The Path of Heterodoxy

I shall be telling this with a sigh
Somewhere ages and ages hence:
Two roads diverged in a wood, and I—
I took the one less traveled by,
And that has made all the difference.

-Robert Frost

I was born an orthodox scientist, plopped out onto a conveyer belt of conventional academic promise. My mom was a renowned molecular biologist and mentor who nurtured my scientific curiosity from a young age, who explained cells while we sat around the dinner table, who told me tales of photosynthesis catalyzing the conversion of carbon dioxide to sugar and oxygen when I was only 5 years old.

“What’s carbon dioxide?” I would ask.

“Great question!” my mom would exclaim with contagious enthusiasm. “The whole universe, everything you can see, feel, and touch, is made up of atoms…” She grabbed my arm and said “if you zoom in to your skin, you’ll find it’s made up of living bubbles called cells. If you zoom into cells, you’ll see they’re made up of even smaller things called molecules - fats, proteins, sugars, nucleic acids - and if you zoom into molecules, you’ll see they’re made up of building blocks called atoms. Carbon dioxide is a molecule, made up of one carbon atom and two oxygen atoms. Carbon dioxide is a gas at room temperature… when we breathe in, we breathe in oxygen, and when we breathe out, we breathe out carbon dioxide…” She would go on, connecting the air I breathe out to the air plants breathe in to the sugars I ate in my cereal.

Suffice to say, born with such raw scientific privilege of a brilliant mom, I did well in school. I loved everything I studied, majored with two degrees - biology and applied mathematics - and almost had minors in chemistry and psychology. I did research as an undergraduate and I was accepted into a prestigious National Science Foundation Graduate Research Fellowship Program. I applied for many PhD programs and was accepted at Harvard, Stanford, Princeton, and more. I received a PhD from Princeton in a quick 4.5 years studying under a famous advisor - Dr. Simon Levin - and I went on to a postdoctoral fellowship at Duke University.

I continued along the path of academic orthodoxy, working on a DARPA-funded team studying bat virus spillover starting in 2017. I helped the team I worked with write a successful DARPA PREEMPT grant aimed at forecasting the spillover of bat henipaviruses and preempting the emergence of these dangerous viruses. I published papers, developed new and innovative methods at the frontiers of both mathematics and biology. I followed a primrose path pointing straight to the halls of the academy lined by portraits of academics past, with high hopes for my own portrait at the end of the hall.

That was then. The orthodox path I looked down in 2019 is now choked in fog, the regal halls and portraits are swallowed by flames, the academic dream is dead.

What happened? Where did I go awry?

At what point did I take the path less traveled by?

My Heterodox Sins

As a scientist, I can’t help but break down what I’ve observed into a series of concrete units of time & causality, events whose mysteries and reasons I can hope to understand. If I understand each event individually, perhaps the whole mess of events can make more sense.

The COVID-19 pandemic was the beginning of the end of my orthodox self-image. A series of events began to reveal that I am not conventional, and in these events not only did I realize I am not conventional, but I realized that the academy is unkind to the unconventional. Below are the events of interests, the unconventional stances I took, and the treatment received.

Predicting a pandemic

My job on the DARPA PREEMPT team was primarily to forecast the evolution of pathogens and forecast the onward transmission of pathogens once they emerge. When SARS-CoV-2 emerged in humans, I immediately began forecasting.

In late January to early February 2020, I had analyzed case growth rates in Wuhan, case reports finding high false negative rates of PCR tests, the classic Bavarian case study of transmission with unspecific symptoms, and more, and I came to believe a pandemic was inevitable - my forecasted chance of a pandemic was so near 100% that I began buying groceries and warning my family. My estimates of case growth rates were much faster, and this was the most important finding that changed one’s view. My methods differed from conventional methods because I had experience estimating growth rates of stocks with methods not commonly used in epidemiology, so I was aware of the many methods, knew the limitations of estimating growth rates with SEIR models, and prioritized other methods that addressed these limitations. Faster case growth rates led to estimates of higher prevalence, higher prevalence increased confidence in a larger subclinical iceberg of cases, higher prevalence sets back the estimated date of outbreak onset in places connected to Wuhan, more subclinical cases decreases the odds of traveler screening stopping a pandemic, and all of this increases the odds of a pandemic.

