Errata-sleuthing for fun and knowledge

Posted on
Tags: hacks, errata, sleuthing, learning, causality

I love inspecting errata, those lists of corrections and desired changes that authors compile for their books, hoping that someday their books will be sent back to the presses and the changes can be worked in.

For many technical books, the errata are mostly boring fixes for publishing errors. Sometimes, however, you can see that the author’s thinking has evolved, and then errata-sleuthing becomes fun. The errata list becomes a kind of “brain diff” for the author’s thoughts. It’s not often that someone can tell you with such precision how their thinking has evolved on important subjects.

By studying these diffs, you can learn something about the way their creators think. Similarly, Seth Roberts wrote on his blog that he learned something about writing and editing by comparing New Yorker articles to the slightly different versions published elsewhere; the New Yorker versions were better, and the differences between the versions explained why they were better. It’s like getting free writing lessons from The New Yorker.

A recent example of errata-sleuting: The errata for Judea Pearl’s Causality are fascinating. Causality research is a rapidly advancing field, and it’s interesting to see, through edits to his two-year-old book, which new developments one of the pioneers of the field thinks are worth working into the text. For instance: the addition of the Mediation Formula to the discussion of indirect effects. There are also many new citations to recent research and a few helpful clarifications. I also like (and am slightly amused) that he’s sharpened some of the poking-sticks he’s using to gently prod reluctant researchers, clinging to traditional methods, toward the advantages of modern causality theory: “some statisticians” has become “traditional statisticians”; a potential-outcomes “fallacy” now has “damaging consequences.” Good stuff.

The next time you’re working through an interesting textbook, see if you can find errata for it, even if you have the most-recent printing. You might learn something unexpected, and for very little cost.