubois, Michael Hill, Smrithi Sunil, Venkatesh Ranjan, for feedback on drafts. Thanks to Stuart Buck for introductory paragraphs 2 and 3. All errors and mistakes are mine.

Cover Image: Ethel Léontine Gabain, Sir Alexander Fleming (1881–1955), FRS, the Discoverer of Penicillin, 1944, oil on canvas, Imperial War Museums, London.

This is important context, because I truly am sympathetic to the triad against ultra-processed food (UPF). In my dictionary, and I know the lines can get blurry here, this includes things like cheese-puffs or high-fructose candy or those little microwave meals that reach unreasonably high temperatures. More granularly you can say that I try to avoid things with palm or soybean oil and any food label where the fat content and calories consumed just do not make sense for the recommended portion size. It’s a very intuitive gray area and it would be safe to assume that we all have different operational definitions of what falls under the ultra-processed bucket.

However, we need to be careful not to over-extrapolate. Processing makes food today safe and convenient in a way that it wasn’t before. This includes practices like pasteurizing milk, which benefits everyone by removing Salmonella, E. coli, and Listeria from our diet. I am aware that this is a rather low and dirty punch to make to establish my point, but there are many such cases where processing is quite obviously a net positive: salting fatty fish in airtight barrels to stop them from going rancid1, freezing vegetables at peak ripeness, or adding iodine to salt to prevent goiter.

Frustratingly, this means that the long debate on processed food requires nuance. I am not a big fan of this word because it can be used as a cop-out – transferring the burden of choice onto the individual. Yet, in this case, it seems like the best solution to the very fundamental problem of what we should be eating day in and day out is to develop better heuristics.

Methodological Constraints:

One reason inexact but well-informed intuition is important is that nutritional science studies can be unusually murky. Take, for example, this overview from 2013 that looked at 50 common ingredients and found that: “Associations with cancer risk or benefits have been claimed for most food ingredients. Many single studies highlight implausibly large effects, even though evidence is weak. Effect sizes shrink in meta-analyses.”

There are many dimensions to why this confusion occurs. For one, research on UPFs suffers from a definition problem. How can instant noodles, chicken nuggets, cookies, and fortified breakfast cereal be lumped into the same homogenized group? There is no shared mechanism that makes all of these things harmful and no consistent marker we can measure. In one recent NIH study, researchers very clearly showed this problem: “The UPFs that were associated with the highest risk for heart disease included sugar-sweetened beverages and processed meats such as hot dogs and deli meat. UPFs associated with the lowest risk of heart disease included breakfast cereals, yogurt, and some whole grain products. Examples of additives commonly used in UPFs include high-fructose corn syrup, hydrogenated oils, sodium nitrite, and artificial dyes.” Which is to say that ultra-processed food is a broad category, and there are a lot of variables that are difficult to untangle from each other.2

Even if we could agree on what counts as a UPF, we can’t actually run a true Randomized Control Trial – we can’t assign people to only eat certain foods for years on end, nor can we constantly monitor everything people consume. This inability to control a variety of confounds and a generally high reliance on surveys (where “participants unreliably recall what they eat”) makes the problem worse.

Despite these issues, (anecdotally) it was not unusual to see methodologically limited nutrition studies being presented as fact in third-party publications3 – adding to the general fear-mongering that surrounds food. But recognizing that epistemic standards in nutrition science are shaky actually helps us navigate the noise because it means we should be skeptical of both the latest superfood craze and the newest dietary demon.

Subscribe now

Drawing the Line:

To think through what this instinct for food should look like I will be using the words processed and ultra-processed a lot. However, there are lots of different ways to slice up the food pyramid and start labeling products good or bad. I want to avoid doing so here and also maintain some consistency when talking about processing so I am going to figuratively poke the bear and quickly explain what processed vs ultra-processed is going to mean in the context of this essay.

If we were to go by the U.S. Department of Agriculture, processed includes “any raw agricultural commodity subjected to washing, cleaning, milling, cutting, chopping, heating, pasteurizing, blanching, cooking, canning, freezing, drying, dehydrating, mixing, packaging, or other procedures that alter the food from its natural state.” This is certainly not a workable definition for the purposes of this article (I don’t think waxed apples or baby-carrots fall under our functional understanding of processed) but I have included it here just to demonstrate how large the Overton window can be.

More specifically, for ease I will be using the standard NOVA classification system: (taken verbatim from source)

• Category 1: Unprocessed or minimally processed foods — such as whole foods, vegetables, fruit, meat and pasta. These foods may have been washed, dried, frozen or vacuum-packed but have no added ingredients.

• Category 2: Culinary ingredients that have been processed, including oil, butter, sugar or salt. They are typically used only in cooking and not eaten on their own.

• Category 3: Processed foods — made by combining Category 1 and 2 foods through preservation or cooking. Examples include canned tuna, fruits in syrup and salted nuts.

• Category 4: Ultra-processed foods are industrial formulations made from food components. They include additives that are rare or nonexistent in culinary use, like emulsifiers, hydrogenated oils, synthetic colors, texture improvers or flavor enhancers. Think chips, soda, instant soup, pastries and mass-produced breads.

My hunch here is that most (but not all) of the foods in Category 4 are giving the word “processed” a bad rap, so much so that we are over-correcting in ways that can only be defined as romanticization of the culinary-past. This should not put us off the evidently safe, nutritious, time-saving items in categories 2 and 3 – think fortified flour, frozen veg, canned beans, or a little mono-sodium glutamate (MSG).

Two small notes, however:

1) I am going to disregard the meat industry for the most part. This is mostly because I grew up vegetarian and frankly have little stake in finding merits for the likes of animal feed that increases hot carcass weight (although if someone has written something good about this, please send it my way).4

2) I also won’t be making any socio-economic claims. One common argument for processed food is that it provides the required calories in a cheap and convenient manner.5 But this should be secondary to whether this food is actually damaging. Ease is good if, on balance, processed food itself is good for us – it cannot be an independent factor in how we set the standards for what we should be consuming.

Some Big Questions:

First let’s talk about safety because, processed or not, the top priority has to be consuming food that doesn’t actively cause harm. This brings up a few popular debates, including those on raw milk, seed oils, and food dyes. Since dairy, fat, and colouring are so ubiquitous in our food, there have been reasonable attempts to separate the signal from the noise. To establish some groundwork and context here, I am just going to summarize the clearest takes on this that I have come across.

Raw milk is very bad for us. There are no additional gut microbiome benefits or additional protein and healthy fats that get lost in the pasteurization process. Drinking milk that was once boiled is not what is causing lactose intolerance either – the heat is not destroying any imaginary lactase. I don’t believe that there are many people that fall in the raw milk camp but again it is a very good example of a widespread product that is markedly improved by processing.

Here on out, things get murkier. For example, with seed oils, it seems like a big reason behind vilifying them is a result of their inextricable tie and excessive presence in a diet where we are deriving an increasing amount of our calories from ultra-processed foods. High-fat consumption can be an independent risk factor for cardiovascular function. Similarly, increasing the amount of seed oil-derived linoleic acid or skewing the omega-6 to omega-3 ratio can also cause some downstream effects (although it is hard to definitively point out in which direction).

Dynomight has written an excellent overview on this problem and comes to the following conclusion:

“A weak version of seed oil theory is that seed oils are highly processed, so why not use cold-pressed olive oil instead? If that’s the theory, fine. In fact, this is mostly what I do myself. I figure it might be useless, but it’s unlikely to be harmful, and olive oil is delicious. And I wouldn’t be shocked if one of the suggested mechanisms for seed oil turns out to be valid. I wouldn’t be surprised at all if some mechanism turned out to be part of a larger, more complicated story. And in practice, avoiding seed oils is probably really good for you, because it forces you to eliminate most of the processed crap you shouldn’t be eating anyway.”6

I particularly like this stance because it gets to the core of the issue: UPFs are very calorically dense. It may very well be that seed-oils themselves are bad but the real confound here is that they are found in high quantities in food that just cannot leave you satiated in amounts where the macro-nutrient make-up is reasonable. It could also be plausible that if we found a cost-efficient way to make UPFs with tallow, butter, or olive oil then they would probably still be bad for us simply because of how calorically dense they are.

Fat, sugar, and calories are easy to measure and watch – but what do you do about food dyes? Recent debates spearheaded through the MAHA agenda (page 8 here) antagonize basically all synthetic colouring but testing whether this is true leads to the same problem of confounds as with seed-oils. From Colleen Smith: “There’s plenty of research on UPFs, but much of it is of mixed quality. Because UPFs bundle sugar, sweeteners, preservatives, flavors, and colorings, diet studies are confounded; therefore, UPF findings don’t automatically implicate synthetic food dyes.”

Here we might be tempted to take the same measure of precaution and switch to natural dyes but these have much looser regulations as compared to the seven legally-allowed synthetic dyes. From the same article: “Natural dyes require FDA approval to prevent inadvertent use of something poisonous. However, once approved, they are exempt from specific labeling and the monitoring inherent in batch testing…Because oversight for natural dyes is less stringent, we are at the mercy of the food industry to provide transparency on its quality control, processing, and sourcing of natural dyes. Without this transparency, a shift to these products could easily repeat the supplement industry’s problems in the 1990s, when purity and safety were far from guaranteed. Consumers shouldn’t have to worry about ending up with heavy metals, pesticides, or brick powder in their food.”

In reading the work that I have summarized here, it is clear that we should consider cutting out at-least some UPF’s. Those that fall into the quite obviously bad category include foods where high fat percentages and sugar content is crammed into teensy little “recommended” portion sizes [seriously though, try eating Nutella on toast and really stick to the 2 tbsp limit, it will be miserable breakfast despite the 21g of sugar 11g of fat]. But rationing your M&Ms should be an individual decision.

Good Ultra-Processing:

One place where ultra-processed food is clearly a net good is when it comes to protein. As I mentioned earlier in the piece, I grew up vegetarian which in and of itself is not surprising. What may be of note is that the Indian state where I was raised is 60% vegetarian. However, I don’t even fall into this massive majority because I eat eggs (which are considered non-vegetarian here). In this context, it is believable that (at least) 27 percent of households consume inadequate amounts of protein. Now this is one of the places where survey results just do not match up to observational evidence and I am not certain that the average urban-dwelling Indian consumes 63.4g of protein per day (as reported). It would be incredibly difficult to do so when a cup of milk has 4-6g and a roti (flatbread) has 2-3g. A growing number of people recognize this and the market has responded with high-protein breads, bars, powders, and yogurts.

These supplements include NOVA category 2 and 3 products such as milk and yogurt where the dairy is recomposed and modified in various ways to have a lower fat percentage and much higher protein content. More importantly, it also includes NOVA category 4 foods such as whey isolates, certain cheeses, and nut butters which are all quite necessary. For many individuals grass-fed steak and butter is just not a viable way of fulfilling their dietary needs – instead, tofu, seitan, paneer, and soy-flour enriched complex carbohydrates are all objectively good inventions that let people hit macronutrient goals while maintaining plant-based diets.

Rejecting Food Romanticism:

Yet there is a growing trend to RETVRN to the good old days – a nostalgic push on the internet with the rise of the “trad-wife” archetype, in restaurants with a steady push correlating expensive with local and organic, and in policy with curbing consumer choice. Together, they transform food choices into moral categories where choosing canned beans over dried ones is a character failing rather than a perfectly rational decision.

The fact of the matter is that most foods we consider “real” are themselves products of centuries of human intervention: selection, breeding, and cultivation techniques that would be unrecognizable to our ancestors. The bananas we eat bear little resemblance to their wild predecessor and the wheat in my feel-good sourdough has been modified through millennia of agricultural practice. Processing, in its various forms, is simply the continuation of humanity’s long project of making food safer, more nutritious, and more accessible.

