Split or Steal’s Real Dilemma is “Why Would You Ever Steal”

The newest free-to-play sensation taking over Steam is Split or Steal, a social game inspired by the British game show Golden Balls. The Golden Balls game is in turn based on the classic prisoner’s dilemma: two players are presented with a sum of money, and can decide privately whether they want to split it between themselves, or steal it. If both choose split, the money is split. If one picks split and the other steals, the splitter gets nothing, and the thief gets it all. If they both choose to steal, neither gets anything. 

After meeting, the players have two minutes to use text chat to sway the other player towards a decision, as well as judging the other player by choices such as their chosen avatar, what organization (read: guild) they’re a part of, and a measure of their lifetime trustworthiness in the form of their “karma.”

Those familiar with the prisoner’s dilemma from game theory are likely already aware of at least one problem with this setup. It’s one that’s shared with the game show this game is based on: the consequences don’t balance out. In the original prisoner’s dilemma, it is (naturally) a case of literal prisoners. Their choices will decide their sentences. If both stay silent about their crimes, they both get one year in prison. If both betray the other, they both get two years. And if one betrays and one stays silent, the betrayer walks free while the silent one gets three years. 

This setup complicates the players’ motives: even if you would naturally want to trust, betrayal starts to feel like necessary self-defense. In Split or Steal, when the other player picks steal, whether you have chosen split or steal yourself doesn’t change what happens to you. Thus, there is no push for self-defense. The only logical reason to push steal if you think the other person is going to steal is to be vindictive: certainly a very human response, but if you can overcome it the game offers little else to push against your goodwill. Besides, you’ll get another shot in a couple of minutes, if you want one. 

Naturally, that’s the next problem: none of these consequences stick. You can get robbed blind, but there’s nothing stopping you from immediately queuing up for another game, and trying your luck again. This isn’t a one time chance, this is a constant grind for virtual currency. You can strikeout as often as you like, and keep swinging. This makes it easy to be a saint: if there is no anxiety derived from missing an opportunity, why not just do the “right” thing? Why antagonize someone when you can gain the same thing over a barely longer time scale by being decent?

And what are you even gaining? Ingame currency. Coins which can be used to buy new avatars, or invest in upgrades to make you gain more money. In less than an hour with the game, I’d earned enough coin to buy a Paramedic avatar with a soft pink spotlight on her. Looking at the other options, it was my favorite possible combination. I had nothing left to buy. So why should I care about winning? There’s nothing to do with the winnings, not even leaderboards or somesuch to brag about your record of superior judgment. The closest thing to that is the global chat… and boy howdy, that is not what you’re looking for.

It is in that global chat that the problems regarding the design become glaringly obvious. People starting organizations called “SplitEveryTime,” where if you don’t split you are kicked from the group. On the other side, thief organizations, where you have to steal every time or you are forced out. Players discussing game strategy, which usually boils down to “split every time unless you’re up against a known thief, and maybe even then.” After all, picking split raises your karma, and if your karma is high people will trust you and be less likely to steal from you. It’s an investment in the future.

It’s a shame, because in rare moments when players are willing to suspend both dispassionate efficiency and gremlin malice, you can have truly amazing interactions. Having previously stolen from someone because they used a slur (which was NOT a great experience, but fortunately there is a large button for reporting inappropriate chat), I went into my next game with the blazing red text indicating that I had “RECENTLY STOLEN.” They asked me why, and I told them, but added, “but honestly, I could be saying anything right now. I’m going to hit split, but you should probably steal. Protect yourself. No hard feelings.”

There was a tense ten or fifteen seconds while I waited for them to reply, and when they did, they simply said “no hard feelings, huh” and locked in their choice. I nodded. A logical choice. And then… it turned out they had split. They believed me. Even when I literally told them not to.

Split or Steal was made in a programmer’s free time, as a hobby. They’re more astonished than anyone that this little pet project has blown up. Doubtless there will be more things to do with money, and further incentives to play more situationally rather than having an immovable strategy. But whether the game will find a way to keep people from being desensitized by the low stakes and constant chances is the real dilemma.