I think both of the following things are true. 1. It is, in principle, good thinking to keep multiple competing hypotheses in mind, and allow each new piece of

I think both of the following things are true. 1. It is, in principle, good thinking to keep multiple competing hypotheses in mind, and allow each new piece of information you get to adjust your level of credence in each one, rather than picking one to Believe and Disbelieving all the others. 2. It is, in practice, a common sort of bad thinking to treat a bundle of mutually incompatible theories as a single thing, and to handwavily treat any bit of information that supports any of them as supporting the whole bundle, even when what the information actually does is to support some bits and undermine others. And also 3. It is also a common sort of bad thinking to adjust one's beliefs not on the basis of looking concretely at how likely a new piece of information is if each of your multiple competing hypotheses is true, but on the basis of vague notions of what feels as if it supports what. I take the New Republic review to be mostly accusing Chan & Ridley of doing 2 and 3 in the guise of 1. I haven't read their book and can't tell how fair that accusation is. Similarly, the following are both true. 1'. It is, in principle, good thinking to adjust the hypotheses you're considering in the light of new evidence, which may promote new hypotheses or new variants of old hypotheses to consideration. 2'. It is, in practice, a common sort of bad thinking to respond to criticism of your beliefs by moving the goalposts: you say A1, someone says "but P!", you adjust to A2 which is compatible with P, someone says "but Q!", you adjust to A3 which is compatible with Q, and at that point you don't check whether it's also compatible with P, so you confuse "at each step I had a version of my beliefs that was nicely compatible with the evidence presented to me" with "there is a version of my beliefs that is simultaneously compatible with all the evidence". And also 3'. Adjusting your hypotheses in the light of new evidence can lead to problems akin to overfitting in machine learning, where you find something that fits very nicely with the specific bits of evidence you looked at but not with the other evidence that's out there, because you optimized it to fit as well as possible with those specific bits of evidence. I take the New Republic review to be accusing Chan & Ridley of doing 2' and 3' in the guise of 1'. Again, since I haven't read their book it's not clear how fair the accusation is. What I would like to see is something that lays out a few specific hypotheses (SARS-CoV-2 deliberately engineered with malicious purposes; SARS-CoV-2 deliberately engineered for study with the goal of understanding/preventing/mitigating future pandemics; SARS-CoV-2 found naturally in bats or something but brought into contact with humans via the WIV; SARS-CoV-2 found naturally in bats or something and brought into contact with humans via the Wuhan wet market; maybe a couple more) and then goes through the available evidence trying to assess what each piece of evidence does to each hypothesis, noting where better versions of the hypotheses are suggested by the evidence and then either assessing those better versions against other evidence or just presenting them as "suggested by the available evidence but now in need of further verification or refutation as new evidence comes in". It doesn't seem to me, from either the Aaronson review or the NR review, as if the Chan & Ridley book does anything much like that. (To a good approximation, no one ever does anything much like that, and it's a pity.)