“The food abundance Americans enjoyed is not because we cooperated with nature. Our abundance is a triumph of human ingenuity over nature’s indifference to us. A potential danger embedded in the worldview of the popular alternative is romanticism with nature and the past. Modern, technologically advanced food is rejected in favor of older, slower, more “natural” food. This mindset is potentially destructive because how we view our future affects not only who we are but what we become.”7

Moreover, the tools that make modern food possible – pasteurization, canning, milling, fortification, cold chains, emulsification, stabilizers – were invented to solve concrete problems: pathogens, spoilage, scarcity, cost, labor, and time. If we start from the lens of these very real issues instead of ‘vibes’, the case for a lot of processing is rather straightforward. It makes food safer, more consistent, cheaper per calorie, and easier to store and transport. It also shifts invisible burdens. Shelf stability lowers household waste. Standardization reduces batch-to-batch risk. Ready-to-use staples compress the time tax that otherwise falls on the people who cook most meals: “In 1965 American women spent an average of 113 minutes per day (almost two hours) in meal preparation. By 2007 they were spending only 66 minutes per day, a 47 percent reduction…While we might have a tendency to romanticize the past, the reality is that cooking was often a monotonous, arduous, thankless task required to keep the family fed.”8

Clearly processed food has its place and to understand what that is, (in lieu of sound, consistent research) we need better heuristics for navigating the modern culinary landscape. I propose that if we accept calories-in/calories-out as a fundamental truth then our intuitive guidelines become straightforward: be suspicious of foods where the recommended portion size seems divorced from satiation or where the macro breakdown doesn’t make sense. At the same time, there is no need to conflate these legitimate concerns with the broader project of food technology. The same processes of human ingenuity that has given us chocolate, cheese, and delicious San Marzano tomatoes year-round.

With thanks to Mike Riggs, Abby ShalekBriski, Adam Kroetsch, Ariel Patton, Colleen Smith, Dhruv Arora, and Nehal Udyavar for their suggestions and feedback on drafts.

Cover Image: Jean-Baptiste Oudry, Still Life with Monkey, Fruits, and Flowers (1724), Art Institute of Chicago.

The high mortality ward was staffed by medical professionals, while the second division was run by midwives. This earned the doctors a rather nasty reputation:

“Many heart-rending scenes occurred when patients found out that they had entered the First Division by mistake. They knelt down, wrung their hands, and begged that they might be discharged… they would protest that they were really quite well, in order to avoid medical treatment, for they believed that the doctor’s interference was always the precursor of death.”2

To some extent this was true. The professionally trained medical staff of the first ward were a lot more likely to carry out examinations for laboring women – simply put, they touched the patient more. Now this would not have been that big a problem if the doctors and medical students refrained from performing autopsies before attending births. Hand-washing between procedures was not common practice, and so the staff became vectors of disease – spreading the deadly puerperal sepsis from cadavers to vulnerable mothers. 

A doctor at the hospital, Ignaz Semmelweis, noticed this pattern3 and mandated hand-washing with a chlorinated lime solution. This dramatically reduced deaths in the high-mortality ward and validated his observational hypothesis that unseen “cadaver particles” were causing illness. 

Shockingly, however, for his life-saving work, Semmelweis was removed from the hospital – he died in a mental asylum aged 47 while his superiors and colleagues remained convinced that the popular theory of miasma (“bad air” that supposedly effervesced from spoiled organic matter) was the true cause of maternal mortality. Moreover, Semmelweis was neither the first nor the last to face such rejection, and the medical community struggled with the particulate nature of infectious diseases for decades. 

It was only by the end of the 19th century that there was some consensus. By then, a group of scientists that we are well familiar with (including Louis Pasteur, Robert Koch, and Joseph Lister) had collectively championed pathogenic microbes as the transmissible, causative agents of disease: germ theory. This late victory is despite the fact that by 1674, Anthony van Leeuwenhoek had already developed microscopes strong enough to see these tiny living organisms: “there were many small green globules. Among these there were, besides, very many little animalcules, some were roundish, while others, a bit bigger, consisted of an oval.”4

Looking at the broad sequence of events, it is truly astounding that about 200 years passed between the likes of Leeuwenhoek and Pasteur – footnoted by individuals such as Semmelweis who observed and worked against preventable loss of life. In this context, it is worth asking: Why did institutional medicine take so long to accept germ theory? 

There doesn’t seem to be a singular factor that adequately explains this delay, and the scope here is quite large, both geographically and chronologically. However, there are a couple of threads that I found particularly compelling and interesting – perhaps examining specific historical moments can partly reveal how gradually, and at what human cost, our understanding of pathogens and disease prevention evolved.

Subscribe now

Technological Hurdles:

One of the easiest explanations for a then-stagnating understanding of germs is practical: the microscopes just were not good enough. Early microscopes, such as those used by Robert Hooke, were essentially just very good magnifying glasses with limited power. They made objects about 20-50 times bigger which was far from sufficient to see what we consider “microscopic” today.5

Such was the state of viewing technology in Leeuwenhoek’s time. He was an interesting guy in that “he was not a doctor. He never published a scientific paper. He did not begin making contributions to science until he was 40. He did not invent the microscope.”6 Yet, today we associate his name with the invention because he was the first person to make advances that allowed us to view the truly minuscule – magnifying objects by 270 to 500 times their original size, giving us our first glance at “animalcules.” The key to this, according to some of his contemporaries, was the exceptional clarity of his glass lenses.

However, Leeuwenhoek jealously guarded this tacit knowledge: “my method for seeing the very smallest animalcules and minute eels, I do not impart to others... That I keep for myself alone.”7 His gate-keeping of methodology, combined with his limited scientific training, is one compelling explanation for why he failed to make the connection to early ideas of germ theory (such as Fracastoro’s theory of contagion)8 himself.

Elsewhere, progress on increasing magnification with a compound microscope was running into its own set of challenges. Here, Smrithi Sunil does an excellent job at explaining those issues: “van Leeuwenhoek’s single-lens microscopes had a problem: a tiny field of view. Early compound microscopes with two or more lenses promised to fix this, but the added optics created severe color aberration, which created a rainbow halo around every specimen because different wavelengths bent at different angles. In the mid-1700s, lens makers solved this by combining different glass types into the achromatic doublet, which canceled out the color fringing.”

All of this meant that the adoption of the microscope as part of the scientific toolkit was generally slow: “Some universities, including Leiden in Holland, had included microscopes among their teaching instruments, but medical instruction in microscopy appeared to actually decline during the 18th century.”9

This brings us back to Semmelweis, who never used a microscope to examine human tissue. His findings were mostly anecdotal. Similarly, a contemporary of his, John Snow, traced London's 1848 Cholera outbreak to a contaminated water pump (where a worried mother had washed her sick baby’s nappies); he, too, relied entirely on observing causal behavioral relationships rather than microscopic analysis. Here, again, while the authorities did take steps to mitigate the immediate disaster, popular belief reverted to the theory of miasmas.10 It would be another 35 years till Cholera would be considered a germ-based infectious disease following Robert Koch’s microscopic accounts from patients in Egypt and India in 1883: “It can now be taken as conclusive that the bacillus found in the intestine of cholera patients is indeed the cholera pathogen.”11

Institutional Hesitancy:

It makes sense that rigorous microscopic examinations helped move consensus. Koch was particularly good at this and had developed a methodology that helped him conclusively discover the causative agents of many infectious diseases including Anthrax and Tuberculosis. Today his process comprise the Koch Postulates: In essence, the hypothesized causative agent must be present in the tissue sample. Then it needs to be isolated in an uncontaminated culture, which can be injected into otherwise healthy animals. If these experimental models display signs of disease and if the same bacterium can be isolated from their bodies, then we can say that X germ is the cause of Y illness.

This methodology was crucial because simply observing microbes in infected tissue didn’t explain their origin:

“Were microorganisms to blame? If so, how did they get there? Many scientists held to the theory of spontaneous generation. If microorganisms could spontaneously appear, say, inside the human, how could a cogent theory of contagion even be considered? Dispelling the concept of spontaneous generation would be crucial for the development and acceptance of the germ theory of disease. Since spontaneous generation breached the church doctrine that God alone could create life, the debate even entered the metaphysical.”12

Koch’s postulates provided a solution to this dilemma. By demonstrating that microbes came from other microbes and not from thin air or non-living matter, his process made germ theory more palatable to both the scientific community and society at large. Moreover, it helped medical practitioners bypass uncomfortable theological questions about creation itself.

I bring up religion here because in some ways it might be fair to think of the rapidly professionalizing field of medicine as a sort of ‘secular ecclesiastical authority’ – one which worked to define itself against ‘irregular’ practitioners like midwives and folk/social healers. Like religious orthodoxy vs. heresy, ‘regular’ doctors (such as those ordained into professional societies and networks) claimed exclusive access to legitimate medical truth and worked to suppress competing traditions.

Share

This is a far-reaching claim but there is documented evidence for this. Take, for example, the case of Martha Ballard, a midwife working in 18th-century America who kept detailed diaries of her practice. Her “ability to prescribe and dispense medicine made Martha a physician, while practical knowledge…as well as willingness to give extended care, defined her as a nurse. In her world such distinctions made little sense.”

Yet as medicine professionalized, these flexible, community-integrated practitioners faced systematic exclusion. As Laurel Thatcher Ulrich notes in A Midwives Tale, “professionals sought to be distinguished from the community they served (hence the title ‘Doctor’). Social healers, on the other hand, were so closely identified with their public that we [historians] can hardly find them.” This created not just professional rivalry but genuine resentment: “Eighteenth century physicians, like twentieth century historians, had difficulty distinguishing one social healer from another, yet they understood the power of their presence…female healers identified with the patients they served in ways male physicians could not. Little wonder that some physicians actively resented their presence.”13

This discord in, and need for, establishing the authority of professionalized medicine perhaps helps explain the continued resistance to germ theory. Going back to Semmelweis, we can see how a finding that suggested that doctors were directly responsible for their patients’ deaths would be both morally difficult to digest and serve as a devastating blow to the medical field. Many prominent practitioners at the time spoke out against Semmelweis, including the highly respectable Rudolf Virchow (i.e. the man who “essentially invented cellular pathology”). In-fact Virchow specifically “was thought to have said [on his death-bed] that, ‘If I could live my life over again, I would devote it to proving that germs seek their natural habitat: diseased tissue, rather than being the cause of diseased tissue.’”14 The fact that the dissenting faction included highly reputable practitioners who had contributed genuine scientific advances made it difficult to dismiss opposition to germs as mere ignorance.

Among men of science, it truly was the doctors who were most hesitant to accept microscopic pathogens as the cause of disease. We see evidence for this in John Drysdale’s 1878 book, The Germ Theories of Infectious Diseases. Here, alongside reconciling the many different theories on the cause of infections, he wrote that:

“in recent times the chief patrons of the [germ] theory, following Pasteur...have been natural historians and physicists, while in the medical profession, which is naturally more familiar with the clinical facts of disease, its adherents have hitherto been in the minority.”

Now maybe by this point the clinicians had just fallen prey to the sunk-cost fallacy, but it is clear that the formal medical community was more resistant to germ-theory than the scientists.

Practical Necessity:

Despite efforts to the contrary, we did reach consensus: slowly at first and then all at once. A telling example of this can be seen in how the events of the Franco-Prussian War (1870-71) played out.

During the course of the conflict the French military performed over thirteen thousand amputations for wounded soldiers with a success rate of just over 24 percent. About ten thousand amputees died because doctors used a dated technique by physician François Broussais, which included bandaging up the stumps with various salves – surgeons could just dip their hands into jars of ointment with their bare, unwashed hands as they moved from patient to patient. We can imagine how that would have been disastrous. In contrast, on the German side, they used the “antiseptic surgery” methods of Scottish surgeon Joseph Lister.15 By cleaning and wrapping the amputation wound in a mixture of carbolic acid (or phenol), linseed oil, and chalk “the Germans would be the first to lose more men directly to battle wounds than to subsequent infection.”16

They were not completely lucky however, and thousands of soldiers both on the German and French sides contracted infectious diseases in the clinics. These could result in sepsis, gangrene, or fever among others unpleasant symptoms with slim chances of recovery.17 Remarkably though, in field hospitals (such as those where a young Robert Koch was stationed) the sick patients were quarantined! There was a baseline understanding that the sick could infect the healthy.

