Have you ever lost confidence in LW after a similar experience? Maybe something where it seemed to you people were "talking a big game" but failing to apply any of that to actually win in real life?

As a stats / machine learning person, a lot of the "Bayesian statistics" talk around here is pretty cringe-inducing. My impresion is that physicists probably feel similarly about "many-worlds" discussions. I think LessWrong unfortunately causes people to believe that being a dilettante is enough to merit a confident opinion on a subject.