3 ms·
I'd like to understand the strengths and weaknesses of the "Bayesian analysis" being used in this article. I'm referring to the long chart of probabilities tha
by nkurz 3y ago
I'd like to understand the strengths and weaknesses of the "Bayesian analysis" being used in this article. I'm referring to the long chart of probabilities that are multiplied together to give the final odds. The article mentions the problems that occur if you fail to account for correlation between the factors, and there are obvious problems about leaving things out, but I suspect there might be other subtleties as well. Is there a good link that covers this on a beginner's theoretical level? Or suggestions of the right search terms to use?
- lamontcg 3y agoThe biggest problem is Garbage-In-Garbage-Out and your final analysis is only as good as your guesstimates for the initial probabilities. So initially the presence of the FCS in the SARS-CoV-2 genome was a "smoking gun" for the lab leak hypothesis way back in 2020 when the sequence was first released for a few days and people were just trying to digest the whole new reality. Now we've found FCS segments and near-FCS segments in other beta-coronaviruses and the FCS that SARS-CoV-2 has is a strange novel sequence that no human would have tried to put there. That has gone from something like 90% evidence for the lab leak to 99.9% evidence against it in the past 4 years.