This growing awareness was further supported by the work of Dr. Edwin Klebs, another frontline physician who conducted 115 autopsies during the conflict. His microscopic examination of tissue samples from individuals who died of infectious cases revealed that they “contained [in almost every case] the lower organisms that are referred to as bacteria, monads, etc.”18 Yet this evidence too, failed to settle the debate. Klebs’s bacteriological technique was allegedly lacking, leading even a sympathetic colleague to wonder: “Does disease follow bacteria, or do bacteria follow disease? We still don’t know the answer.”19

This is quite reminiscent of Virchow’s alleged statement on germs “seeking” out diseased tissue and representing the effects of an illness rather than its cause. Such push-back also makes sense because Virchow was indeed a firm opponent to Kleb’s findings, arguing against a pathogen first explanation. Christoph Gradmann explains the tension within the medical community particular well here: “Once Virchow had—to put it briefly— committed himself to the causation of inflammations by mechanical irritants and defined them as internal processes, any view that gave a biological explanation for the same phenomenon and at the same time assigned an important part to an exogenous pathogen was bound to represent a clear opposition to the dominant doctrine of the age.”20

Despite this ongoing scientific debate, practical necessity was clearly able to drive change on the battlefield. War made for rather clarifying venue and military leaders could not afford to be embroiled in academic debates:

“In the civilian world, the germ theory was a radical notion...But on the battlefield, such disputes were academic. Lister had demonstrated that his approach saved lives, so the Germans, keenly aware that disease had historically always been more deadly to military forces than actual battle, made Lister’s antiseptic procedure compulsory at all medical facilities.”21

This pragmatic acceptance of sanitation practices, even without full theoretical consensus, suggests that perhaps somewhere a silent majority understood the life-saving importance of sanitation (even if they struggled to explain why it worked and reconcile it with the dominant world-view).

Final Thoughts:

Today, it is odd to imagine a world that lacks the very mundane understanding that germs can cause disease. After all, we do practice basic infection control without really thinking about it: we wear masks when necessary, we wash and sanitize our hands, we filter our water and so on. This was not the case for most of human history where death from preventable diseases was part and parcel of life – as recently as 1900 “the three leading causes of death were pneumonia, tuberculosis (TB), and diarrhea and enteritis, which (together with diphtheria) caused one third of all deaths. Of these deaths, 40% were among children aged less than 5 years.”

Just over a century later, this graph looks very different and germs are no longer the largest drivers of human mortality. Yet, as I have tried to demonstrate, we have travelled a long road to reach this point – circumventing technological barriers, moral discomfort, and institutional hesitancy. Taken together, these factors slowed down our collective scientific understanding and resulted in real, preventable loss of life.

Hindsight can be particularly illuminating and we can see that similar frictions exist in science today, that we still fail to update our priors as quickly as we should. However, crisis can force us to act: take for example the Covid-19 pandemic when emergency use authorizations fast-tracked mRNA vaccines that might otherwise have spent years in review. Sometimes we know things work but only feel comfortable adopting them as circumstances evolve.

In this light, the two-century journey towards modern germ theory feels like a necessary reminder of how our innate nature shapes scientific progress – and how that nature remains unchanged (unless rigorously challenged).

With thanks to Mike Riggs, Emma McAleavy, Lesley Gao, and Ariel Patton for their comments and feedback on drafts.

Cover Image: Antoine-Jean Gros, Bonaparte visitant les pestiférés de Jaffa [Bonaparte Visiting the Plague Victims of Jaffa] (1804), Musée du Louvre.

Suggestions for further reading:

Laurel Thatcher Ulrich, A Midwives Tale.

Thomas Goetz, The Remedy.

René Dubos, Louis Pasteur; Free Lance of Science.

Robert P. Gaynes, Germ Theory: Medical Pioneers in Infectious Diseases.

This was not always the case. For example, “between 1848 and 1872 tuberculosis was the leading cause of death in Britain, accounting for 15.0 per cent of all deaths.” Similarly, the microbiologist who first identified the causative agent of TB, Robert Koch, said in his 1882 lecture that

“If the importance of a disease for mankind is measured from the number of fatalities which are due to it, then tuberculosis must be considered much more important than those most feared infectious diseases, plague, cholera, and the like. Statistics have shown that 1/7 of all humans die of tuberculosis.” 3

More recent estimates state that globally, “the annual risk of infection at the beginning of the twentieth century was so large [...in the order of magnitude of 10 or more per cent] that it was very unlikely that a person could escape infection by the time of reaching adulthood.”

Clearly, TB has been a massive problem for a long time, ravaging the bodies of its victims without much concern for things like socio-economic status or geography. However, by the mid-twentieth century, following improved sanitation practices and the discovery of curative antibiotics, this had become a much more manageable disease in many countries. Why then does TB still kill over 1.2 million individuals every year (making it the largest infectious cause of mortality)? Why is this a hard problem for us to solve?

One of the reasons that makes TB so insidious is the nature of the pathogen itself. Mycobacterium tuberculosis grows slowly, replicating about once a day under ideal lab conditions, compared with minutes for many common bacteria. It takes time to grow because it wraps itself in an unusually fatty, waxy cell wall rich in mycolic acids, which makes it difficult for immune cells to penetrate. Instead, white-blood cells surround the bacteria, creating tissue balls called tubercles. Now, the bacteria enter slow or non-replicating states where the disease is uninfective and dormant to the point where the vast majority of people live and die oblivious to the presence of tuberculosis in their bodies: this is called latent TB. In India for example, the government estimates that “more than 40% of the population” has latent TB, yet active cases are much fewer. This does not make the disease harmless.

In about 5 to 10 percent of cases, the immune system cannot produce enough white blood cells to contain all the bacteria. M. tuberculosis then grows unchecked within the lungs or elsewhere, slowly overwhelming the body, causing a physical wasting away of tissue (earning it the name of “consumption”).4 Identifying these active cases early is notoriously difficult. Most patients today come from densely populated, low-income regions (which makes transmission via aerosols easy) where TB is already endemic. Malnutrition is common, so one of TB’s early symptoms, loss of appetite, may be misread as a welcome sign of feeling less hungry.5 This means that diagnosis is then delayed until the disease has progressed and patients are already highly infectious, ensuring that transmission continues.

Even in hospitals, infection control can be inconsistent. When I visited a TB ward in a government-run clinic in India, the first thing I noticed was that no one, not even a single staff member or patient, wore a mask. Instead, in the crowded waiting rooms, coughing patients stood shoulder-to-shoulder with those awaiting unrelated care (including maternal screenings). Such settings allow the disease to continue spreading freely. 

However, while infection control was sparse, I was pleasantly surprised to see that alongside chest X-Rays, the clinic took sputum cultures and tested each sample for drug-resistance before handing out the usual four-drug combination of antibiotics: rifampicin, isoniazid, pyrazinamide, and ethambutol (RIPE).6 This is a necessary advance, especially as the clinic reported that 20% of patients who are currently being treated came in with a drug-resistant variant of the disease and did not respond to the normal first-line interventions.7

Drug resistance is a predictable consequence of how M. tuberculosis interacts with treatment systems. The bacterium’s slow growth means that even with the right drugs, patients are forced to run a therapeutic marathon, taking the RIPE antibiotics daily for at least six months (or longer for more complicated cases). This prolonged timeline creates excessive challenges. 

For one, almost nobody is prepared or able to isolate for the few weeks it takes for the disease to return to a dormant state – business as usual, despite having an active infection, increases community transmission. Patients also often stop treatment mid-way because of excessive side-effects such as nausea, reflux, and sudden onset of rampant hunger. Others pause the drugs once they start feeling better, unwilling to come into the clinic weekly to pick up their next few doses. Moreover, with hundreds of patients being treated at any given time, the hospital’s tracking mechanisms are also stretched thin, and patients inevitably fall through the cracks despite efforts to avoid unnecessary attrition. Regardless of cause, each interruption gives the surviving bacteria the chance to adapt, and over time, strains emerge that no longer respond to one or more of the primary drugs.

Once resistance sets in, the problem compounds. Treating multi drug-resistant TB (MDR-TB) and extensively drug-resistant TB (XDR-TB) requires second-line antibiotics that are more toxic and less effective. Side effects ranging from hearing loss to kidney damage drive further non-adherence8, feeding the cycle of resistance. These regimens are also far more expensive, straining public health budgets that are already insufficient for universal coverage of drug-susceptible TB.

Tuberculosis creates a real financial burden on the patients and the systems designed to help them. Even as we continue finding ways of making testing cheaper and bringing down the price of drugs, the material realities of treatment add up. There is a real cost to missing work because of side effects, to commuting to the clinic on a weekly basis (if not more frequently) to pick up medication, to finding adequate nutrition, to staffing hospitals with enough personnel to track down infected individuals. This economic reality is part of the reason why tuberculosis, despite being entirely curable with proper treatment, continues to claim more lives than any other infectious disease – the systems designed to defeat it require sustained, substantial financial commitment that has proven elusive in practice.

The United States has historically provided about one-quarter of the total international funding for TB response with most of it being directed towards low- and middle-income countries (LMICs). However, since the recent dissolution of USAID and following the 90-day pause, a KFF analysis finds that of the 770 global health awards identified, 162 included TB activities, and 79% of those (TB-specific) programs were terminated. Similarly, the Center for Global Development estimates a 56% cut to the total dollar amount given by USAID for TB programs.

These cuts are expected to exact a real human toll: “An internal memorandum by the former acting Assistant Administrator for Global Health at USAID estimated a 28–32% increase in tuberculosis cases in the next year” if the funding cuts continue.9 Similarly, a more specific forecast from Asterisk Magazine uses Covid related disruptions as a proxy for how this shortfall might impact disease incidence and fatalities. In 2020 there was a 10-percent drop in global TB funding: this led to 59,000 additional TB-related deaths compared to the previous year.10 Today’s cuts are more concentrated (in LMICs) and may lead to 98,000–175,000 excess deaths in the first year as screening, diagnosis, drug delivery, and treatment support stall in the places most dependent on aid. Worse still, alongside increased mortality, these interruption to TB programs may increase the risk of people developing and transmitting the more difficult, drug-resistant strains of the infection. 

Given the projections are correct, we are consciously condemning hundreds of thousands of human beings to a painful and preventable death. The prognosis is horrifying. Extensive weight loss, weakness, fevers, difficulty breathing, and coughing up blood all herald a long, slow, torturous death – and that is if the disease doesn't spread to the bone or brain. In Everything is Tuberculosis, John Green offers a particularly chilling description: “If left untreated, most people who develop active TB will eventually die of the disease. Their lungs collapse or fill with fluid. Scarring leaves so little healthy lung tissue that breathing becomes impossible. The infection spreads to the brain or spinal column. Or they suffer a sudden, uncontrollable hemorrhage, leading to a quick death as blood drowns the lungs.” This is a terrible way to die.

Moreover, per a 2019 analysis, roughly 25 percent of the world’s population has latent TB – we are truly not going to eradicate the illness anytime soon. In fact, it’s not really a thing of the past anywhere in the world. However, what many countries have accomplished is a TB incidence rate low enough that the scale of the disease is no longer worrying, and we have made strides in steadily whittling down incidence globally, too. Knowing that we are capable of doing so should encourage us to press on with fighting tuberculosis.

There is no new, flashy solution to the disease – the nature of the bacterium makes creating a high-efficacy vaccine challenging, and its prevalence in poorer markets means that there is a lack of financial incentives to invest heavily in R&D here. But even so, there is merit in staying the course and fighting this disease the old-fashioned way. In rolling up our sleeves to increase screening, expand BCG access, track drug-resistant variants, ensure sustained medication compliance, hand out and enforce masking – the un-sexy but vitally important stuff. The things that save lives.

Thanks to Mike Riggs, Abby ShalekBriski, Steven Adler, Andrew Burleson, Allison Lehman, Kelly Vedi, and Samarth Jajoo for feedback on drafts; and to the staff at Gomtipur Urban Health Center, Ahmedabad for providing observer access to a government funded TB ward. 

Some reading that was particularly beneficial:

• Our World in Data’s 3-part series on TB [here, here, and here]

• John Green, Everything is Tuberculosis

• Thomas Goetz, The Remedy: Robert Koch, Arthur Conan Doyle, and the Quest to Cure Tuberculosis

• Vidya Krishnan, Phantom Plague: How Tuberculosis Shaped History

• Saloni Dattani’s We don't have to sit back and just watch the horror unfold

As an additional note: I focused quite heavily on India as my model country for high-burden TB because living here provides me with insight on how things play out on the ground. If you have context on how TB infrastructure has been impacted in other countries (especially on the African continent) and would like to chat, please email me on hiya [at] mundane [dot] beauty.

Cover image: The Sick Child (Norwegian: Det syke barn). 1885-1886. Oil on canvas. 120 × 118.5 cm. Norwegian National Gallery (Oslo, Norway) [Wiki Media Commons]

Edited to fix copy errors.

