The expected value isn't flawed, but it's often not an useful guide to what action you should take. Imagine a lottery with an accumulated jackpot of $11 million, and there's one million $10 tickets (this isn't all that rare). Should you take all your savings and buy 1000 tickets? You have a positive expectation value, and 99.9% chance of ending up penniless. Is one that buys the tickets really a much more rational person than one who does it when the jackpot is $8 million? One can make just positive expectation bets and be certain to lose it all. The Kelly Criterion is a nice little formula for calculating the optimal bet size from the risk and the expectation.
"The expected value isn't flawed, but it's often not an useful guide to what action you should take."
Yeah, I agree with that - if good decisions could be made that easily we'd all be phenomenally successful. There are mountains of complications (both qualitative and quantitative) that must be heaped on top of expected values if they're going to be any use at all in deciding what action to take. Perhaps this was the author's point, but perhaps that's sort of obvious anyway.
