Gain of Function

What it is, who built it, and why the people who built it were the last ones who could admit what it produced.

Gain of function research wasn't a rogue program. It was the logical endpoint of a twenty-year American biosecurity strategy, built deliberately, funded consistently, and defended against critics by the same people who ran the February call. Daszak didn't cause this alone. The institution he served might have. Here's how it was built.

---

The theory is not obviously crazy.

Nature produces novel pathogens on its own timeline, indifferent to our preparedness. A bat coronavirus in a cave in Yunnan is one mutation event away from efficient human transmission at any given moment, and we will not know it is coming until it arrives. The argument for gain of function research is that if we can identify which mutations produce that transition, study them in controlled conditions, and develop countermeasures in advance, we are net ahead of the natural timeline. We know what the threat looks like before the threat knows it exists.

That is a real argument. Serious scientists made it seriously. The research program it produced was not the product of negligence or malice. It was the product of a theory of biosecurity that captured an institution, built a funding apparatus around itself, and then became too large and too load-bearing to interrogate honestly.

The interrogation happened anyway. The institution suppressed it. When the thing the interrogators warned about potentially materialized, the institution suppressed that too.

This is the piece about how the institution was built.

---

Gain of function research, in the relevant sense, means taking a pathogen and deliberately enhancing its transmissibility, virulence, or immune evasion properties. The stated purpose is to get ahead of natural evolution: if you know what an enhanced version looks like you can develop vaccines and treatments before nature produces it.

The unstated premise is that you can do this safely. That the enhanced pathogen stays in the lab. That the containment protocols hold. That the humans running the experiments don’t make mistakes, don’t get infected, don’t carry something home on a surface or in a breath.

That premise has never been stress-tested at scale. It has been asserted, funded, and institutionally protected. The scientists who questioned it were not, for the most part, arguing that the theory was wrong. They were arguing that the gap between “this works in principle” and “this works in a BSL-2 lab in Wuhan with the containment protocols that facility actually has” was large enough to matter.

Fauci’s apparatus did not find this argument compelling. The grants continued.

---

The apparatus

Anthony Fauci has run NIAID since 1984. Forty years. In that time NIAID’s budget grew from roughly $320 million to over $6 billion. The coronavirus research program at WIV was a small line in that budget. EcoHealth Alliance’s total NIH funding across all programs ran to tens of millions over the relevant period. In absolute terms, not large.

In structural terms, the coronavirus program was not about the money. It was about the paradigm.

Fauci had spent two decades building the intellectual and institutional case that emerging infectious disease was the central biosecurity threat of the 21st century, that the American research apparatus needed to be positioned ahead of that threat, and that gain of function research was a necessary tool for that positioning. NIAID’s budget growth was downstream of that argument. The revolving door between NIAID, the major research universities, and the pharmaceutical companies developing the countermeasures was downstream of that argument. Fauci’s position as the indispensable expert, the man who understood the threat better than anyone, was downstream of that argument.

The argument needed to be true. The people whose careers it had built needed it to be true. When it potentially produced the thing it claimed to be defending against, the cost of admitting that was not just reputational. It was structural. The entire edifice rested on the premise that this research was net positive. One pandemic later, the premise was a question.

They decided the premise was not a question.

---

The interface

Peter Daszak understood something that pure academic researchers often don’t: the funding interface is the power position.

EcoHealth Alliance is nominally a wildlife conservation organization. Its actual function, in the relevant period, was to serve as the conduit between American biosecurity research dollars and research programs that NIH couldn’t fund directly, in countries where American researchers couldn’t operate openly, under a conservation and public health framing that attracted less congressional scrutiny than a weapons-adjacent biosecurity framing would have.

Daszak was not a virologist. He was an ecologist and a fundraiser. His specific skill was translating the gain of function research paradigm into grant language that moved money from Fauci’s apparatus to Shi Zhengli’s lab at WIV, and then translating the results back into publications and press appearances that justified the next round of funding.

That is a valuable institutional function. It made Daszak irreplaceable to both sides of the transaction. WIV needed the American funding and the publication access it enabled. NIAID needed the field research and the BSL-4 lab access that EcoHealth’s WIV relationship provided. Daszak sat in the middle and collected the rent.