I tried to share my findings of 2-3 day doubling times with academics on my DARPA PREEMPT team, but one professor from Oxford memorably said he doesn’t have time for me and, simply put, has far more confidence in teams from Harvard, Imperial College London, and the London School of Hygiene and Tropical Medicine than “a postdoc in Montana”. The large institutions all estimate 6.2 day doubling times, low subclinical rates, and high odds of successful containment, so on their credentials alone he - my own teammate - did not have time for my heterodoxy. Others in my orbit were similarly averse, with one professor claiming that if I shared my results and I was wrong, it could get picked up by Fox News, trigger complacency, and I could be directly responsible for millions of deaths.

In this one instance, I chose to keep my findings private. These are the warnings the world never heard, the heterodox Cassandra was just a postdoc in Montana, told they are not an epidemiologist and pressured to stay in their lane. Instead of the academy discovering and appreciating diverse perspectives, or finding talent irrespective of the institution where talent is enjoyably employed, a social beast reared its head and discouraged me from sharing my findings, so rather than these findings informing people of an oncoming pandemic and helping folk shop for groceries in February, these findings were used to help a hedge fund short the market.

The Influenza-Like Illness (ILI) Paper

Because I believed there was a higher prevalence and faster growth rate, I hypothesized there would be a surge of patients seeking care in major US metro areas like NYC by March-April 2020. As a consequence of this mathematical foresight, Justin Silverman, Nathaniel Hupert, and I kept our fingers to the pulse, monitoring the CDC’s influenza-like illness dataset reporting the fraction of patients visiting sentinel primary care providers with influenza-like illness (ILI).

We found a massive surge of patients in March of 2020. We converted this surge in the fraction of patients experiencing ILI to an estimate of the raw number: 8 million excess patients, far more than the <100,000 cases documented at that time. Our paper was picked by The Economist, and immediately we were attacked by fellow scientists (not random trolls) who accused us of disrupting the public health message, minimizing COVID (for more cases meant fewer deaths per case), and triggering complacency. Even within my own workplace, some of the scientists who discouraged me from sharing my February 2020 findings went further and began piling on in Twitter rants or even in our Slack channels, finding many opportunities to criticize me or frustrate my efforts to belong in the scientific community.

The ILI episode happened at a time when other heterodox scientists were also being lambasted on Twitter, with academics pulling levers at media outlets to increase the scale and scope of academic beef from a snide email to a public outcry delegitimizing some for the sin of being different, diverse. Any one academic may say they did little - they only cast one stone on any given week, retweeted a few of the attacks, rejected one paper, and posted a nasty thing here or there, yet the collective effect of this high volume of attacks lobbed toward a small group of scientists wasn’t only a swarm of hostility felt by the heterodox scientists, but a broader normalization of hostilities by the orthodoxy against scientists who strayed from “The Science” and “the message” it supports. Normalizing such hostile rhetoric against any who diverged from the norm allowed every person to arm themselves with a small quiver of flaming arrows, and as they shot their arrows up and down the halls of the academy, striking curtains and portraits on the wall, the temperatures rose. The social dynamics felt as if the academy had burst into flames. As wooden beams burned and fell from the ceiling, the orthodoxy was so fixated on violence their first instinct was that such flaming beams could be weaponized and used as bludgeons. In the minds of the orthodox, everything needed to be done to protect science and scientists, even if that means attacking science and scientists with the foundations of what was once our shared home. Everything, from voices and votes to the pillars of research ethics and seats at the table of editorial boards, was weaponized to defend The Science.

Avoiding the flames, dodging the bludgeons, ducking the flying chairs, and coding beside the burning portraits, I continued to watch the data and evaluate competing theories.

