Viewquake: interpreting obvious mistakes
In his poker manual Ace on the River, Barry Greenstein writes something like: When you see a successful player make an obvious mistake, do not dismiss them. Rather, start trying to learn more from them. It means that they must be compensating for the mistake--a mistake you know well enough to diagnose third-personally--by doing other things particularly well.
A few reflections:
- In the poker context, this advice was a bit more relevant in an era when excellent training material was less available; scouting your opponents for implicit strategy advice was more important. This is a lot less true of poker now than it was when Barry was writing (almost 20 years ago!), but it's very true of software engineering.
- Sometimes the person in question is directly compensating for the mistake (or seeming mistake) in a nonobvious way. The canonical poker example here is playing too loose preflop and making up for it in big-money decisions later in the hand. There are many less obvious examples involving passive play in order to entice weak players into the hand. It's more interesting, I think, when there's a real, all-things-considered mistake happening, but with offsetting skill in some other subdomain.
- This advice shows up in a different way in Cowen and Gross's Talent, where they advise you to dwell on the "Groucho Marx" question of why a particular candidate is available to you. One of their conclusions is that an obvious defect somewhere allows for the possibility of extraordinary skill somewhere else (while remaining within the threshold of the talent you can actually get). There's also the story Ben Horowitz tells in The Hard Thing About Hard Things about how he was able to hire Mark Cranney because Cranney, despite being excellent at every strategic and organizational element of sales, did not "look the part" and so was passed over by other firms.
- I found this so insightful in part because poker players are so eager to find fault in their opponents. This is both because finding and exploiting faults is exactly how you become a winning player and because the dominant psychology of the game simply involves a strong desire to believe yourself better than other players. Other domains are a lot less like this, but I'm reminded of the philosophy literature on disagreement, which takes up questions like: "if I believe X and an epistemic peer believes not-X, must I change my belief in X?" So much of that discussion begins (or seems to me to begin) from a place of extreme desire to hang on to one's beliefs in the face of disagreement. Not to elide tricky questions of the structure of belief, but a lot of those papers felt like stories with desperately insecure protagonists. (Why not ask "what may I learn?" instead of "must I change?".)