Looking back, this is probably one of the reason why I failed to engage with the discourse so far (and this is true for many of my friends too). I find it uncomfortable to read about the many reasons for declining fertility which can, in part, be attributed to the increasing access to education, employment, and the internet for women. Irrational as it may be, it feels callous to critique the choice that women have today to take on ambitious, fulfilling, if ‘greedy careers’ even if it comes at the cost of late marriages and fewer children. [This is perhaps the reason why Ruxandra Teslo’s article, Fertility on Demand, was so refreshing to read because it argued for choice. A ‘both and’ paradigm that allowed the reader to envision a future where she could both have the career she wants and all the children she desires.]

For the pro-natal argument to succeed, it needs to be presented with more empathy. Less as a hypothetical and with more care for the very individuals who might be most affected. My primary reason for thinking this is true comes from growing up in India and interacting rather fluidly with women across a wide range on the age and socio-economic axes. That is to say, this is a limited snapshot (hence ‘Notes’) but hopefully an interesting one.

Some context first:

• Total Fertility Rate (TFR) in India is sliding and recently hit 1.9 [global replacement rate is 2.1 children per woman]

• This decline is striking because:

  • Indian female labour force participation (LFPR) is not particularly high even when compared to demographically similar countries. Although this does not hold among the diaspora (i.e. once women leave India, LFPR increases quite a bit: see Alice Evans excellent article on How Migration Enables Love & Economic Prosperity).

    

  • Marriage rates continue to remain relatively high. As Alice Evans notes coupling is somewhat stable in South Asia because of caste-networks, arranged marriages, and social stigma against divorce (and late-to-marry young people). Furthermore, “between 1992-93 and 2019-2021, changes in non-marital fertility explain 0% of fertility change because it’s more-or-less zero in both cases.” My personal takeaway from this is that, at-least in broad terms, we cannot ascribe that much weight to the falling fertility is caused by falling coupling logic when talking about India – rather people are choosing to have less children within marriages.

    

  • Female literacy rate in the country is rising and the gender gap between in educational attainment is dwindling:

• In India, like in most countries, there is a preference for two children. This is especially important because “in the Indian context, childlessness accounts for only 6% of the difference between high-fertility and below-replacement districts” [source here] – i.e. again, the main driver of lower fertility in India is an increase in women having less children not women having no children.

  

  • And interestingly, part of this tendency can be ascribed to pressure from health-care workers (and society at large) to have fewer children. [Tangentially, another graph from the same source shows how women in India don’t end-up with their desired family size because of limited access to fertility care at higher than average percentages.]

To quickly summarize: Indian women are increasingly educated, matching male counterparts, though this hasn’t translated to workforce participation. They continue marrying at high rates due to social pressure, and married couples account for virtually all people who are parents. Both women and men converge on two-child households, influenced by a society-wide push toward smaller families.

Many of the facts presented above can be ascribed to a pervasive tendency towards female seclusion and honor in India. This is very much an oversimplification, but it is not false to say that:

• Educated women bring credentials and prestige to the family.

• Employed women have financial freedom, agency, and move outside the domestic sphere in spaces where traditional gendered interaction is broken down.

  • Limiting employment thus helps curb career ambitions that might delay marriage and prevents romantic encounters where women might choose their own partners

From the BBC: “In a 2018 survey of more than 160,000 households, 93% of married Indians said that theirs was an arranged marriage. Just 3% had a "love marriage" and another 2% described theirs as a "love-cum-arranged marriage", which usually indicates that the relationship was set up by the families, and then the couple agreed to get married.”

(Though anecdotally, many romantic relationships are retroactively presented as family-arranged to preserve honor).

• Caste restricted marriages are important for the honor of the family, for maintaining patrilineal ties, for business interests and so on.

Therefore we can say that an educated woman fetches better marital prospects for the family and her lack of employment helps maintain purity culture as the status-quo. But why smaller families?

I think much of the credit should be ascribed to the immense success of the “Hum do, Hamare do” campaign. Literally translating to “two of us, and two of ours”, this state sanctioned family planning initiative was aimed at eliminating excessive poverty courtesy of large families. A modernizing initiative of sorts that played out on all levels of society:

They [Indian bureaucrats] invented methods of communication that steered the popular imagination of the family towards a specific direction: the small, cohesive unit that came to epitomize postcolonial modernity. Their success may be gauged from the fact that the family of four contained in the red triangle not only reflected the state’s need to control birth rates, but also came to represent an aspirational standard of life for the middle classes. The idealized “small family” was one that could afford modern comforts, amenities and leisurely activities. [Source here]

Although never an authoritarian push like China’s one-child policy, the government did definitely succeeded in “steering the popular imagination” of the country. I want to present two example of how they did so by looking at television advertisements from the 1960’s.

Now some context. These clips probably ran on Doordarshan which was India’s primary television channel and a national phenomenon. Per my father, who has fond memories of the channel from the 70’s:

• It had limited hours of run time and was seldom on during the 9-5 slot when most people were working.

• There was a fixed telecast schedule most days with 15 minutes of news followed by 30 minutes to an hour of a show of some sort that came with at least 2 ad breaks.

• TV thus was firmly a family event where everyone in the community waited and watched together: predictable, high reach, limited choice.

It is within these circumstances that we see “Wives and Wives,” a full color animated short that was narrated in English – it first aired in 1962. I urge you to take the four minutes necessary to watch the full clip [linked here].

In essence, a man is picking out is his ideal wife through a biodata (a document akin to a relationship resume that is still in use today). He finds two women to his liking, both more or less identical in how they look and the man envisions life with them.

Wife number one dresses fashionably and loves to shop. She goes to the market and buys in abundance without much care for her husband’s monthly budget. She cooks elaborate meals with food waste that goes quite literally to the dogs. And moreover, meals in this household are chaotic with five children at the table (this was the average TFR when the commercial was made):

In contrast, wife number two is dressed more plainly and traditionally (with a bindi on her forehead), goes to the store only for what is necessary, grows her own tomatoes, stitches all the children’s clothes and so on. At her table there are only two kids:

Her habits and dispositions mean that the family can afford a bigger house, a better lifestyle and the husband then asks — why not have more children?

To which the woman responds with a firm no, stating that they should instead invest in the two children they already have:

This more elaborate and implicit indication towards family planning was later followed up with a much more famous advert in 1969 titled “Umbrella” [linked]. Again, full color, narrated in English, but simpler and more direct. The premise is straightforward – a man is standing in the rain with his family but is unable to offer all of them protection from the elements, his umbrella is too small and there are too many children (again we see five kids):

What if there were just two children instead? Now everyone is warm, dry, and happy:

See also in the above image the red triangle — this is the symbol for birth control in India and is still used today in all official communication regarding family planning.

Clearly something here worked. My dad is one of five siblings, but me and most of my cousins come in pairs of two.

Moreover, in 1990 the rate of birth control usage was ~36 percent but this has since risen to above 50 percent [52-56% depending on source] in women of reproductive age living with a partner. One caveat here is that many women, especially in rural parts of India, resort to female sterilization as a way of family planning – complete and final. This is often because the onus of contraception falls on the women, which is a separate issue to contend with, but I point it out here because this indicates that increasing the TFR in India has to be a long term campaign. Many families with two kids quite literally cannot have more children. [I have written about this sort of family planning as part of a biographical study on rural India previously (see pages 7-9 of the document linked here if of interest)].

Where does this leave us?

Today Indian’s across the board are all for family planning, for having less children but those who lead more prosperous lives. One only needs to look at the way out falling TFR is being reported in the news and on social-media to see this. Below is a snapshot (from last week) of a front page article in India’s leading English language newspaper, The Times of India. It states that “India has made significant progress in lowering fertility rate, from nearly five children per woman in 1970 to about two, courtesy improved education and access to reproductive healthcare…” – lowering fertility clearly is a positive thing here.

Similarly, the same news on social channels, where the audience tends to be younger, states the following: “Total fertility rate (TFR): India's TFR has fallen to 1.9, below the replacement threshold of 2.1.Long-term outlook: The population is projected to climb to 170 crore before beginning a gradual decline in roughly 40 years. The UN attributes this decline to the increase in educated population, improved reproductive healthcare, and women gaining a voice in the decisions that affect them…” [Source: fayedsouza on instagram]

The comment section of this post was filled with appreciation for how we are solving the over-population problem. The text itself lauds lowering TFR to a more literate, healthier, happier populous where women have more agency. This is a modernized India!

In this paradigm, why should young women in a developing nation even consider having more than two children? We grew up in a culture that prioritized female purity, where we heard stories of stigmatized divorced women and unhappy married ones, of unsupportive husbands and overworked wives, of distant fathers and dependent mothers – why would we surrender ourselves to the same fate? What makes it appealing? Moreover, we would be working against half a century of social conditioning where smaller families symbolized modernization and a more liberal atmosphere for women.

If pro-natalist policy-makers want to shift the tides, especially in India, then they must make larger families more aesthetically appealing to women. Make it a part of happy, new India instead of a cold and calculated economic problem to solve. There are many issues and debates I have not touched upon and this is by no means an all encompassing solution but it cannot hurt to have more empathy for the demographic in question.

With thanks to Stuti Iyer, Mahi Patel, and Harsh Aditya for feedback on drafts.

The insistence on ‘amateur’ was previously noted in an Asimov Press profile on Cocioba, which highlighted that the word comes from the French amour or Latin amare – to love. This relationship with biology makes sense given his interest in reviving the “gentleman sciences” of the nineteenth century1, demonstrating that valuable findings can come from the public domain. His guiding question to that end is simple: Can you be curious about the natural world and do publishable research as a hobby? Everything in his DIY lab is an attempt to prove that the answer can be yes.2

Cocioba started off like most hobbyists, running experiments for “the love of the game,” and then began sharing them on the internet openly, abundantly, and conscientiously. Soon, he was receiving inbound messages from academics and fellow enthusiasts alike: “Hey, I saw that you are doing something that aligns with my research. Could you help me with this thing and I will pay you?” Outside the ivory tower but suddenly privy to its inner workings, he describes these projects as “therapy” for the researchers, a chance for them to answer questions they have always had without being bound by a grant. Over-time, it became his primary job – running a “one-man contract research organization (CRO).” 

However, unlike traditional CROs Cocioba maintains a crucial clause in his contracts: if he builds a “fundamental tool, such as a really convenient vector that has nothing to do with [their] IP, then it should be open.”3 These methodological advances then become public knowledge, with due credit to the company that funded them. It is a model that benefits everyone – clients get their specific work done while gaining some goodwill and the broader community gains access to new techniques.

Cocioba is equally forthcoming with his own projects, which he documents publicly on Flowers for Everyone, a lab-notebook that he has previously described as an “open-source diary meets Julia Child cookbook.” Entries include experiments with Petunias engineered to turn deep red as a visual marker for successful genetic modifications, and progress-updates on modular plate-readers. Beyond the site, he is also working on a modernized twist to a floral dip transformation protocol for Arabidopsis that makes transgenic seeds glow while still inside their seed pods (siliques). This work, inspired by a 2003 paper, is part of his broader effort to mine expired patents for low-cost tools that can democratize biology. Taken together, his work represents “one example of how somebody wades through science day by day” and has become a reliable resource for a growing community of amateur scientists.4

He doesn’t just make public his successes, though, quoting Edison when probed about how he chooses what to document: “I didn’t make a light bulb. I learned a thousand and one ways how not to make a light bulb.”5 For Cocioba, failures are an important part of the scientific process – “That tacit knowledge is what industry has because they know how they failed. But in academia, we don’t actually know how everyone has failed. We only get the final result.” To circumvent this problem he publishes his shortfalls as open-questions on X: “Instead of failing in a vacuum, be like, ‘This failed, but I wonder why?’ And get the nerds who are on their lunch break going, ‘Oh, I actually know.’”

This philosophy of radical transparency also drives one of his proposals: bounties for biology. He imagines a Department of Defense-style system where researchers could earn money for validating published promoters or curating the thousands of unverified parts in the iGEM registry. The key difference from traditional funding? “If multiple people chase that bounty and converge onto it from different angles, each one of them should be paid the bounty in full.” Moreover, “it would be great if we can compensate for negative results by giving academic credit.” This approach could potentially incentivize validation while creating learning opportunities for students.6 He calls it “janitorial science” because it is unglamorous but essential. “Sanitation is the reason why civilizations work, and biodata sanitation is honorable work... if we actually give people credit and say ‘you are a sanitation engineer of biology’, that’s actually an incredibly respectable thing.”