Fall 2020 Forecasts

Under my theory, the theory which predicted a pandemic and the March 2020 surge with 2-day doubling times of ICU admissions across NYC providers, it was probable that the March-April 2020 surge in New York City ended because of the depletion of the susceptible population, a state of the population that prior to COVID we referred to as “herd immunity”.

If this were true, then we’d be able to compare future outbreaks to the NYC outbreak and find that future outbreaks saw cases peak at a population fatality rate similar to that of NYC. Jutsin Silverman and I developed a dashboard we would use to monitor outbreaks for comparison with the NYC curve and, in the summer of 2020, we saw many unmitigated outbreaks corroborate “The NYC Line”. The figures below take some time to understand, but they are essentially time-series on a timescale of burden, allowing us to compare at similar burdens the growth rates of outbreaks that take place on different dates.

In August 2020, I shared these findings with the CDC forecasting team, providing context for future outbreaks - should the US outbreaks see cases peak at 1 death per 1,000 capita, my theory developed since February 2020 would provide an explanation why: herd immunity.

The fall 2020 outbreaks across the United States universally corroborated “The NYC Line” as an upper-bound quantile of outbreak intensity (both speed & burden), especially those outbreaks in states or counties that did little to mitigate the spread of COVID-19.

The lagged population fatality rate, D(t+11), at peaks in cases across 3,000 US counties all converged to the NYC line prior to the arrival of vaccines. This peak is anomalously below the 4x more severe dashed-line scenarios of orthodox models.

I made pretty figures that illuminated this finding beautifully. However, in the Fall of 2020 the Great Barrington Declaration (GBD) was written pointing out that our models may be overestimating the severity of the pandemic (which my findings also suggest) and containment policies may cause harm. While the GBD was called the “herd immunity” strategy, the authors didn’t mention herd immunity, although I personally believe that Fall 2020 outbreaks peaked due to herd immunity, as evidenced by their convergence with the NYC line. Below is the inferred phase diagram of all >3,000 US counties’ outbreaks funneling towards their peaks underneath the NYC line long before vaccines arrived and at burdens far less than conventional estimates indicated by the dashed line.

In October 2020, my postdoc advisor demanded I not share my findings. She was worried that my claims about herd immunity could disrupt her relationship with the state government, as the state government was giving her $1 million to do test & trace experiments on the campus population, experiments that were not needed under my theory. Under my theory, the state government ought to allocate resources towards tribal and rural communities that lack non-congregate housing and access to care, not toy around with experiments quarantining college students.

I left my position at Montana State University over these disagreements, ultimately drawing on the lesson I learned from the COVID short: if seeing something and saying something within my expertise goes against convention, then I would rather share my findings and be criticized than not share my findings and watch the world suffer in unprepared ignorance.

Defense of John Ioannidis

Dr. John Ioannidis is a renowned professor at Stanford. He wrote articles early in the COVID-19 pandemic arguing that we were making decisions in the face of uncertainty, that our policies could cause harm and the pandemic may not be as bad as forecasted by conventional teams.

Dr. Ioannidis became reviled by vocal academics on Twitter. These academics believed Ioannidis was a minimizer, and they would criticize him at every opportunity.

At one point, Dr. Ioannidis published a paper finding that academics with many Twitter followers didn’t actually have that many papers or citations, in other words the people centering themselves as “the experts” on Twitter may not actually the best people to consult for expertise. Dr. Ioannidis published the paper with Twitter handles, and many academics began a crusade claiming that publishing a paper with Twitter handles is unethical, a violation of IRB rules, and a form of harassment that could lead to scientists being targeted.

I dissented. I argued that this rule was being made up - Twitter handles are publicly available data and there is no need for an IRB to analyze and publish data in the public domain. Furthermore, many of the same people criticizing John Ioannidis for harassment of scientists had broadly shared and supported an MIT group publishing handles they labelled “anti-mask”. I argued that if there were any ethics at issue here, it’s the selective invention of research ethics outrage to attack Ioannidis, and the questionable ethics of scientists attaching derogatory labels like “anti-mask” to groups of people they color as anti-mask in their diagrams.

