Bill Gurley on COVID Origins and Failure Analysis

Every disaster leaves two kinds of damage: the collapsed building or grounded fleet, and the unanswered question of why. Bill Gurley, the venture capitalist known for backing Uber and for a widely shared 2021 talk on regulatory capture, spent his latest appearance at the All-In Summit walking through six catastrophes to make a point about the second kind of damage. Buildings can be rebuilt. Questions, if nobody asks them, just sit there.

Gurley's method comes from industry, not journalism. It is called CAPA, short for corrective and preventive action, and it is a legal requirement for failure analysis in pharmaceuticals, medical devices, and aerospace (09:24). The process has five steps: detect the failure, investigate it, confirm the root cause, correct it, and prevent it from happening again. The order matters. Skip the middle steps and the fix is a guess. As Gurley put it,

Finding the failure cause doesn't help if you don't prevent. [04:34]

He then ran the tape on five events where CAPA, in effect, played out in public. The 2021 collapse of the Champlain Towers South condominium in Surfside, Florida, killed 98 people (01:49); the cause was a mix of design flaws, deferred maintenance, and an ignored inspection report, and investigators from The New York Times, the National Institute of Standards and Technology, and the Miami Herald all worked the story until they agreed on what happened. The Herald's team grew to 37 people and won a Pulitzer. The findings became new condominium safety rules across Florida and other coastal states, which is the point of the whole exercise: a root cause that never becomes a rule was wasted effort.

The Boeing 737 MAX crashes followed the same shape. Boeing bolted larger, more fuel-efficient engines onto an old airframe to compete with Airbus, which shifted the plane's balance and required a piece of flight-control software called MCAS to compensate. Pilots were never told it existed, in part, Gurley says, to avoid the cost of retraining (05:41). Two crashes later, six separate investigators, including two congressional committees, Indonesian air-safety officials, and 57 whistleblowers, converged on the same failure. Boeing and the FAA, by contrast, spent the early weeks stonewalling document requests and blaming pilots. Senator Ted Cruz's line to the FAA administrator, asking whether he knew he didn't work for Boeing, captures the dynamic Gurley calls regulatory capture: the regulator and the regulated had become friends.

Hurricane Katrina and the Fukushima nuclear disaster get the same treatment, and both end the same way. The Army Corps of Engineers wanted the Katrina story to be an act of God, water simply overtopping the levees. Independent teams from the National Science Foundation and Louisiana State University found the levees had failed from underneath, a design and construction problem, and the Corps' own hired reviewer at the American Society of Civil Engineers eventually got absorbed into the institution it was supposed to be checking (11:25). At Fukushima, the plant's sea wall stood 5.7 meters high even though internal reports had projected tsunamis of 10 to 15 meters within a 30-year window; the report was buried (12:40). The commission that eventually said the plant's disaster was, in its own first line, "profoundly man-made," was led by a physician named Kiyoshi Kurokawa, chosen partly because he had no ties to the nuclear industry.

Gurley's favorite searcher, though, is Richard Feynman, the Caltech physicist added to the Rogers Commission investigating the 1986 Challenger disaster mostly, Gurley suspects, for the credibility of his Nobel Prize. Feynman cleared his calendar for six months, showed up at NASA facilities uninvited, and, on live television, dropped a piece of O-ring material into ice water to show the commission exactly how it lost flexibility in the cold (18:05). When the final report smoothed over his findings, he told the chairman to take his name off it. The commission caved and printed his notes in full as an appendix.

The Investigation That Never Started

Here is Gurley's turn. All of that, he says, happened for disasters that killed, at most, thousands of people. COVID-19, by his tally, caused about 15 million deaths and cost the global economy roughly $45 trillion (18:55). Add up the Surfside collapse, the 737 MAX crashes, Katrina, Fukushima, and the Challenger disaster, and the combined toll is a fraction of one percent of that, about 3,000 times smaller by his math (20:36). And yet, six years on, there has been no equivalent investigation: no newsroom assigned 30 reporters, no bipartisan congressional commission, no NIH-funded third-party inquiry, no Kurokawa or Feynman standing apart from the institutions involved (21:30, 21:54). He put it bluntly:

Forgive me if I'm unafraid of what some AI guy's chatbot is gonna do to humanity when you can't tell me what this thing already did. [19:24]

Gurley's explanation is that the people treated as the relevant experts, virologist Anthony Fauci, former NIH director Francis Collins, EcoHealth Alliance president Peter Daszak, former Wellcome Trust director Jeremy Farrar, and University of North Carolina virologist Ralph Baric, were not positioned to investigate impartially. All five, he argues, had been public advocates of gain-of-function research, the deliberate modification of a virus to make it more dangerous, before the pandemic began; Fauci wrote in a 2011 op-ed that its benefits outweighed its risks (23:49). A person who spent years arguing a practice was safe, Gurley says, cannot then be the neutral judge of whether that practice caused a catastrophe. His verdict:

What I'm here to tell you is they weren't experts, they were blockers. [22:45]

The most specific piece of evidence he raises is a document called the DEFUSE proposal, uncovered three years after COVID by an informal research collective called DRASTIC. It describes a plan submitted by EcoHealth Alliance connecting a virologist at Duke with the Wuhan Institute of Virology to insert a furin cleavage site, a genetic feature that can make a virus more infectious, into a bat coronavirus (30:08). In a survey Gurley cites, only 22 percent of scientists who believed COVID originated from an animal had even heard of the proposal (29:41). He says he ran the document past an AI system, asking it to apply CAPA reasoning with no prior context, and it flagged the proposal as top-tier investigative evidence. Documents related to the research remain the subject of unresolved requests to the University of North Carolina, which Gurley says the university and the state have been slow to produce (31:28).

What the Searchers Deserve

None of this, Gurley is careful to say, proves a lab origin. He explicitly declines to take a side, noting that China has tested 80,000 animals without finding a natural crossover event, a fact he says cuts against easy certainty in either direction (37:37). His argument isn't about the answer. It's about the absence of the process that, in every other disaster he named, eventually produced one. He credits a handful of people who kept investigating without institutional backing: molecular biologist Alina Chan, journalist Katherine Eban, reporter Emily Kopp, Princeton professor Zeynep Tufekci, and author Alison Young, all of whom, he notes, have faced public attacks for the work.

He closes with a line attributed to Plato, chosen, it seems, because it reframes the entire talk as a question of temperament rather than politics:

We can easily forgive a child who is afraid of the dark. The real tragedy in life is when men are afraid of the light. [32:54]

Gurley isn't asking anyone to accept a conclusion. He's asking why, after every other disaster on his list produced a Kurokawa or a Feynman, this one, the largest by orders of magnitude, still hasn't.