DEFUSE was the natural product of that position. Having established himself as the indispensable interface, Daszak proposed to expand the program: more viruses, more enhancement, more countermeasure development, more publications, more funding. The DARPA application was an attempt to diversify the funding base, to bring defense money into a program that had been running on public health money, under a pandemic preparedness framing that the post-9/11 biosecurity apparatus should have found compelling.

DARPA looked at it and said no.

---

What DARPA’s no actually means

The DARPA rejection memo is worth reading in full if you haven’t. It is written in the affectless language of program officers who have evaluated a lot of proposals and know what they’re looking at.

DARPA’s reviewers concluded that DEFUSE proposed work that could potentially create “enhanced potential pandemic pathogens,” that the proposed biosafety standards were insufficient for that risk profile, and that the gain of function approach was “not fully in compliance” with the relevant regulatory frameworks. They were not saying the science was bad. They were saying the risk was unacceptable.

Here is what that judgment actually represents.

DARPA is the institutional heir to the Manhattan Project. It was created in 1958 as the direct response to Sputnik, but its intellectual lineage runs straight through the postwar national security apparatus: through the people who watched what happened when you gave scientists an open-ended mandate to build the most dangerous thing imaginable and then tried to contain what they built. The Manhattan Project produced two lessons that are directly relevant to this story.

The first: containment of novel dangerous capabilities degrades over time and across institutions. The Soviets had the bomb within four years. Not primarily through espionage, through the fact that the physics was knowable and the engineering was solvable once you knew it was solvable. The assumption that you can build a dangerous capability and hold it inside a defined perimeter is repeatedly falsified by history. DARPA was built by people who knew this from direct experience.

The second: the scientists closest to the work are structurally the worst judges of whether it should be built. Oppenheimer’s team thought hard about the ethics. They built it anyway, because the work was compelling and the theory was sound and the threat framing was real. The lesson DARPA took from this was institutional: you need a layer of judgment that isn’t captured by disciplinary enthusiasm. The program manager structure exists precisely for this. A virologist’s job is to understand viruses. A DARPA program manager’s job is to ask whether understanding this particular virus in this particular way produces a downside scenario that the upside doesn’t justify.

When DARPA says a biosafety proposal is inadequate, that is not a bureaucratic finding. That is the heir to Trinity telling you the blast radius is wrong. These are people whose institutional memory includes having built things that escaped their intended containment envelope, at scale, with civilization-level consequences. Their “no” carries epistemic weight that Fauci’s “yes” simply doesn’t, not because DARPA is more virtuous, but because DARPA has the relevant scar tissue and NIAID does not.

NIAID is optimized to say yes to science. Its selection pressure runs toward “yes, with oversight.” DARPA is optimized to say yes to capability and no to uncontrolled risk. They are different institutions solving different problems. When the two looked at the same proposal and reached opposite conclusions, the correct prior was to weight DARPA’s “no” more heavily. The specific question being decided, “is this risk class manageable in this facility with these protocols”, is exactly the question DARPA’s institutional structure was built to answer. It is not the question NIAID’s institutional structure was built to answer.

Nobody weighted it that way. That is not just an error in judgment. It is a category error about which institution has the relevant expertise on the question actually being asked.

The DARPA rejection was dated 2018. The NIH/EcoHealth funding channel to WIV was already running and continued running after DARPA passed. The work DARPA declined to fund was adjacent to work that was already funded. The biosafety concerns DARPA put in writing did not produce a review of the existing program.

Fauci’s apparatus had its own assessment of the risk. It differed from DARPA’s. The grants continued.

---

The moratorium

The moratorium is where the suppression of internal dissent becomes visible.

In 2014, enough scientists raised enough alarm about gain of function research that the Obama administration paused federal funding for certain categories of it. The pause was not a ban. It was a holding pattern while the relevant agencies developed a review framework for determining which GOF research was acceptable and which wasn’t.

The review framework was developed by the same institutional apparatus that had been funding the research the moratorium paused. NIAID had a seat at the table. The framework that emerged was called the P3CO (Potential Pandemic Pathogen Care and Oversight) framework. It was implemented in 2017. The moratorium lifted.