Waning immunity

Under the hypothesis that Fall 2020 outbreaks peaked at herd immunity thresholds, the only way we can explain follow-up outbreaks is by a mix of waning immunity and immunoevasion.

I briefly touched my feet on the path of orthodoxy by working with a team at The London School of Hygiene and Tropical Medicine as they made sense of the first variant of concern, Alpha. I was working at hedge funds analyzing SPACs (special purpose acquisition companies, basically private equity vehicles for IPOs) and provided a punny “IPO” or “initial phylogenetic observation” analysis of the transmissibility of B.1.1.7.

Under my theory, however, we would expect to see Alpha cause larger waves in places with earlier outbreaks, presumably places that had lower Fall 2020 surges. That is exactly what we saw, and this provided evidence of waning immunity.

Unfortunately, waning immunity was considered misinformation at the time of this discovery as waning immunity implied vaccine-induced immunity may wane. If natural and vaccine-induced immunity waned, then we’d expect to see surges in the US maybe 6 months after the Fall 2020 outbreak and the transmission rate in these follow-up outbreaks would be most negatively associated with the cumulative burden in these outbreaks and have no negative correlation with vaccination rates.. Indeed, by July 2021, that is what we saw.

If immunity wanes, then the cost-benefit ratio of a policy mandating vaccines changes. One might justify mandating vaccines if the Fall 2020 outbreaks were not natural herd immunity thresholds and immunity does not wane. Unfortunately, from my theoretical vantage point, it was likely that Fall 2020 outbreaks peaked at natural herd immunity thresholds and immunity waned, calling into question the wisdom of mandating vaccines that carry some risks and weren’t going to prevent future infections.

Eventually, in August 2021, the CDC published findings of an outbreak in Provincetown demonstrating indisputable evidence of 100% immune evasion, where patients with vaccines had the same incidence as patients without vaccines.

This same theory of natural herd immunity thresholds in the Fall of 2020 and immune evasion in Delta led to the theory that future outbreaks would have characteristic timescales as they deplete the susceptible populations in relatively unmitigated outbreaks burning through populations whose immunity waned. Above, we see the Omicron outbreaks across South African provinces defining the characteristic timescales of Omicron outbreaks across US states, again corroborating this theoretical path I travelled by.

COVID Origins

Recall that before COVID I was studying pathogen spillover from bats to people, and statistical methods for attributing pathogens to hosts or sources. Once I was done with outbreak forecasts, I returned to the question of pathogen spillover, of attribution of this biological agent to the source whence it came.

I read the literature and found it to be lacking, to be making uncharacteristically strong conclusions based on flawed data, flawed analyses, or unjustified assumptions. I wrote an article critiquing one paper, Pekar et al., claiming the two big branches at the base of the SARS-CoV-2 evolutionary tree is evidence of two spillover events. We showed that the authors’ methods did not justify their conclusions, and additionally they excluded sequences which invalidate their premise of two big branches at the base of the SARS-CoV-2 evolutionary tree. This angered zoonotic origin proponents and put a target on my back - peruse the comments of our preprint to get a glimpse.

Far more heterodox was the next paper I helped write with Valentin Bruttel and Tony VanDongen. The lab origin theory revolved around a furin cleavage site hypothesized to have been inserted, but to insert a furin cleavage site in a coronavirus one would need to make a DNA copy of the virus. We found evidence in the SARS-CoV-2 genome consistent with the dominant methods for making DNA copies of coronaviruses.

This, too, was picked up by The Economist as well as Telegraph and more. In each case, however, these journals elevated the controversy manufactured by the guardians of the orthodoxy, all of whom have close ties to either the Wuhan Institute of Virology believed to have conducted this work, or NIAID, the US-based health science funding agency found to have funded in 2019 the exact researchers who proposed to insert a furin cleavage site inside a DNA copy of a SARS coronavirus in Wuhan in 2018.

