The Voting Problem

Grains of Sand

A recent twitterconversation with Tommy Rousse about “Rawlsian game design” evolved into a broader discussion of the illusion of choice, situations where I think my actions are affecting the system, but they aren’t really. Many (most?) gambling games have this problem. Game Designer Michael Brough snarkishly added that another system with an illusion of choice problem is voting. Michael’s point is actually a valid and important one, and I think it’s worth a closer look.

It’s hard to criticize voting because it has a sacred status as an essential component of democracy, that klugy system that Churchill famously described as “the worst form of government, except for all those other forms that have been tried from time to time.” I honestly do believe it would be a catastrophe if there was a total loss of faith in the sacred status of voting, however there are plenty of people maintaining that status and that’s not our job. Our job, as game designers, is to think deeply about systems, choices, actions, motivations, and outcomes, and from this perspective voting is terribly flawed.

There is a lot of fascinating thinking about the knotty formal problems of group decision making, for example Arrow’s impossibility theorem, which suggests that there is no consistent and fair way to express the preferences of a group of individuals. But what I have in mind is a problem that is far simpler and perhaps more difficult.

What I see as the essential problem of voting is that it doesn’t scale. I think voting is a great way to pick lunch places and (ahem) department chairs, but as it scales up to the size of cities, counties, states, countries (and planets?) its failure to scale becomes more and more problematic.

The essential problem is that, in a group of a million or more people my individual vote doesn’t have a meaningful impact on the outcome. I know we want it to, I know we’d like to be able to claim that it does, we want to believe that there is some magical property that our tiny grain of sand contributes to the sunny tropical beach of group preference. But it doesn’t. If you were struck by lightning on the way to the polling booth it wouldn’t affect the outcome of any election you ever voted in. Worse – if I went back in time and took every vote you ever cast in every election you voted in and switched it with a vote for the opposite candidate it wouldn’t matter one bit. The impact of your individual vote is so small compared to the scale of the overall election that it is utterly insignificant. This is basically the public choice version of the Sorites paradox.

This should trigger your game designer alarm system, because one of our important jobs is making systems where choices and actions do matter. Meaningful actions and meaningful choices are the bread and butter of good game design, so when we see a choice that doesn’t matter we should want to take a closer look.

What are the consequences of the voting problem? Well, here’s one story that I find pretty convincing: because my individual vote has essentially no impact on the outcome, both the benefits of voting wisely and the penalties of voting foolishly are essentially non-existent. As a result, there’s no incentive for me to think deeply about how I should vote. I am free to vote for other reasons, and because voting does take some effort and I want to get some value from it, I end up just voting in a way that makes me feel good about myself, that expresses my values, or that impresses other people. I end up using my vote as an instrument on which to perform my values, instead of as a tool with which to achieve my values by actually affecting the world. I can vote Republican to signal my respect for tradition, I can vote Democrat to signal my compassion for others, and I never have to worry about what the actual results of my vote will be. Instead of a process for solving group decision problems, voting becomes a theater for expressing idealized versions of our values (most of which we have inherited unconsciously from our parents and peer groups) in a way that makes us feel good about ourselves without having to think too hard. (This description of the consequences of the voting problem is taken pretty much directly from economist Bryan Caplan’s terrific book The Myth of the Rational Voter.)

Actually solving real-world problems is a challenging process involving hard research into complex systems that have lots of uncertainty and plenty of unintended consequences. Even when we all have the same preferences it’s a process of trade-offs, compromises, and negotiations. Add into the mix that we have different preferences, and it becomes a nightmare. That we manage to do it all is a miracle, and make no mistake – voting-based democracy is miraculous. But could we do it better? We should at least think about it.

I would say that most people who play gambling games realize that what they are doing is a sick pleasure and a dangerous vice, most of us understand that when we blow on the dice and beseech the gods of luck we are indulging in a pleasant but foolish form of primal superstition. But when we vote we are fooling ourselves about what we are doing and why. And compared to gambling, the stakes here are exponentially greater.

That’s why I don’t think its unreasonable to claim that voting is worse than gambling.

One way I think game design could contribute to this discussion is through more research into the design of large-scale games where the actions and choices of each player continue to matter, not just symbolically but in a direct, concrete way, on the behavior of the system. Most existing large-scale games are really just social settings for small-scale games where our actions do matter. Could there be a million-player game where each player’s actions have an important, significant impact on the outcome? Where, as a result, each player is incentivized to think deeply about the consequences of their own actions and those of other players? Some MMOs, like EVE Online and Tale in the Desert, hint at possible directions, but most MMO design doesn’t seem to care much about this question.