Rock Breaks Scissors: A Practical Guide to Outguessing and Outwitting Almost Everybody
William Poundstone (2014; Little, Brown and Company)
We hear a lot these days about algorithms behind the scenes of our digital lives: the minute we shop or browse for news online, Big Data draws a bead on us. Research shows that it would take a great deal of effort not to be predictable in this way; in fact, we probably couldn't achieve it if we tried. This is not just true of my taste in L.L. Bean sweaters - and what I might be willing to pay for them - but all the way down to the smallest atoms of choice, like a friendly game of Rock, Paper, Scissors over a bar tab. (Not always so friendly: did you know that there are international tournaments of RPS?)
There's actually a game one step more primitive than that, in which two players make individual choices, such as heads or tails on a penny they hold, and award the win to either matching coins, for one player, or non-matching coins for the other. Any binary will do: odds and evens, or yeses and noes. When playing against a machine, the ideal would be to play a random sequence of guesses, and avoid falling into predictable patterns.
Practically nobody can do this; our vision of randomness turns out to be guessable by the most primitive machines. Back in the fifties, a couple of the masterminds at Bell Labs built such machines; Claude Shannon's had sixteen bits of memory, fewer than you have in your fingers and toes. "A stream of brilliant and egotistical scientists, engineers, and mathematicians passed through Bell Labs in the 1950's. Shannons' outguessing machine was the sword in the stone. Anyone with pretensions to being smart could hardly leave without matching his genius against it. Few could resist the challenge, and few left without sacrificing a bit of ego." Shannon himself, naturally, had the only consistent winning record.
Poundstone's book looks into research on the prediction problems we encounter in everyday life: multiple choice tests, March Madness brackets, or tennis serves; and some that are rare but consequential: picking unguessable passwords, or offering the right price on a house. His advice won't make you unbeatable, but these are cases where moving the odds a little bit is at least pleasing, and possibly profitable.
Some of the advice amounts to: be humble. You're human, so your intuition about what randomness looks like is probably terrible. Your intuitions about football betting, in a point-spread tournament, are probably so bad that you'd win just as often betting the opposite of your preferred card, because if your contrarian Evil Twin wins, there are probably no other bets like that to split the pot with.
The stock market is the classic case of needing to outwit other humans as ill-informed as you are. Poundstone's long term investment strategy, based on a ten-year trailing average of price-to-earnings ratios, is the polar opposite of day trading. It's austere in the extreme, but he's probably right.
Even if you have no skin in such games, Rock Breaks Scissors is an invigorating book. The writing is clean and clear, and Poundstone follows the questions he raises to a reasonable depth. And his goal is salutary: "To understand the misperception of randomness is to gain power, not over fate but over ourselves. Fortuna keeps spinning her wheel, and no one gets out of the game."
Any Good Books, April 1, 2021 by email.