My Punishment

While recent evidence has corroborated our theory, much like out of sample evidence corroborated my theory of COVID outbreaks and herd immunity thresholds, corroboration does not seem to be sufficient to allow reentry to the academy. The toxic tone of disagreements, the bruised egos of big-name professors who were proven wrong by the data itself, has created a community of toxic, indignant orthodoxy. There is no hope of readmission for a sinner like me, for after enough battles and bruised egos, there will be on every committee, every editorial board, every conference board, someone I have offended with my heterodoxy.

Now, in engagements with orthodox academics desperate to defend their turf from the overwhelming hordes of data and evidence, I am labelled by a Scarlet Alphabet of heterodoxy. Antivax, Brownstone Boy, COVID minimizer, Denialist, and so on. I am in the eyes of many a pariah because I had the insight to discover things and the courage to stand by my discoveries. Few things worsen an indignant senior’s ego more than a junior’s refusal to yield, it’s as if I’m blamed for the boxer’s fracture in the fists of those who punched down at me.

I wanted to share this journey for future scientists looking ahead at their careers, unsure which path to take. The forks in the road I’ve encountered here are going to be encountered by others. You may find something, something uniquely visible to someone with your expertise, and you may encounter everything from discouragement and recommendations to not share your work, to outright hostility and, if you do share your work, alienation from people you once thought were your colleagues or friends.

You will see signs pointing you to orthodoxy. You will look up the miserable, lonely, thorny thicket of controversy along the path of heterodoxy, and you may understandably be tempted to not bear the Scarlet H. If you seek to maintain your career as you always imagined it, if you seek comfort and collegiality, then by all means, do not stick your neck out. Avoid controversy. Never criticize people who occupy positions of power. You may get your portrait raised on the halls of a building, and you will never know anything but love for the building that has included you, or at least the uncontroversial version of you.

My departure from the downhill slope of orthodoxy and ascent up the grueling hills of heterodoxy has, indeed, made all the difference. In my heart, I’m still my mother’s child curious about carbon dioxide, fascinated by photosynthesis, in awe of atoms. I’m not quite the scientist I was before... I am better.

By following the path of heterodoxy, I have been scraped and bruised, punched and kicked, beaten and harangued, posted on a digital pillory and had insults thrown at my face. I would do it all again, for only on this path do you discover who you are. If you are going to be a heterodox scientist, you will need to be the best scientist in the room, with results that withstand the scientific fury of your detractors. If you are going to be a heterodox scientist, you need to have the self-confidence and self-love to survive the loneliness of going against the world. With time, however, you will find others who have also gone against the world, others also courageous enough to stand by their principles. A single heterodox friend is a more powerful ally than 100 orthodox colleagues, for the orthodox colleagues with slither away the second you need unconventional help whereas the battle-hardened heterodox ally will understand the value of friendship in times like these. Here on the windy, rainy hilltop of heterodoxy, we stay warm thanks to the company, we don’t live in darkness thanks to the glow of those around us. If you survive the path of heterodoxy, you will find a community of the most brilliant minds, true friends who prioritize belonging over belief, and you will learn to laugh while you are being punched, to get stronger while you are being attacked, to love while you are hated.

I only recommend the grueling ascent of heterodoxy to the strongest, most brilliant, most authentic, and courageous friends. If you can survive the ascent, then on the other side you will discover your strengths, your peers, and your purpose.

You will be despised by despicable people who found you useful when you molded your soul to fit their grasp, and you will be admired by the admirable heterodox scientists who climbed this mountain before you. You will be loved and respected among a diverse community of authentic, untrammeled souls, and even if we disagree, we will do so politely and in a way that ensures you still belong in our community, for few understand the corrosiveness of intolerance better than those of us who have been exiled to this heterodox city upon a hill.