During 2023 November, Rootclaim (represented by Saar Wilf) and Peter Miller held a public debate on the origins of covid. The debate took place via video conference over 20 hours in 3 sessions, with written supplements, during which Rootclaim held that sars-cov-2 escaped from a lab where it was created artificially through gain-of-function research, and Peter contended that sars-cov-2 evolved naturally in animal hosts from which it infected humans without involvement of a lab.

The debate was co-judged by Will van Treuren and myself; we decided independently without conferring, solely on the basis of the materials presented during the debate (including references). We were to decide which of the two hypotheses we found more likely, or a 50-50 decision; the overall debate outcome and its $100k stakes was to be determined by the average of our decisions.

Debate

The debate took place over three sessions: expand the tags below for the videos and slides. Additional material (including the written questions asked by the judges and their answers) can be found scattered across several comments here.

The videos have been edited for time; there are plans to produce a condensed version but this has not been done yet.

The 2021 WHO report on covid origins (and its annex) is the most important source that will be used by both sides. The Project DEFUSE proposal is also relevant. Additional sources can be found linked in the slides.

In my opinion, the debate presentations often assumed an audience that is already familiar with details of the covid origins controversy, as well as familiarity with Bayesian analysis. Viewers may wish to refer to the background section of my written decision, in the next section. An updated discussion of Bayesian computation and hypothesis testing is on my blog.

Decision

Will and I each found that Peter was better able to support his position, and decided strongly in favor of zoonotic spillover. Our decisions approached the problem from different angles, with Will emphasizing the biology, while I focused more on mathematical modeling, as accords our background.

The primary basis for my decision was the relative epidemiological proximity of the earliest indicators of covid to a plausible animal source, rather than a potential laboratory source. While Rootclaim identified unusual characteristics of the genetics of sars-cov-2 and the potential for laboratory work to create a similar virus, I found this insufficient to overcome the proximity evidence and the prior against a laboratory origin.

Overall I found the evidence in favor of zoonotic origin to be strong though not insurmountable, whereas I do not think there exists significant evidence in favor of a lab leak.

An updated copy of my decision in written form is available as a pdf. (The original version of my decision is also available, but there is no reason to read that. Note that within the rules of the debate I was to make a decision without influence from the other judge, other people, or my own research; this original decision reflects that purpose. Any updates may reflect influences I have had from discussion with others or reading the other judge’s decision. There are several minor changes in the update.)

Will’s written decision can be found in the links in the video description for his decision video.

Responses

Rootclaim has responded to the claims made by Will and I.

Here is an interesting big-picture discussion of the debate and its format (with a slightly sensationalized title); it includes some background information and context for the debate I was unaware of.

UIUC physics professor Michael Weissman authored a Bayesian analysis of covid origins, including a brief discussion of the Rootclaim debate outcome.

While not exactly a response to the outcome, the virtual-money betting market Manifold had a prediction market for the outcome of the debate. Actually they had more like 10 markets related to the debate, but only one had any significant trading.