Illustrative epistemic mistakes people made with the J4 hand
The Robbi Jade Lew / J4 episode brought out some good analysis and much bad analysis.
Many of the analytic mistakes were not about strategic or other specific evidential details, but structural errors in reasoning under uncertainty.
Here is a partial list.
- Ignoring base rates. (I discuss this more here and here.)
- Conflating unconditional and conditional probabilities. Many people on both sides said that some piece of evidence was very unlikely given some scenario--either that there was cheating or that there was not cheating. Often, these claims were perfectly true but ignored the fact that the evidence was also very unlikely under any scenario. What mattered was the likelihood ratio of the evidence, not simply its likelihood given some hypothesis or other.
- Ignoring correlation. (See here.)
- Ignoring the context. Expected behavior is simply different at very high stakes. I am not an expert about the nuances of high-stakes culture or the politics of getting into streamed games, but far too many people extrapolated from their own poker experience or what they see of the highest-stakes games they know a lot about. This caused people to reason poorly about, among other things, (i) the probability of a misread in a big pot, (ii) the likelihood that an honest player or a dishonest one would be given a 50% freeroll on a stake in a game, (iii) the likelihood that a weak amateur could win over two sessions against an uneven lineup including very top professionals, and (iv) the motivation to cheat. I was guilty of this in not thinking about (ii) as rigorously as I should have: although I initially thought the stake was suspicious, I mostly shrugged my shoulders and thought that very strange financial arrangements often happen in poker.
- Ignoring history. This is closely related to ignoring base rates. It is irresponsible to claim that a cheater "would never" do something that is in fact what cheaters have been observed doing. As is so often the case, a bit of research is worth a lot of raw intelligence when you're trying to reason accurately.
There were also the usual emotional and political issues: getting stubborn, ignoring inconvenient evidence, disguising expressions of political affiliation as dispassionate analysis, and so on. But what is striking to me is the poor display of epistemic fundamentals even among people who seemed not to be suffering from those.