Bending THE RULES TO break
the system
From Jeopardy! wagers to artificial intelligence, our culture is obsessed with exploiting inefficiencies and ‘gaming the system.’ But in calculating how to get ahead, we haven’t stopped to consider the consequences
This coming January, professional gambler James Holzhauer, one of the most successful Jeopardy! contestants in the show’s long history, will face off against Ken Jennings and Brad Rutter, the only two people who have ever won more than him on the gameshow. The so-called Greatest of All Time tournament will be an anticipated test of wits and knowledge — but perhaps most of all, it will be a test of Jeopardy! strategy.
Earlier this year, Holzhauer made international headlines when he won 32 games in succession between April and June for an accumulated total of nearly $ 3 million in prize money. Holzhauher attracted interest as much for his success as for his intriguing, aggressive approach. From the outset, it was obvious Holzhauer was not playing Jeopardy! the conventional way: He’d select the most valuable clues first, in a bid to raise money early, and he would bet enormous amounts on Double Jeopardy and Final Jeopardy clues, maximizing rewards.
The strategy worked very well: Holzhauer is the first player in the history of the game to win more than $ 100,000 in an episode, and he did so six times. While Holzhauer was good at trivia, as all Jeopardy! victors must be, what stood out from his run was the cunning scheme. Holzhauer had figured out a way to use the rules of Jeopardy! to his advantage, working the odds and making features such as Double Jeopardy into reliable windfalls.
One would be hard-pressed to call Holzhauer’s victories unfair. He didn’t cheat; he barely exploited. And yet, in crucial ways, he wasn’t playing Jeopardy! the way the creators of Jeopardy! would prefer the game to be played, nor indeed the way that it had been played for more than 50 years. He gamed the system, in other words.
What does it mean to game the system? It’s not exactly cheating. If a cheater breaks the rules, systems are usually built to punish them. Someone gaming the system exploits the rules. They maintain the illusion of obeying the letter of the law while subtly seeking an artful advantage within its framework. To do so, one must perceive a microscopic flaw or discrepancy within a system’s seemingly impermeable mechanics and then seize upon it quickly and ruthlessly, cashing in until the people in charge get wise. And although gaming the system is rarely considered a crime, or even really unacceptable, it’s undertaken with the understanding that the upper hand secured isn’t quite right.
One who games the system follows the rules, technically. Whether it’s ethical only the system- gamer knows in their heart. It can be a foolproof shortcut to fame and fortune, an easy road to triumph and success. But in the grand scheme, it can also be ruinous — because every time one exploits the system, the system itself is at risk of being destroyed.
Even while Holzhauer’s scheme represented a major boon for the show — millions tuned in each night to see how the guy would fare, it was never particular interesting watching Holzhauer slay his would- be competitors day after day. Precisely because of his unique strategy, Holzhauer made serious Jeopardy! domination seem rather dull. And now, his innovation is an almost uniform style of play.
When he was eventually defeated, after 33 episodes, it was by someone using the same strategy against him. Everybody wants to use that trick now. Going forward, players can either follow Holzhauer’s lead and aggressively gamble their way to triumph, or they can ignore his precedent — at their own peril, if their competitors act otherwise. One way or another, Holzhauer’s approach has changed Jeopardy! forever. And that’s always the thing with gaming the system. Once a loophole has been identified, it’ll keep being exploited until it’s closed.
One morning in 2003, a man in his early 60s named Jerry Selbee noticed a sign in the window of a convenience store in his small town of Evart, Mich. advertising a new state lottery called Winfall. An amateur math wiz, Selbee scanned the rules and right away picked up on what he called a “special feature.” Unlike the ordinary way in which a lottery’s jackpot builds week after week until somebody correctly guesses every number, Winfall would pay out its grand prize whenever it reached $5 million, distributed evenly among the closest partial winners. The odds of actually winning
W infa l l were still extraordinarily l ow.
But every few weeks, Selbee realized, the odds would go up. Way up.
“Here’s what I said.
I said if I played $ 1,100 mathematically I’d have one four- number winner, that’s 1,000 bucks,” he explained to CBS in an interview earlier this year. “I divided 1100 by six instead of 57 because I did a mental quick dirty and I come up with 18. So I knew I’d have either 18 or 19 three-number winners and that’s 50 bucks each.” A thousand-dollar ticket and a whole bunch of $ 50 tickets meant Selbee stood to come out $ 800 ahead on his $ 1,100 investment, on average. “It’s actually just basic arithmetic,” he insisted later. The first time Winfall landed on its $ 5 million limit, Selbee bought $3,600 worth of tickets. He won almost double that.
By the time Selbee’s strategy was uncovered by reporters at the Boston Globe in 2011, he and his wife, Marge had won more than $ 26 million, for a profit of more than $9 million. Some math majors at MIT had discovered the Selbee loophole around the same time; they managed to score almost $4 million over a similar period. When the Globe exposé broke, the state treasurer launched an investigation. “We really looked at this, looking for corruption,” the inspector general for the state, Greg Sullivan, admitted. “We used subpoenas, we looked at documents, we interviewed dozens of people to look at this in detail with a hypothesis that something illegal had happened.”