The scientists who had raised the original alarm did not find the P3CO framework reassuring. Marc Lipsitch at Harvard, Richard Ebright at Rutgers, and others argued that the framework’s review criteria were too narrow, that the review process lacked independence, and that the institutional incentives of the people running it pointed in one direction. Their papers were published. Their arguments were on record. The funding continued anyway.

Ebright had been making the specific argument that gain of function research at WIV posed unacceptable biosafety risks for years before 2020. He was not a fringe figure. He was a credentialed biosafety expert at a major research university. His argument did not require the virus to be engineered: even natural pathogens enhanced in a lab with insufficient containment protocols could escape. The question was not whether the research was malicious. The question was whether the safety infrastructure matched the risk.

The NIAID-funded apparatus had a response to Ebright’s argument: the research was necessary, the safety protocols were adequate, and the critics were outside the relevant expert consensus. The grant cycle continued.

The critics turned out to be right about the risk profile. Whether they were right about this specific event is still, technically, unproven. The behavior of Fauci’s apparatus in February 2020 is consistent with people who believed they might be.

---

The load-bearing assumption

The gain of function paradigm had one catastrophic load-bearing assumption: that the enhanced pathogen would stay in the lab.

Everything else in the theory is sound. The identification of potential pandemic pathogens is real science. The development of countermeasures in advance is real preparedness. The collaboration with researchers who have field access to natural reservoirs is real epidemiology. The assumption that all of this could be done safely, in perpetuity, across multiple institutions and multiple countries with varying safety cultures and varying infrastructure, without a single containment failure, was the assumption that the entire edifice rested on.

It was never proven. It was assumed, because the alternative was too expensive to contemplate: if the assumption was wrong, the paradigm was wrong, and if the paradigm was wrong, the funding apparatus built around it was wrong, and if the funding apparatus was wrong, the careers and the institutions and the publications and the advisory positions and the speaking fees were all downstream of a mistake.

Fauci didn’t build this alone. Collins didn’t build this alone. Daszak was a consequence of the structure as much as a driver of it. The structure produced people like Daszak because the structure needed people like Daszak: entrepreneurs who could move money across the boundaries that pure academic researchers couldn’t cross, who could maintain relationships that government employees couldn’t maintain officially, who could absorb the institutional risk of the more aggressive research programs while the government funding apparatus maintained plausible distance.

When the question of whether the assumption had failed became live in January 2020, the answer threatened everything. Not just careers. The paradigm itself. Twenty years of institutional construction. The intellectual foundation of the argument that had grown NIAID’s budget twentyfold.

They had a call on February 1st.

---

The choice

The first three pieces in this series are about what was done and who did it and why the philosophical frame being assembled to absorb the accountability question doesn’t fit the facts.

This piece is about something harder: the people who built the machine that may have produced the catastrophe were not stupid, were not simply corrupt, and were not running a rogue program. They were running the program their paradigm required, with the funding their paradigm had generated, toward the goals their paradigm had defined.

The paradigm was wrong.

Not wrong in the sense that the theory was obviously crazy from the start. Wrong in the sense that the load-bearing assumption didn’t hold, and the people who should have stress-tested it had too much invested in it to do so honestly, and the people who did stress-test it were managed out of the relevant consensus, and when the assumption potentially failed the people running the program chose the program over the accounting.

DARPA stress-tested it. Ebright stress-tested it. Lipsitch stress-tested it. The institution had the information. It chose not to use it, because using it would have required asking whether the entire program was built on a premise that didn’t survive contact with the Manhattan Project’s actual lessons — that dangerous capabilities spread, that containment fails, and that the people most invested in building the thing are the last ones who can see that clearly.

That choice is on them. The structure that made the choice predictable is on all of us.

Daszak wrote DEFUSE because DEFUSE was what his career was for. Fauci funded the program because the program was what his paradigm required. The February call happened because the call was what you do when the thing you built potentially becomes the thing you said you were protecting against, and you have to decide whether to say so.

They decided not to say so.

Everything in the first three pieces flows from that decision. Everything in this piece is why the decision was predictable and why predictable does not mean excusable and why the paradigm being wrong does not make the people who suppressed the question any less responsible for the suppression.

The machine had a logic. They chose to run it.