Monday, 3 February 2014

Playing one game when everyone else is playing a different one

Arthur Chu has won a few games of Jeopardy. This, alone, would not garner much attention, though it's a very good performance and nice take-home pay. Full disclosure: I was on the show once, and came third, so I ain't criticizing.

But Arthur has been winning by breaking some time-honoured Jeopardy practices in order to play smarter. Daily Doubles are the way to go, for they let you put a dollar figure on how confident you are, in particular vastly increasing the the valuation you can put on a question, and do not give anyone else a chance to answer. Consistent with this, Arthur jumps around high-paying categories, looking for Daily Doubles. Then, once, after hunting around and finding a Daily Double, he bet the minimum ($5) because it was about sports and he didn't think he'd know the answer. Then why the hell did he look for it in the sports category? Answer: to prevent someone else from getting it who knows something about sports; such is the value of the Daily Double.

And he's played to draw: in the lead heading into final, he's bet $X = the difference between his current total and twice the second place player's total, such that if they both get it right, the worst he can do is draw; he will win if the other player bets less than his or her maximum, or gets the answer wrong. Now, ordinarily, leaders bet $X + $1: same deal but you win outright if you respond correctly. Presumably Arthur's logic is this: he would still get to keep all his money if he draws rather than wins outright; he doesn't think the second-place player is very good, and wouldn't mind seeing them continue into the next round, rather than a randomly chosen player; and he runs slightly less risk of not winning at all. William Spaniel explains the logic very well here. (I have one quibble with William's analysis: if the gap with the second-place player is small enough, i.e. they've played a good enough game, Arthur should conclude that the 2nd-place player is likely better than the median replacement player, and should go for the win. But I digress.)

This is quite different from what you ordinarily see on Jeopardy. Plodders like me, who keep going down a category like mules rather than looking for Daily Doubles (I flubbed the one I got), let the Arthurs of the world run rampant. In deviating from the norm, it has bothered some viewers: it's harder to follow when someone is jumping around. As for Arthur's Final Jeopardy wagers, it just... feels wrong, by some norm of what a game should look like, not to go for the outright win. There is a sense that it's just not cricket, as they say. Here, Arthur is exploiting a rule that creates tension with the way that a typical viewer thinks a game should happen: there should be one Jeopardy champion.

There are a lot of games out there, and they have rules, but they also have norms that help define those games. These are easy to see in televised games, because there are leagues (game shows) with an interest in making it good TV, and teams (contestants) with an interest in winning. Hockey's rules worked one way in the 1980s, generating a golden age of scoring. But once the great Jacques Lemaire (among others) figured out a way to exploit them to win hockey games, those same rules meant that the game was played totally differently: with stifling defense through the neutral zone that dramatically cut down on scoring chances.

This doesn't happen all the time. It's hard to know how to win in complex games, and so you learn some shortcuts. Jeopardy fans can figure out, by watching enough, that it "makes sense" to bet $X + 1, and then do so if they get on the show, even though there's a better strategy out there. It becomes received wisdom that sometimes you need to steal bases to win baseball games, so that's what you do, even though some nerd is telling you that that's a bad bet. (Hell, sometimes it even works.) Jeopardy is a pretty straightforward game, apart from the matter of actually knowing the answers, but even then the nerds that go on don't really seem to play it well, usually... and I fully include myself in that critique.

There is also socialization that isn't even about winning. You learn one way to play a game, even if you don't adopt that way of playing as a shortcut to winning. Sean Avery worked out that you could, under the NHL's rules, be a jerkface and help your team score a goal. Nobody had ever thought to do that before.

When I went on Jeopardy, they told us a bit about the mechanics of the game. There are employees who press buttons to reveal the clues on the screens, and listen to the responses to make sure that they can adapt when it's formulated a little differently than expected or is otherwise questionable, but permissible, and signal to Trebek what to do. The show employee who dealt with us explained that these guys' job is easier when people plod through a category, because they just focus on the next answer down the line. When there's a player like Arthur, they look at each other and warn: "Jumper!" This is, of course, hilarious. More than that, I don't know that it was an intentional effort to get us not to be "jumpers"... but in any case, it made it clear that "jumpers" are weird folks to the Jeopardy personnel. So norms get perpetuated, new players learn the ropes in one way rather than another.

Leagues can change the rules to adapt to the norms. The NHL tried to deal with the Dead Puck Era by, among other things, eliminating the rule against passing the puck up two lines. (At about the same time they put other foolish rules in place, including one that makes actually winning games Pareto sub-optimal--the opposite of every football league in Europe, where winning gets you three points and a tie gets you one.) Basketball banned zone defense and put up a shot clock for the same reason. Immediately after Avery did his thing, NHL immediately banned... whatever it was he did. If Jeopardy wants to conserve the norm of one winner, they should do something like rule that you get half of your total in case of a tie.

It's all a reminder that the thing about games is this: there are really two games going on. There's what the rules say, and what we act like the game is. And Arthur Chu is winning because he doesn't care about the game the rest of us are playing.

1 comment: