'Thinking fast and slow' - Daniel Kahneman
I'm now halfway through and not disappointed. At first I thought "I already know this stuff, the gist of it being that we're all bad at statistics". Since then I had to revise that thought: it's more like we think we know a lot, but we don't know shit. Even after we were just shown (and agreed with) that we don't know anything.
I've spoken to quite a few people about the book already, and I introduce it with my favorite part: Kahneman goes to a guy training fighter pilots and says "it's better to reward people for their actions than to punish". To which the trainer answers: "Hell no! If I praise a pilot for an outstanding maneuver, he'll do worse next time. But if I yell at someone that screwed up, he's bound to do better!" To which Kahneman starts thinking... the trainer is both right and wrong. What he sees is regression towards the mean. The pilot that did really well did better than his average, and did so by luck. The guy that flunked did worse than his average, also by luck. Next training their performances resembled their average more closely. It had nothing to do with the trainers response.
Reading the book gives you the feeling we're all stumbling through life. Some more successful than others, but more often than not this is through sheer luck.
I've always been a big believer in modelling: if you want to master a skill, copy from those that have already mastered it. Now I realize that this can only be applied to situations were there is sufficient evidence they have mastered it. Trading horse races: some of us here have traded over 100.000 races, something that can be seen as a repeatable task. Eventually the role of luck is diminished. But if you copy the actions of the founders of google, would you get an equally successful company? I doubt it.