New book draft of "Bayesian modeling of perception and action", with @KordingLab @dangoldreich, now available at bayesianmodeling.com.

We incorporated 100s of comments from the community - thanks so much everyone!

To be published by @MITPress but e-book will remain free. New comments spreadsheet also on that website. We will incorporate comments concerning typos or factual mistakes, and other comments to the extent that our publisher allows. Remaining comments will be taken into account for the second edition of the book.

This is only going to appeal to a niche audience, but here we go.

When you fit a simulated (sampled) model using ML estimation, do you ever encounter log(0), and then replace that by log(some small number) so that your code runs?

I used to do that all the time. Bad idea!! 1/ A few years ago, Bas van Opheusden explained to me that this inevitably leads to biases, because log(true probability) can go all the way to negative infinity, whereas log(small number) cannot.

And this is just one symptom. Even if you don't run into log(0) issues,

2/

Many chess players believe that women play worse chess than men. Here is how to respond.
1/ Their argument is typically a variant of "the best male player is much higher rated than the best female player". This was also the argument in a recent article on the Indian website @livemint, "Why women lose at chess" (livemint.com/mint-lounge/fe…)
2/

Partners of physicians often live with a baseline of anxiety. Over the last week, that anxiety has exploded and become mixed with outrage.

I tell my wife not to be a hero, but if her job requires that she sees COVID patients without proper protective gear, 1/5 then heroism is part of the job description.

And then I realize that much of this was preventable. That every physician's act of heroism plugs a hole left by a system that failed them at all levels. 2/5