The question of who does science, and how they are compensated, remains central to Cocioba’s work. Referencing the current rate of funding cuts, he recounts how friends have watched awards disappear because a proposal contained the word “climate,” or lost money because their proposals “had the word ‘diversity’ in it when they were talking about microbial diversity.” But the true damage, he says, will reveal itself later when research programs are abandoned and promising students leave science because the price-reward trade-offs no longer makes sense: “Cutting costs for basic research is like cutting off your leg to lose weight.”

The one potential upside he sees though, is more scrutiny of research budgets from scientists who begin taking a “hard look at the expense sheet.” Preferred-vendor purchasing, he argues, “turns into a mafia really quick” because no one knows the baseline. One example he shares is of a five-liter ‘reaction vessel’ which is listed at $300 in a lab catalog but when flipped over, reveals the same maker’s mark as a $23 IKEA bucket. He has made a sport of such investigations, which allow him to make custom lab-hardware for labs at a fraction of the cost: “cottage-core biotech” he calls it. 

These methods have become increasingly necessary. Prototype PCBs that once landed for five bucks plus shipping now cost almost ten-times that because of fluctuating tariffs, and Cocioba is forced to run even leaner – “Before, I would do seven designs and pick the one that works; now I have to do one and it better work.” To economize, he mills his own electrophoresis combs on a 3-D printer, swaps platinum electrodes for paper clips if the steel will do (“because thousands of boxes of paper clips are still going to be cheaper than one gel box”), and uses an Instant Pot as a foolproof autoclave.

Being able to DIY hardware is also useful because, “finding used laboratory tools can be hit-or-miss and it really depends on where you live. Now, because of tariffs, it’s even harder to get OEM7 stuff from China…and sometimes you just need something that holds a tube at a particular angle.” This prompted Cocioba to publish a paper where he cataloged some of these work arounds – a repository of 3D-printable lab tools that allow others to access wet-lab tools quickly and at a fraction of the cost.8 With consumer-grade printers now cheap and ubiquitous, 3-D printing has quietly democratized bench science, making it easier for curious individuals to start experimenting.

Cocioba makes a compelling case for why that matters.“There are industrial-grade microbes on your shoe right now,” he says, organisms that will die underfoot because no one bothers to swab the tread. With overnight, zero-prep sequencing now cheap enough for amateurs, he argues the real hazard is not sampling the puddle, not plating the mystery colony – there is a real opportunity cost to not being curious about the world. 

His open notebook, his bounty proposal, his insistence on documenting failures are all attempts to lower barriers to doing science. To show that the question has moved beyond whether amateurs can contribute to science to whether science can afford to ignore them.

All quotes and opinions from conversations dated 2nd and 27th May, 2025. Wording lightly edited at times for clarity.

With thanks to Alex Danylyszyn and Samarth Jajoo for feedback on drafts.

In 1869, alone in a damp kitchen of an old castle in Germany, Friedrich Miescher spent months looking into the components of pus cells, and coincidentally ended up isolating a new substance from their nuclei. We would eventually come to know of this material as DNA, but the true magnitude of the discovery somewhat evaded the young researcher. Although Miescher hypothesized that it could play a role in the generation of new life, he wrote up his findings as a simple biochemical observation: unusual, but not urgent.

His protocol was published two years later in dense German scientific prose with a title that translates to “On the Chemical Composition of Pus Cells” – hardly anything to be excited about. True to its name, the paper was mainly a methodological explanation for extracting the many components of white-blood cells. In fact, Miescher’s deliberation on the discovery of the DNA molecule goes but a little beyond calling it “nuclein” and noting its unusually high phosphorus content alongside its distinctiveness from proteins and fats. Unsurprisingly, the paper generated little immediate interest and was seen as yet another biochemical characterization among many of that era.¹

What is truly interesting however, are the notes that were never published. He speculated in letters and notebooks that this molecule might be central to both reproduction and proliferation.2 His mentor Felix Hoppe-Seyler, however, insisted on repeating every experiment before allowing any publication and was skeptical of broader theorizing, delaying Miescher’s original 1871 paper and leaving his bolder hypotheses confined to private correspondence. Regardless, he soon left his makeshift “lab” to train under another scientist before accepting a professorship where his research veered in a different direction, not returning to the question of ‘nuclein’ for several years.3

Had Miescher’s insights about nuclein’s role in reproduction been more widely recognized or had he pursued them more aggressively, the trajectory of biological science may have been dramatically different. The foundations of molecular genetics might have been established earlier. Instead, it would be almost eight decades until Oswald Avery, Colin MacLeod, and Maclyn McCarty conclusively demonstrated in 1944 that this material, Meischer’s ‘nuclein’, and not a protein as many had assumed, was indeed responsible for heredity – finally launching the decade-long endeavor to understand the structure of DNA.4

This story is a good example to get us thinking about science as a sprawling tree where each discovery opens up new branches to explore. Researchers start at the roots, and keep going until they find themselves at their chosen leaf on the frontier — but the intermediate nodes, which sometimes were truly at the cutting edge themselves, will often not be published. This pattern of scientific discovery, neglect, and rediscovery is more common than we would like to think.

Moreover, scientific stagnation stems not only from the loss of fundamental insights like Miescher's, but also from the disappearance of incremental and procedural knowledge that never makes it into formal publication. Because for every ten experiments, perhaps one ends up in the final paper. The rest – the intermediate steps, the ambiguous results, the dead ends, the subtle effects that didn’t replicate – remain tucked away in lab notebooks or forgotten entirely. Crucial technical ideas: better ways to calibrate an instrument, a trick to reduce noise in measurements, modifications to a protocol that saves hours, all become tacit knowledge. They are immensely valuable in getting things to work but are rarely written up, remaining hidden to everyone save the scientist themselves.5

Therefore, the story of scientific development tends to get lost and this is true even for publications describing incredibly exciting discoveries. Science can be messy but papers present a polished narrative where one logical step precedes the next and every key result can be summarised in a neat figure. In this framework miscellaneous experiments are pushed-out, sometimes to the appendix, and often from the paper entirely, for the sake of tidying up the manuscript.

In many respects, for scientists, choosing to go to graduate school is a sacrifice: a large opportunity cost paid in time and money that is mediated only by genuine curiosity. Upon entering the system however, even the best scientists are faced with a tough dilemma where working on the most ambitious projects comes with the risk of spending years being labeled ‘unproductive’ (i.e. not publishing).

While everyone values high-risk, high-reward science in principle, the practical reality is that researchers who take big risks need to consider the potential impact on their careers if those risks don't yield results.

Consider the plight of a newly enrolled graduate student: they might realistically be able to tackle about three substantial projects during their PhD. For an average scientist, each ambitious project might have only a 3% chance of yielding publishable results. Even if our researcher in this scenario is exceptionally talented and can boost those odds to 15% per project (or in other words is 5x better), the limited ability to sample gives them about a 61% chance of none of their projects resulting in a published manuscript (assuming they have the ability to take 3 of these bets during their PhD).6 It then becomes plausible that despite the scientist’s potential brilliance, they might have nothing tangible to showcase when they graduate, effectively shutting the door on their future scientific career. This is a devastating outcome for someone who chose this path out of genuine curiosity, passion, and ambition.

Clearly, this scenario is worth mitigating and so, there is a need to be pragmatic — finding a middle ground of ideas which are more easily possible, yet still exciting. We get a lot of good science this way: even incremental progress will often compound to solid expansions to the frontier. However, this system of publishing does create a self-perpetuating cycle of bad incentives, opaque science, and unhappy researchers.

This cycle of exchanging big ideas for stable careers is not new to American science. David Kaiser, for example, describes how the physicists who were once seen as solitary thinkers and tinkerers, got swept into the ‘suburbanization’ of the Cold War university system. Their departments ballooned and they ended up “trading cafes for cafeterias.”7 Graduate students arrived in waves, not to pursue wild ideas but to train for steady jobs to meet the increasing demands of industry or government research. The result was a more professional, productive, and well-funded science – and also one that could be read as a shift towards what Gerald Holton calls an “S2 system”: public, institutional science grounded in shared standards of empirical verification and logical rigor. What it displaced in some ways was “S1 science”: the private, intuitive domain that is often anchored in a sense of beauty and individual calling. As physicists themselves noted at the time, something got lost in the shift – a little curiosity, risk, and perhaps even the motive to do science in the first place.

We should consider different ways to measure scientific productivity and think about how we can change our metric from the number of prestigious publications to creating the largest “surface-area of shared information.” By incentivizing the documentation and publishing of failures and explorations of the intermediate leaves in scientific trees, we not only increase our collective understanding, but encourage graduate students to pursue their most ambitious ideas in a conducive environment.

In 2009, Fields Medalist Tim Gowers decided to attack a complex open problem in mathematics entirely in public. He wrote a blog post outlining his approach and invited anyone online to contribute. What followed is known today as the Polymath Project, a real-time experiment in massively collaborative problem solving. Over a few intense weeks, over a hundred comments were posted, and ideas were proposed, revised, discarded, and synthesized – often within hours. The end result was that a true solution emerged remarkably quickly and Gowers later described the experience as “to normal research [what] driving is to pushing a car.”

The project also surfaced a key idea that there is latent ‘micro-expertise’ scattered across the scientific world.8 This refers to people who, in narrow contexts, might have precisely the insight needed to move a problem forward and online tools enable this expertise to be activated in ways that conventional publishing or lab structures cannot. Leveraging micro-expertise also allows collaboration to benefit from probabilistic advantage – with more minds working on a problem, there is a greater chance that someone will have the specific perspective needed for a breakthrough.

In short, the Polymath Project reimagined what academic collaboration could look like: open, fast, chaotic, and remarkably effective. Importantly, it succeeded where others stalled because it combined open participation with just enough structure – clear ground rules, active moderation, and visible leadership – that kept momentum high. This mechanism was abundantly visible when the project tackled the Density Hales-Jewett problem, which exemplified what Gowers might have considered the “middle ground”: a problem significant enough to be challenging yet structured in a way that allowed multiple mathematicians to contribute different insights and approaches simultaneously.

However, Gowers himself raised a critical caveat on how far this model could generalize: could problems that “do not naturally split up into a vast number of subtasks” be effectively tackled by more than just a handful of people? In essence, the large-scale structure of Polymath makes it less effective for problems requiring a single transformative insight or those needing months of private contemplation to develop what mathematicians often call “mathematical intuition” – that pre-theoretical sense of a concept that precedes formal articulation. While the Polymath Project effectively amplified certain types of expert attention through parallelization, it struggled to cultivate the deep, sustained rumination that drives conceptual revolutions.

These limitations of Polymath reveal that different collaboration infrastructures might be needed for different types of intellectual activities. A compelling parallel here is Github, a platform that has transformed how large-scale software development happens by making previously invisible work processes completely transparent. On GitHub, the entire lifecycle of software development unfolds in public view: feature proposals through issues, implementation attempts via pull requests, the evolution of source code across versions, and the discussions that shaped critical decisions.

A developer might spot a bug and open an issue describing the problem. That report prompts others to investigate and suggest fixes. Then, if someone proposes a solution, they can share it through a pull request to invite open review and feedback before it is integrated. This process creates a massive, searchable repository of problem-solving: for almost any technical challenge, one can find discussions of failed approaches, successful implementations, and the reasoning behind design decisions. The platform therefore is not just an archive of final products but preserves the complete archaeology of how solutions evolved, with collaborations sometimes scaling to thousands of contributors.

Taken together, what both these examples demonstrate is that when the structure is right, collaboration can not only be broader but smarter – allowing expertise to find the problems it is best suited to solve. Michael Nielsen articulates this idea in his analysis of online scientific collaboration where he points out that expert attention is the ultimate scarce resource in scientific progress. Nielsen also writes about “restructuring expert attention”: the bottleneck is often the cognitive bandwidth of those with the specialized knowledge to interpret, connect, and advance ideas at the frontier.

This is where the true promise of both Polymath and GitHub-like systems lies: in their ability to create “attention markets” that operate with lower friction. These markets function by making problems and potential solutions visible to a wider audience and creating mechanisms for efficiently matching expertise to challenges. When Terence Tao posts a mathematical insight on his blog or when a developer submits a clever issue with a thorough spec, they are making a bid for expert attention in a more direct, immediate way than traditional channels allow.

