Reader KW pointed me to this rant essay from Ruby on Rails enfant terrible Zed Shaw on what computer programmers don’t know about statistical analysis, but should. (Spoiler alert: a lot, apparently.) Perhaps surprisingly, building complex software systems often involves a lot of simulation, experimentation, and measurement for which statistical methods would be an asset. But according to Shaw, many programmers often have no idea how many iterations to run a test for, or why an average is often meaningless if you don’t also consider the variation, or how confounding factors can mess up an experiment. There’s actually some good statistical advice here, illustrated with examples from R.