The Global Catastrophic Risks Institute conducted an anonymous survey of relevant experts on whether they thought COVID was more likely caused by a lab accident (aka lab leak) or zoonotic spillover. Their summary, bolding is mine:
The study’s experts overall stated that the COVID-19 pandemic most likely originated via a natural zoonotic event, defined as an event in which a non-human animal infected a human, and in which the infection did not occur in the course of any form of virological or biomedical research. The experts generally gave a lower probability for origin via a research-related accident, but most experts indicated some chance of origin via accident and about one fifth of the experts stated that an accident was the more likely origin. These beliefs were similar across experts from different geographic and academic backgrounds.
The experts mostly expressed the view that more research on COVID-19’s origin could be of value. About half of the experts stated that major gaps still remain in the understanding COVID-19’s origin, and most of the other experts also stated that some research is still needed. About 40% of experts stated that clarity on COVID-19 origins would provide a better understanding of the potential origins of future pandemics. Given clarity on COVID-19’s origin, experts also proposed a variety of governance changes for addressing future pandemics, including measures to prevent initial human infection, measures to prevent initial infection from becoming pandemic, and measures to mitigate the harm once the pandemic occurs.
The vast majority of the experts express the belief that a natural zoonotic event will likely be the origin of the next pandemic.
The experts also provided a set of clear recommendations for preventing, preparing for and responding to future pandemics, which generally align with many previous studies.
Link to the main report is here, and link to their (much longer) methodological and analytical annex is here.
[EDIT: I currently think there are enough problems with the survey that they should be mentioned alongside the results.
Firstly, the sample seems to have been based on personalized outreach, rather than mass emails. This runs the risk of selection bias. [EDIT 2: I'm told that 'personalized outreach' looks more like "sending individual emails to everyone on a big list" than "the authors emailing their friends".] Also, some participants were recruited by recommendations from other participants, which may reduce the effective sample size by over-drawing from pools of people who agree with each other. [EDIT 2: But this was mitigated to some degree by disciplinary and geographic diversity.] [EDIT 3: See this comment and replies by one of the authors of the report on the survey.]
Secondly, the survey asked people if they were familiar with a few different papers that analyse the evidence for or against a lab leak. They also included a fake paper to see how often people lied about their familiarity. A third of the sample claimed to be familiar with a fake study, more than were familiar with one relevant piece of evidence, the DEFUSE proposal, but much fewer than were familiar with most other pieces of relevant evidence.]
Surprised to read these threads without any reference to the Defuse proposal, documents of which were recently revealed to include even more detailed descriptions of features found in COVID-19 than the ones known previously.
Another thing that continues to surprise me in this discourse is not to have seen spelled out in terms of Bayes factors what to me seems the most straightforward way of thinking about the significance of the Defuse proposal as evidence, namely, that instead of thinking about the likelihoods of the features of the virus, we ask the competing hypotheses to explain the existence of the proposal itself and try to assess the Bayes factor P(Defuse|lab)/P(Defuse|natural).
Specifically, this involves estimating the probability that there exists a grant proposal, written less than 2 years before the outbreak by collaborators of the lab located in the city where the outbreak was first detected, proposing (among other things) to engineer a feature found in the natural virus:
not seen before in closely related viruses
defined by a 12 nucleotide insert when compared with closest relatives (i.e. the closest know relatives were not in any way close to evolving such a feature)
in the same location as the feature appears in the natural virus in the spike protein at the S1/S2 junction
These are not the only detailed correspondences between the supposedly natural virus and the grant proposal, as is made clear in the US right to know piece. To me, the existence of such a plan seems to be a fluke of cosmic proportions under the natural hypothesis and totally unsurprising under a lab engineering hypothesis.
The typical likelihood based argument I have seen argues (correctly in my view) that the likelihood|lab of certain features mentioned in Defuse is vastly elevated when conditioning on Defuse and that therefore P(feature|lab, Defuse) >>P(feature|natural). One thing about this form of reasoning that many people don't seem to get is that the argument for P(feature|natural) being small is not based on some general view of what evolution is capable of achieving; instead it is based on the specificity of the feature in question. With Defuse, the challenge for the natural origin hypothesis is to argue for an elevated probability of a very specific outcome under the natural origin hypothesis. This can't be done by appealing to the general awesomeness of evolution, as this form of argument only underscores the vastness of the sample space of natural viruses of which the subset of the viruses with the specific feature is a tiny subset.
What becomes clearer if instead of P(feature|lab, Defuse) we use up the Defuse evidence by thinking about P(Defuse|Hypothesis) is that the lab engineered virus hypothesis is the only hypothesis H one can think of under which P(Defuse|H) is not tiny.
The Global Catastrophic Risks Institute conducted an anonymous survey of relevant experts on whether they thought COVID was more likely caused by a lab accident (aka lab leak) or zoonotic spillover. Their summary, bolding is mine:
Link to the main report is here, and link to their (much longer) methodological and analytical annex is here.
[EDIT: I currently think there are enough problems with the survey that they should be mentioned alongside the results.
Firstly, the sample seems to have been based on personalized outreach, rather than mass emails. This runs the risk of selection bias. [EDIT 2: I'm told that 'personalized outreach' looks more like "sending individual emails to everyone on a big list" than "the authors emailing their friends".] Also, some participants were recruited by recommendations from other participants, which may reduce the effective sample size by over-drawing from pools of people who agree with each other. [EDIT 2: But this was mitigated to some degree by disciplinary and geographic diversity.] [EDIT 3: See this comment and replies by one of the authors of the report on the survey.]
Secondly, the survey asked people if they were familiar with a few different papers that analyse the evidence for or against a lab leak. They also included a fake paper to see how often people lied about their familiarity. A third of the sample claimed to be familiar with a fake study, more than were familiar with one relevant piece of evidence, the DEFUSE proposal, but much fewer than were familiar with most other pieces of relevant evidence.]