Despite the general absence of public, collaborative infrastructure, scientists do have private mechanisms to document years of ideas, experiments, and iterations. The most ubiquitous of these is the lab notebook which has been a long held tradition where the researcher chronicles, with immense granularity, the true lives of scientific ideas. Within this journal one finds snapshots of quiet triumphs written down with as much sincerity as moments of failure and reflection. Thus, one can witness the false starts and gradual refinements that lead to illuminations at the frontier.

These notebooks really do have it all: photos of gels, observations, screenshots of analysis outputs, next steps. Unfortunately, the notes are typically trapped in isolated systems – bound notebooks, private Slack channels, or emails between collaborators. What we need instead are open, real-time, lab notebooks.

Think about a biologist, Dave, as he optimizes a PCR protocol for his single-cell atlas project. Dave writes down the settings, the samples, maybe takes a photo of the setup. Maybe he runs into an issue, and documents the troubleshooting steps he took, along with changes to the protocol. All of this is probably sitting in his lab notebook in Iowa City right now. But if it were in a shared, searchable system, it could help hundreds of other researchers dealing with similar setups. It is important to highlight that there is no extra work for Dave, except simply uploading his notes — and that being public does not imply being participatory — to interact further is a separate choice.9

Where this gets really powerful though is in the connections that could be built. Say a researcher in Boston, Julia, is now tweaking a statistical analysis technique for scRNAseq data that is very similar to Dave’s, and because the system understands the connection, Dave gets a same-day notification about the new entry in Julia’s lab notebook — instead of two years later when it appears as an appendix to the methods section. Or maybe, Julia is stuck on a calibration issue with a specific instrument, and the system connects her with another biologist, Alan, who documented solving the exact same problem three months ago. A language model can ingest updates, and we can build semantic search infrastructure to have a big impact on Dave, Julia, and Alan’s research.10

Just a couple of years ago, these frequent updates would generate data that became nearly impossible to search through. Today, we have seen that language model based tools can help us find relevant scientific literature. For example Future House, a SF based non-profit, recently launched PaperQA 2: a LLM powered agent that can search over a large chunk of all the scientific literature out there, and out-performs postdoc-level biology researchers on retrieving relevant information [since time of writing, Future House has also released three AI Scientist Agents]. Tools like PaperQA can become even more powerful when they have access to information that doesn’t even make it to the publication. Imagine a world in which a researcher posts experimental results, and after a quick NotebookQA run, gets a notification stating their results look pretty different than runs from their colleagues, prompting them to repeat and verify their experiments. These are the kind of improvements that can save weeks on end for scientists.

With the same infrastructure, we can create efficient attention markets where researchers (and perhaps even the public) can follow updates on scientific subfields. Instead of peer review happening on a manuscript or on the results of science, peers can provide feedback as the science progresses. They can point out potential issues in real-time, suggest controls that might have been missed, or offer alternative interpretations of the data. Researchers can incorporate this feedback and credit those who helped, or they can decide it is not relevant and move forward with their original approach. Either way, the entire exchange is visible and becomes part of the scientific record.

In this world, a project that fails to yield anything publishable would still create value by showing what approaches were tried and abandoned. The dead ends themselves become valuable signposts, and we appreciate the earnest scientists who really tried to work on crazy ideas instead of pointing to their lack of a publication record as reason to not hire them. Imagine a shared‑lab‑notebook ecosystem in which every branch climbed, every node explored, remains visible. A null result from an antibody that failed to bind or an organic synthesis that plateaued at sub-optimal yield all become tagged, timestamped reference points that future researchers can query. They stop someone in another lab, on another continent, from wasting three months repeating the same cul‑de‑sac, and they create a new, positive metric for the original scientist: the “surface-area of shared information” they contributed.

Review committees could count how many technically solid, substantial, but ultimately negative results someone contributed to the commons. Hiring panels would see evidence of ambitious swings rather than a thin CV filtered by the survivor bias of publishable results. Over time, whole subfields would acquire a “terrain map” of what has already been tried, encouraging bolder exploration because the personal cost of striking out drops dramatically. We can, and should, turn failure from a career liability into a public good.

This could also have a large impact on conventional funding pipelines, overhauling how resources are allocated. Currently, funding decisions are based on grant applications and publication records – lagging indicators that tell funders what someone has already done. But with science happening in the open, resources could flow dynamically to promising work as it emerges. A project showing unexpected results could attract additional funding, equipment access, collaborators, or computational resources, within weeks of the discovery, instead of two years later when the paper is published.

We think there are two core parts to making these open, real-time lab notebooks a reality: software and funding.

The process of doing research is chaotic, and the software should embrace that. Stream-of-consciousness notes, quick observations, voice memos — all fair game. No need for templated write-ups and rigid structure. The system should automatically extract key concepts and methods, making them discoverable without manual tagging. We have seen early versions of this idea: communities like OpenLabNotebooks tried to normalize transparency in day-to-day science. But the friction remains – logging entries still feels like a chore, and finding relevant past work is often harder than it should be. That is why it is crucial that the next iteration of these tools make it very easy to create, share and interact with notes. Well built software will make a huge difference here.

Ideally, this interface will also have a graduated approach to visibility for those who prefer updates being private, or shared with small groups to start with. We should aim to build software that is a value add for scientists even if they want to keep their findings close to their heart until they publish — essentially, a great blogging platform for research findings, even if the blogs are (semi) private.

If given permission, the software will make the daily work of science discoverable, and systematically create productive connections between researchers. It will surface these connections naturally, creating a map of scientific exploration that grows more valuable with each new entry. Like we showed earlier, this is tractable to build — and places like Future House and the Allen Institute have already been developing part of the stack which will enable it.

Meanwhile, funding is becoming a widespread and increasingly pressing issue with federal science budgets contracting as we are writing this. NSF director, Sethuraman Panchanathan, recently resigned from his post after facing calls to slash the agency’s budget for the coming year by 55%. The NIH is not faring any better: a Politico scoop details an internal proposal to slash the agency from 47 billion to 27 billion (around ‑40%) which comes alongside a push to consolidate its 27 institutes into just fifteen. This disruption in the traditional funding scheme creates challenges for the scientific enterprise at large, giving us an opportunity to experiment with how it operates: we think graduate students offer the perfect starting point for this transformation. They generate novel data daily, are actively building their scientific identity, and often most acutely feel the limitations of the current system.

A good way to kick-start the movement with some momentum would be to launch a fellowship program for them. Instead of piecing together funding through teaching and grant support, fellows would receive stable, flexible support to pursue their most interesting questions. At a time when federal science budgets are in retreat, this model would offer a timely and tractable way to sustain early-career researchers while piloting new norms for open, ambitious science. This financial and structural support creates the freedom to experiment not only with research ideas, but with how science itself gets done.

The fellowship’s mandate would be simple: work in public. Fellows share their journey – the data, the code, the half-formed hypotheses, the failed experiments. Not in a performative way, but as a natural extension of their research process. Instead of publication anxiety, fellows build a visible record of scientific contributions that go beyond the scope of traditional papers. The explorations themselves – not just the projects that make it into journals – become recognized as valuable scientific work.

Seeding the movement with an active core, and building a community around the process is essential to it taking off. This is, in some ways, a collective action problem, and the fellowship is a good way to counter academic pressures and competitive dynamics: things which could lead to the idea dying out before it is even tried out at any sort of scale.

Where the printing press gave rise to the scientific journal, the internet provides a similar opportunity to build something new entirely: something more dynamic and honest. We can capture the texture of research as it unfolds. And we now have the tools to make this possible, systems that render science searchable and collaborative.

If Miescher was a fellow working within such a paradigm, his early observations and speculations about ‘nuclein’ would not have remained hidden in letters and notebooks. Those insights, even as they were forming, would have been visible to others studying heredity – potentially accelerating one of biology’s most important discoveries by decades.

Scientists already recognize many of the problems that we have outlined here. The challenge is no longer awareness, but cultural imagination: how to shift the norms and incentives of science toward a new methodology. The current moment offers a rare alignment. On one hand, recent advances in software make this kind of infrastructure tractable, on the other, the instability of traditional funding has made the scientific community more open to rethinking how research is organized and supported. The friction is lower today than it has ever been — we just need to act.

Additional Notes

• Erika DeAlden Benedictis has a series of posts which share project updates from her PhD: these updates are very well contextualized, and readable by non-biologists — they illustrate the kinds of information we want to collect.

• In 2007, Jean-Claude Bradley was already talking about open notebook science but the movement never really took off: information was scattered across different blogs and wikis, annoying to post, and hard to search through.

• Another reason for open notes is simply to make science less lonely. It is nice to have people following what you’re working on, and encouraging you when things look grim. A good example is how several makers follow each other on wip.co.

• It’s probably a bad idea to position the fellowship as an incentive to work in the open. We should find people who already want to (or already do) work in the open, and help them survive.

With thanks to Jase Gehring, Eric Gilliam, Linus Lee, Neeraja Sankaran, and Nim Tottenham for reading early drafts.

References

There really is no better place to see this phenomenon unfold than my home institution of Columbia University where roughly 40 percent of the student body hold non-U.S. passports. Chinese nationals alone make up more than 6,500 students and researchers – about one in five across the institution – most of them enrolled in graduate programs in the School of Engineering and Applied Science. The one-year Master of Science degree, in particular, has become a popular, self-funded track among international students, driven largely by the opportunity it offers to enter the OPT pipeline. In fact, as Adam Tooze observes, the school is so well known among China’s educated middle class that Columbia has earned a Mandarin nickname: 哥大 (gēdà), short for 哥伦比亚大学 (gēlúnbǐyà dàxué).

While still striking, these numbers feel like a holdover from an early-2000s ideal – when students came, studied, and stayed. They earned advanced degrees, joined U.S. research institutions, or took jobs at firms driving the next wave of innovation. For decades, this model helped power the American technological ecosystem, which leaned on foreign-born scientists to reliably staff its labs, launch startups, and support discovery. But the tides are shifting. Immigration barriers and political uncertainty have led a growing number of Chinese graduates in particular, to take their skills and return – often not as junior employees, but as lab directors, company founders, or venture-backed entrepreneurs. And crucially, where returning once meant catching up, now it often means competing.

This is not a trend that occurred by chance, and much of it instead is part of a collective push by the Chinese government to bring home the “sea turtles.” This term is a literal translation of the word haigui which is a linguistic play in Mandarin that emerged around 2002 as shorthand for overseas returnees. It is a homophone of 海归 (meaning “returning from overseas”) that employs the image of sea turtles that instinctively return to their birthplace to nest. But China’s success in reversing its brain drain has been anything but natural instinct. It represents perhaps one of the most coordinated attempts at talent repatriation in recent decades.

The foundations were first laid in 1978 when Deng Xiaoping urged sending students abroad and welcoming them back. By 1993, this evolved into an official Communist Party policy of granting “freedom to come and go” to overseas scholars. Since then, it has broadened from academic recruitment to a whole-of-government approach towards economic development. The scale of this shift is striking. Since the late 1970s, more than 6.5 million Chinese have returned after studying or working overseas. What began as a trickle became a wave: from fewer than 15,000 returnees in 2000 to over 580,000 in 2019. Today, more than 80% of Chinese who study abroad eventually return home. According to China’s Ministry of Education, by the end of 2019, 4.23 million out of 4.90 million overseas graduates had returned, a cumulative return rate of 86%.

Rather than playing to patriotism alone, the Chinese government developed an interlocking set of policies and incentives to make returning home not just possible but appealing. These measures range from national-level talent recruitment initiatives to localized perks, and they operate in tandem with reforms to institutional infrastructure and research environments. The centerpiece in this giant policy machinery is the Thousand Talents Plan that was launched in 2008. It provides competitive salaries, startup grants, and housing subsidies to lure top-tier researchers and entrepreneurs back from overseas. Its junior branch, Young Thousand Talents, targets scientists under 40, offering them fast-tracked positions and research independence rarely afforded to early-career academics elsewhere.

These national-level initiatives are reinforced by regional efforts that address logistical and financial barriers. Grants between $70,000 and $420,000, housing subsidies and tax breaks all reduce friction for returnees. Even China’s notorious household registration (hukou) system, which typically restricts access to urban benefits, has been modified. Chinese cities and provinces actively compete for this influx of overseas talent by offering fast-tracked hukou approval for those with foreign degrees, effectively removing one of the most significant barriers to geographic mobility within China. Degrees from prestigious Western universities provide automatic points in these systems, virtually guaranteeing residency rights in top-tier cities like Beijing and Shanghai. 