Sullivan eventually came to the conclusion that nothing illegal had happened. Selbee’s methods, while certainly cunning, were technically above board. He hadn’t broken the the rules. He’d simply exploited them — slyly honing in on defect of the lottery and capitalizing on the mistake for his personal gain. He wasn’t really winning the lottery, in any meaningful sense. He was working an exploit. And once that exploit was discovered, the whole thing was shut down.
In the wake of the Globe article and the state’s investigation, the Massachusetts lottery discontinued the Winfall game and the rule that capped jackpot payouts at set intervals. Thanks to Selbee, no sensible lottery is likely to implement that special feature ever again.
All games by nature have rules, and the person who can exploit a weakness in those rules will have a serious advantage over those hemmed in by them. The key point is that, by gaming the system, one wins according to strategies the creator of the system never had in mind. Loopholes like the ones discovered by Jerry Selbee have serious material benefits: they made him millions, just as Holzhauer’s effort to game Jeopardy! earned him a fortune as well. But of course games aren’t the only systems capable of exploitation for material gain.
When an investor detects an imbalance in the price of an asset across different markets and quickly trades on the difference, they are expressly earning profit on an inefficiency. Arbitrage is the market’s most common form of gaming the system, and it happens every day. More rapacious attempts to “hack” the market, meanwhile, might yield shortterm gains, but sometimes have drastic consequences: Witness credit default swaps and the men and women who bet against collateralized debt obligations in the mid- 2000s. In some ways, the 2008 financial crisis was a consequence of a lot of savvy people trying to get rich by gaming the system in the worst possible ways.
But even without major financial incentive, human beings are uniquely inclined to find holes in the rules of play. Even when we all agree to play by the rules, there are inevitably those who try to manipulate them under the pretense of playing fair. You see it everywhere across professional sports: in the soccer player who falls down in agony to draw a foul or feigns an injury in order to run out the clock when his team is up a goal, in the basketball player who attempts a higher percentage of three-point shots to optimize each of their possessions, in the baseball team stacked with players who either hit home runs or strike out. It even comes up with frustrating regularity in the realm of esports: There are dozens of otherwise acclaimed fighting and action games that can’t be played on a competitive level, simply because too many players know their game-breaking technical defects and glitches.
In sports and in video games, a sense of injustice can go some way to discouraging the kind of wily rule-bypassing that might otherwise go unchecked. The groaning, whinging football star lying prostrate on the pitch, feigning a terrible injury or unbearable pain, will of course be commensurately disrespected by his opponents and viciously booed by fans — and the greater the advantage won by the deceit, the more fervent the animosity. The gamer who knows how to string together the one unbreakable chain of attacks and spams it unapologetically to win a battle, similarly, won’t soon be asked for a rematch. It may be allowed, according to the rules; it may even represent your best chance of coming out on top. But it won’t win you much in the way of goodwill.
Though perhaps a winner, the system- gamer is something of a poor sport. No one likes to feel cheated. Doubly so when the cheater isn’t technically breaking the rules. The sense of lively competition, of honest rivalry, is punctured by these hacks, and the player on the other end can’t help but feel obscurely betrayed. Nevertheless, it can be difficult to resist the impulse.
While the urge to exploit the rules of a system for gain feels uniquely human, it’s hardly exclusive to us. Computers instructed to win a game will always seek shortcuts within the acceptable parameters. A researcher at Deepmind named Victoria Krakovna assembled a list last year of the creative ways artificial intelligence ( AI) systems accomplished the tasks set for them, almost all of which involved the AI in question exploiting loopholes in the programming. AI tasked with winning a chess game might deliberately make moves complex enough to overwhelm the opponent’s computer, causing it to crash (and therefore forfeit). Another, tasked to not lose a game of Tetris, simply paused the game indefinitely, to avoid ever losing.
A computer’s conception of efficiency differs from ours in one especially important respect above all others: It has no moral centre, unless programmed to. Elon Musk has spoken of the danger inherent in programming for efficiency alone, where for instance an AI tasked strictly to increase the value of a portfolio might elect to invest in defence companies and start a corresponding war. The extreme version of gaming the system is not limited by scruples. Computers don’t care about anything but following the rules — and they remind us that following the rules in and of itself isn’t necessarily good enough.
We alone have our sense of sportsmanship and gamesmanship; we have a sense of what is right and what is wrong, outside of what is merely permitted; we have a sense of fair play, and know intuitively how to spot a kind of cheating that isn’t actually cheating.
We understand the critical difference between a completely authentic victory and one claimed through crafty or somewhat devious means, and we can hold ourselves accountable to a higher standard of conduct than what happens only to be written down. It’s not just a matter of our advanced moral sense or our strict feeling for ethics. More than that, it’s our sense of what makes games worthwhile in the first place. We know that gaming the system can spoil the fun.
In this we find an all too- typical irony. Our drive to find and exploit secret efficiencies winds up ruining the very thing at which we hope to excel. In gaming the system, we often game ourselves.