Other local municipalities coordinate with state-owned enterprises and tech parks to offer further support. Although these programs in non-target cities have varying levels of success depending on the source describing them, I mention them here to demonstrate that there is consistent investment on all levels to lure back talent. 

Importantly, the government’s efforts to encourage return migration have not relied solely on financial incentives. Structural reforms that address logistical burdens such as family resettlement, administrative complexity, and institutional rigidity have been equally instrumental. For example, China’s “double first-class” university initiative was launched in 2015 and has played a central role in this strategy. Universities designated as such receive concentrated state investment to develop “world-class” status (which has made them hotspots for recruitment in tech and bio), with policies emphasizing increased research and faculty productivity.

Thus, over the decades, we can see how the Communist Party’s organizational muscle helped make policy around returnees a main-stay pillar of the country’s national development strategy. In 2003, the Party formally declared itself the primary authority over all talent-related policies, creating the Central Leadership Group on Talent and integrating efforts across ministries, universities, and regions. This centralization allowed for rapid mobilization, especially during global economic downturns like the 2008 financial crisis, which China seized as an opportunity to bring its diaspora home – the director of China Service Center for Scholarly Exchange, Bai Zhangde explained in 2009 that:

“In terms of overseas students and scholar[s], the influence of the financial crisis is a mixed ‘blessing’...with the implementation of the domestic strategy of ‘rejuvenating the country through science and education’ and the strategy of ‘strengthening the country with talent,’ the adjustment of the country’s industrial structure and the acceleration of scientific and technological innovation have provided broad space and more opportunities for overseas students to return to China.”

Nowhere has this push paid greater dividends than in biotechnology. Between 2010 and 2020, over 140 returnee-led biotech companies were founded in China, many by researchers trained in America in fields like oncology and gene therapy. Several of these firms – Biosion, EpimAb, and Harbour Biomed among them – have leveraged U.S.-acquired expertise to build drug discovery pipelines in China. The migration of this talent therefore reflects not only career incentives abroad, but gaps in the American system that make long-term retention harder than it should be.

The effects extend far beyond individual companies. A recent analysis found that participants in the Young Thousand Talents program significantly increased their research output after returning – publishing 27% more papers than their peers who stayed abroad, including more work in high-impact journals. These returnees also assume leadership roles at a higher rate, as indicated by a 144% increase in last-authored papers, suggesting they have independent labs and research agendas. 

At the same time, China’s biopharmaceutical landscape has become more attractive for multinationals and global investors alike. Merck, for instance, recently acquired a preclinical GLP‑1 molecule from Hansoh Pharma, soon after rival Summit Therapeutics inked a partnership with a different Chinese company to develop a competitor to Merck’s blockbuster lung cancer drug. In fact, in just the first months of 2024, the ten most lucrative licensing and collaboration deals between Chinese firms and global pharmaceutical giants neared $26 billion in total value.

These developments reflect a broader reconfiguration of global innovation. China is now at par in areas like artificial intelligence and accounts for nearly one-fifth of all CRISPR-related patent families worldwide. In sum, it can be said that these trends reflect more than just individual career decisions; they signal a shifting center of gravity in global science.

In this new and volatile world order, America risks losing its scientific and technological edge not through dramatic collapse, but through quiet erosion. For one, immigration policy has not kept pace with the realities of global talent competition. The H-1B visa cap remains stuck at 1990s levels despite a vastly expanded economy and growing demand for STEM talent. And the transition from a student visa to permanent residency remains long and uncertain, often requiring years of employer sponsorship, which in turn limits mobility, discourages entrepreneurship, and deters risk-taking among early-career scientists.

Compounding these structural issues are deeper cultural and geopolitical currents. For one, the now-defunct “China Initiative” created a climate of fear and suspicion that has yet to fully dissipate. A 2023 Stanford study found that the Initiative alone triggered a 75% jump in departures of China-born scientists from U.S. institutions. More than 1,400 Chinese researchers left between 2018 and 2021, and many others now avoid applying for federal grants out of fear or frustration.

To be clear, national security concerns about technology transfer are real and merit careful policy responses. But blunt instruments also have the capacity to produce unintended consequences. When highly trained scientists avoid grant applications that might fund breakthrough research, when promising graduate students choose other destinations for their education, or when talented faculty depart for opportunities elsewhere, the loss becomes material in very real ways. 

Retaining global talent has long been one of America’s most formidable advantages. Chinese nationals alone account for 17% of all U.S. doctoral degrees in science and engineering, and historically, the vast majority (~75% of all international PhD students) have stayed to work and teach in the U.S. after graduation – mostly to national benefit. But that pipeline is no longer guaranteed.

The shifting landscape demands at minimum a reconsideration of how the United States approaches global talent. This is not merely about competing with China’s incentive packages, which would be neither feasible nor necessarily desirable given different economic and political systems. Instead, America must leverage its existing position as a world-leader in training scientists, and make it easier for them to stay here (especially because researchers already want to).

There are some solutions. First, the U.S. could streamline pathways from graduate education to permanent residency, especially for STEM PhDs. Proposals to staple green cards to diplomas have circulated before; it might be time to offer this idea at least some consideration if nothing else. The annual cap on employment-based green cards (set at ~161,000 for FY2024) is not only inadequate but hampered further by per-country limits, and forces nationals of countries such as India and China into decade-long queues. These bottlenecks lead to significant attrition: when PhDs graduate into uncertainty, many return to their home countries. Modernizing the green card process by exempting STEM doctorates from annual caps or creating a fast-track system for graduates of American research institutions would be a concrete step toward retaining the very talent the country has already invested in training.

Second, federal funding agencies should specifically increase support for early-career researchers, providing the autonomy and resources that many now search for elsewhere. The median age at which researchers receive their first major NIH grant has risen to over 42, a striking lag given that many biomedical discoveries emerge from younger labs willing to take conceptual risks. ​​As the Institute for Progress recommends, one promising solution is the expansion of postdoctoral and graduate fellowships that are institutionally portable, i.e awarded to individuals rather than tied to a particular lab and open to temporary visa holders. These reforms would make it easier for young scientists to move between institutions, form new collaborations, and launch independent careers, thereby unlocking “America’s latent scientific capacity.

Finally, the U.S. should treat talent retention as a matter of national competitiveness. That means integrating talent policy into broader strategies on industrial policy, research infrastructure, and immigration reform. It also means fixing problems that affect all researchers, such as administrative bloat, rigid academic hierarchies, and declining federal support for basic research. But it especially means recognizing that highly trained, mobile scientists have choices now that perhaps they did not historically. Therefore, ensuring that America remains “the best place in the world to do high-impact science” is no longer a self-maintaining status – it requires good legislation and a turn away from the current trajectory.

Despite the many challenges that students and researchers face in the U.S., for now, it remains the place to be. In writing the story of the “sea turtle” I aimed to not just make a point about China, but to remind that talent is mobile, and that systems shape decisions. America still has unmatched strengths in higher education and research (and these homegrown advantages need not be exported to foreign countries because of domestic policy inaction), but this clear upper-hand is eroding at the margins. Many international students would prefer to build careers in America if given a realistic path to do so – drawn by its universities and a culture of unfettered freedom. Yet, when that ambition meets structural friction, rational choices often lead them elsewhere.

This tide is not irreversible though and America can continue to be competitive by restoring clear blue waters for talented individuals from across the world. 

With thanks to Samarth Jajoo and Adam Lehodey for feedback.

Lead image: William James Bennett. Weehawken from Turtle Grove. ca. 1830. The Metropolitan Museum of Art, https://www.metmuseum.org/art/collection/search/13117.

Today we benefit from a remarkable piece of technology that has improved both human productivity and quality of life. Yet, this was not always the case. Strikingly, right up to the turn of the 20th century our tools for dealing with pain were limited and our understanding of the underlying mechanisms even more so – the path to making pain a choice was a long one. Over the centuries, we have fermented poppy plants into opium, shaved willow bark into tonics, and inhaled strange vapors in front of large audiences. Now, a new generation of biotech companies wants to isolate the very “switches” in our nerves and flip them off. Here is a short history of how we got here:

Graph inspired by Saloni Dattani’s excellent article on the Golden Age of Antibiotics.

Opium and Willow Bark in Antiquity

The opium poppy, or the “joy plant” as the Sumerians called it, emerged as one of the first breakthroughs when discovered in the Fertile Crescent. By the time of the Egyptian empire, opium had become a staple remedy. Its ability to induce sleep and numb suffering was so pronounced that doctors attributed its powers to gods and goddesses, gradually weaving opium into Egyptian mythology. [3]

In the Greco-Roman world, the poppy plant featured prominently in literary tradition, with both Homer and Virgil drawing connections between opium-induced sleep and death. In Virgil's Aeneid (IX, 433-37) [4], for example, he writes of the one whose “neck droops: limp as a crimson flower…frail as poppies, their necks weary, bending.” Hippocrates too reportedly used opium along with white willow bark, which he recommended for fevers and aches. This willow bark is particularly significant as it contains a precursor to aspirin’s active ingredient, acetylsalicylic acid.

These ancient remedies, however effective, carried major burdens. Opium easily triggered dependence; prominent historical figures, such as Marcus Aurelius, reportedly consumed daily doses [5]. For centuries, civilizations accepted this trade-off because better options simply didn't exist. When sedation was needed – even for procedures as basic as tooth extraction or treating battlefield wounds – opium remained humanity’s best offering.

Chloroform, Ether, and the Birth of Modern Anesthesia

I want to take a short detour here to talk about anesthesia because while opium relieved certain aches or helped patients doze through lesser procedures, the agonies of major surgery remained horrifying until the mid-19th century. Surgeons worked as quickly as possible while assistants restrained the writhing patient. Speed was prized: the best “sawbones” performed amputations in mere minutes [6]. The notion that we could completely extinguish pain and any memory of it by rendering someone unconscious was, for most, inconceivable.

That changed in 1846, when a Boston dentist, William Morton, demonstrated the use of Ether for a tumor removal in what became known as the “Ether Dome” demonstration at Massachusetts General Hospital. Journalists and doctors were astonished: the patient lay still, felt no pain, and awoke unaware of the knife. Soon after, a Scottish obstetrician, James Young Simpson, began experimenting with Chloroform, testing the sweet smelling liquid on his friend in his dining room. While today it is most commonly associated with a mysterious white handkerchief in works of crime fiction, for many decades it was the go-to method of inducing unconsciousness. [7] [8]

However, intriguingly it was not widely accepted at first. The public was scandalized and some religious authorities condemned the effort to abolish pain in childbirth as meddling with the natural order. But the tide turned definitively in 1853, when Queen Victoria used chloroform for the birth of her eighth child. Knowing that it was good enough for the monarch, put most moral objections to bed and almost overnight anesthesia gained broad acceptance in Britain. [9]

Chloroform did have risks though. It came with a narrower safety margin than ether and an overdose could cause sudden cardiac arrest, claiming a patient’s life within minutes. Over time, anesthetists refined dosages, and the presence of a specialized “anesthesiologist” gradually became standard in operating rooms. For the first time, surgeries could be performed meticulously which helped continue the rising trend of a professionalized medical class.

Morphine, Heroin, and the Rise of Modern Opioids

While anesthesia was part of the medical puzzle, for most everyday uses opium still ruled, but it was the isolation of morphine in the early 1805 that launched a true pharmacological revolution. German pharmacist Friedrich Sertürner extracted potent crystals from raw opium resin and named them “morphium,” referencing Morpheus, the Greek god of sleep and dreams. Morphine quickly found its way into doctors’ kits: it was stronger than opium’s crude preparations, easier to measure, and when injected, provided rapid, dramatic relief [10]. Around the same time, the hypodermic syringe was perfected (1844-1851). Combined, morphine plus the syringe offered near-instant sedation of severe pain – an appealing prospect, especially in warfare. During the American Civil War (1861–1865), tens of thousands of soldiers received morphine injections for gunshot wounds or amputations. Although they recovered physically, many returned home with a new affliction: morphine dependence, or what was labeled “the soldier’s disease.”[11]

This kickstarted the quest to find an “even safer, less addictive” form of morphine led to more chemical feats: diacetylmorphine (later dubbed heroin), hydrocodone, oxycodone, and so on. Each debuted with optimism within a rather lax regulatory environment. Heroin, for instance, was sold by Bayer in the late 1890s as a fix-all cough remedy.

By the 20th century, opioids had branched into countless variants, from the moderate (like codeine) to the terrifyingly strong (like fentanyl). The analgesic power was never in doubt, making them indispensable for severe acute pain or end-of-life care, but so was their capacity to create a vicious cycle of addiction.

Aspirin, NSAIDs, and Over-the-Counter Relief

Despite opium’s saga, a more humble contender was waiting in the wings for simpler pains: derivatives of the willow bark. Moreover, while Bayer & Co. might have misstepped with heroin, Felix Hoffmann, a scientist at the company developed acetylsalicylic acid, soon marketed as Aspirin (1899). Compared to opiates, aspirin had fewer addictive properties and was particularly good at reducing inflammation and fever.

Aspirin caught on at a scale rarely seen before in medicine and started something of a revolution in the class known collectively as non-steroidal anti-inflammatory drugs (NSAIDs). Ibuprofen, introduced in the 1960s, claimed to be gentler on the stomach lining. Naproxen, diclofenac, and others soon followed. Collectively, these “over-the-counter” analgesics became a staple of everyday life, providing relief from headaches, muscle sprains, period pains, minor arthritis flare-ups and everything in between. [12]

Yet, most would soon be dethroned by a much earlier discovery – paracetamol. It lacks the anti-inflammatory label its counterparts boast but generally outsells them because it spares the stomach. While it is the go-to treatment for fevers and aches in all age groups, it is also the leading cause of acute liver failure and responsible for “more than 100,000 calls to Poison Control Centers, 56,000 emergency room visits, 2,600 hospitalizations and nearly 500 deaths” in the United States. [13] [14]

Where does this leave us? Opium and its derivatives are addictive, NSAIDs can cause gastrointestinal bleeding or kidney strain at high doses, and acetaminophen can destroy the liver. 

Nav1.7 (and Nav1.8) Inhibitors

The recent approval of Vertex Pharmaceuticals’ Suzetrigine (VX-548), marketed under the brand name Journavx, marks the beginning of what could become the next major chapter in analgesia. Unlike previous drug classes, Suzetrigine targets the Nav1.8 sodium channels to silence pain directly at its source in peripheral nerves, circumventing the brain’s reward circuits entirely.

The inspiration for this approach came from a rare genetic anomaly known as Congenital Insensitivity to Pain (CIP), where individuals lack functional Nav1.7 sodium channels, rendering them completely unable to sense pain. The idea was that if scientists could block this specific channel pharmacologically then they might be able to replicate this “pain-free” state without the addictive and sedative consequences of opioids. However, early clinical trials of Nav1.7 inhibitors produced mixed results, revealing that targeting just that channel was insufficient in reliably extinguishing pain – it turns out that the bodies of individuals with CIP also compensate by producing heightened levels of natural opioids (like endorphins and enkephalins). [15]

Vertex Pharmaceuticals then shifted focus slightly, concentrating on Nav1.8 which is a closely related channel that also transmits signals for inflammatory and neuropathic pain. Their new compound, VX-150 demonstrated notable efficacy in early human trials, paving the way for their next-generation molecule, VX-548 (Suzetrigine). Clinical studies from 2022 onwards confirmed VX-548’s effectiveness, showing significant pain relief following surgeries like bunionectomy and abdominoplasty, rivaling opioid analgesics without the same side effects. [16]

In general but also in context of the story so far, the FDA’s 2025 approval of Suzetrigine thus represents a landmark moment. It introduces the first non-opioid oral analgesic class in decades, capable of substantial acute pain relief with minimal adverse effects – no respiratory depression, no sedation comparable to opioids, and crucially, no addiction risk. 

Challenges remain however, with Phase-3 trials of Biogen’s Nav1.7 inhibitor, vixotrigine, withdrawn after mixed results during Phase 2. Despite these setbacks, Vertex’s breakthrough with Nav1.8 inhibition represents a necessary step forward; perhaps overcoming pain is within our scientific reach. While history has repeatedly shown that pain relief is rarely simple, we might just be able to make pain a thing of the past for everyone. [17]

References:

[1] Timothy D. Warner and Jane A. Mitchell, "Cyclooxygenase-3 (COX-3): Filling in the Gaps Toward a COX Continuum?," Proceedings of the National Academy of Sciences of the United States of America 99, no. 21 (2002): 13371-13373, https://doi.org/10.1073/pnas.222543099.

[2] Anti-inflammatory Drugs Market Size to Hit USD 272.35 Bn by 2033," BioSpace, June 18, 2024, https://www.biospace.com/anti-inflammatory-drugs-market-size-to-hit-usd-272-35-bn-by-2033.

[3] Victor B. Stolberg, "Chapter 3," in Painkillers: History, Science, and Issues (Santa Barbara: Greenwood, 2016), 29-65.

[4] Virgil, Aeneid, Transl. Robert Fagles.

[5] Thomas W. Africa, "The Opium Addiction of Marcus Aurelius," Journal of the History of Ideas 22, no. 1 (1961): 97-102, https://doi.org/10.2307/2707876.

[6] Matt Soniak, "'Time Me, Gentlemen': The Fastest Surgeon of the 19th Century," The Atlantic, October 24, 2012, https://www.theatlantic.com/health/archive/2012/10/time-me-gentlemen-the-fastest-surgeon-of-the-19th-century/264065/.

[7] Massachusetts General Hospital, Archives: https://www.massgeneral.org/museum/exhibits/ether-dome.

[8] Nicole Erjavic, "James Young Simpson (1811–1870)," Embryo Project Encyclopedia, Arizona State University, July 23, 2017. https://embryo.asu.edu/pages/james-young-simpson-1811-1870.

[9] Joe Schwarcz, "Anesthesia a la Reine: Chloroform Used to Be a Widely Used Anesthetic, Dating Back to the 1800s," McGill Office for Science and Society, September 21, 2022, https://www.mcgill.ca/oss/article/medical-history/anesthesia-la-reine.

[10] Rudolf Schmitz, "Friedrich Wilhelm Sertürner and the Discovery of Morphine," Pharmacy in History 27, no. 2 (1985): 61-74, http://www.jstor.org/stable/41109546.

[11] Norman Howard-Jones, "A Critical Study of the Origins and Early Development of Hypodermic Medication," Journal of the History of Medicine and Allied Sciences 2, no. 2 (1947): 201-49, http://www.jstor.org/stable/24619591.

[12]V. Wright, "Historical Overview of Non-steroidal Anti-inflammatory Drugs," Rheumatology 34, suppl. 1 (1995): 2-4.

[13] Samir S. Ayoub, "Paracetamol (Acetaminophen): A Familiar Drug with an Unexplained Mechanism of Action," Temperature 8, no. 4 (2021): 351-371, https://doi.org/10.1080/23328940.2021.1886392.

[14] Cristina I. Ghanem, María J. Pérez, José E. Manautou, and Aldo D. Mottino, "Acetaminophen from Liver to Brain: New Insights into Drug Pharmacological Action and Toxicity," Pharmacological Research 109 (2016): 119-131, https://doi.org/10.1016/j.phrs.2016.02.020.

[15] Michael S. Minett, Vanessa Pereira, Shafaq Sikandar, Ayako Matsuyama, Sonia Lolignier, Alexandros H. Kanellopoulos, Flavia Mancini, et al., "Endogenous Opioids Contribute to Insensitivity to Pain in Humans and Mice Lacking Sodium Channel Nav1.7," Nature Communications 6 (2015): 8967, https://doi.org/10.1038/ncomms9967.

[16] Vertex Pharma, "Vertex Announces Positive Results From the VX-548 Phase 3 Program for the Treatment of Moderate-to-Severe Acute Pain," January 30, 2024. https://news.vrtx.com/news-releases/news-release-details/vertex-announces-positive-results-vx-548-phase-3-program.

[17] Annalee Armstrong, "Biogen's Non-opioid Painkiller Offers Mixed Results in Phase 2 Trial, This Time in Sensory Pain Disorder," FierceBiotech, September 16, 2021, https://www.fiercebiotech.com/biotech/biogen-s-vixotrigine-fails-phase-2-small-fiber-neuropathy-trial.

I call this space “Mundane Beauty” because of a belief that humans have an innate propensity to imbue the world with their contemporary understanding of plaisir (delight/joy/contentment/pleasure). History is but our method of investigating these changing notions and expressions of beauty so that we may better know ourselves. Similarly, science, at-least for a long time, was thought of as a tool to understand the wonders of the natural world (ref; the idea of God-given beauty in the West). Naturally, there is some (significant) overlap between the aesthetics we hold dear and those we are able to observe in the universe. I guess there is a reason why people who made this intersection of the past and the future their vocation were called ‘humanists’ in the Renaissance.

However, just because Ancient Greek poetry is no longer a prerequisite for a scientific education (as was the case for the humanists), does not mean that modern-day STEM draws any less from the ‘standard of beauty’ we express so freely in our artistic traditions. We are just less conscious of it.

This became apparent to me when a friend of mine (Hi Samarth!) started working towards teaching himself math equivalent to doing an “unofficial undergraduate major.” Often, he would use words like ‘elegant’ or ‘beautiful’ to describe certain proofs he found pleasing just as he would use ‘ugly’ and ‘non-intuitive’ to characterize those that did not seem as clear to him. This vocabulary is but a feature of the field of mathematics which relies heavily on non-verbal, implicit ideas of beauty, often to epistemic ends. Generalizable, symmetric, eliciting-the-feeling-of-placing-the-final-piece-in-an-incomplete-puzzle (and I am sure there are plenty of other ways of defining mathematical beauty) proofs have consensus around their truth sooner. Ugly ones less so, often being subject to many iterations of reframing till a certain standard is met [Euclid’s fifth postulate is a very good example and I will talk about it more in a later essay]. More astoundingly, this superficial heuristic tends to be a generally good indicator of both verity and usefulness. Is this phenomenon not worth investigating?

It is in following this train of thought that I stumbled into the works of Henri Poincaré — a giant in the French mathematical scene in the late 19th and early 20th centuries. He is best known for being the ‘father of topology' due to his vast contributions to that sub-field but he also dabbled in early non-euclidian geometry, math related to the study of relativity, geodesy, and most importantly for my purposes: publicly accessible philosophy of science. It was this philosophy that made coherent the underlying mechanisms of doing math and attempted (at-least briefly) to elucidate why exactly aesthetics mattered in STEM.

“It is, therefore, the quest of this especial beauty, the sense of the harmony of the cosmos, which makes us choose the facts most fitting to contribute to this harmony, just as the artist chooses from among the features of his model those which perfect the picture and give it character and life.” — Poincaré (Science and Method, transl. George Bruce Halsted)

Concurrently, while I was making my first pass through his written work, I was asking those around me to explain one mathematical concept they found beautiful and what made it so [I would be incredibly grateful here if you took a moment to fill out the following with more of the same questions: form]. Here there was an interesting conflation between geometry (both Euclidean and not) and beauty — most likely because of its visual nature. More specifically, a lot of the math Poincaré influenced elicited these feelings of delight.

My hunch here is that those who were conscious of the philosophical value of their work often implicitly perpetuated those ideas even in their more technical presentations.

This leads us to what I am trying to do: In following this hunch, I am trying to write a history of mathematical aesthetics. For practical reasons (time mostly), I am limiting the scope to finding the ‘standard of beauty’ within the brief history of non-Euclidian geometry. I will move from Gauss to Poincaré to see how the evolution of the field demanded a focus on aesthetics. Subsequently, I will show the link between that story and Poincaré’s philosophy of science (relying on my own close readings alongside secondary scholarship). Lastly, time and space in my senior thesis permitting, I want to provide an overview of Poincaré’s influence on mathematical teaching after him — i.e. why is he still relevant today and why should we care about beauty in STEM more broadly?

This work has led me to France, and I am in Paris till the 20th of July doing archival research: if you happen to be here and are interested in the questions I am asking [or think they are not particularly novel/worthwhile], I would be very happy to discuss things further over coffee. Please send me an email at jainhiya1212@gmail.com, alternatively, my DMs are open on Twitter.

I will also continue writing about the story on this blog as it becomes clearer to me, often including full excerpts from the final work. Feedback would be very helpful!

With special thanks to everyone at Archives Henri Poincaré at Univ de Lorraine, Nancy 2 (especially Prof Laurent Rollet and M. Pierre Willaime) for their help over the past week. Not only were they very helpful with resources, but conversations with them were immensely productive in finding the story-of-best-fit so to speak.