Days ago on Reddit, a user suggested testing the viability of anarchocapitalist theory with a microcosmic video game society and economy. The prospect of being able to test political, social and economic theories without risking devastation of quality of human life is an attractive one. The ability to test human interaction the same way we test chemical reactions would eliminate arguments based on conjecture and would settle centuries-old disagreements, but has technology yet enabled us to do this? Immediately I was reminded of the Chicago v. Austrian School “Debate” between David Friedman and Robert Murphy at last year’s PorcFest, where Friedman asserted that we do indeed now have the technology to conduct at least some of these sorts of experiments in the social sciences. Brief but unambiguous, he said,
On reproducible experiments: It’s not clear you don’t have any of them, because there are natural experiments. But in fact we now have the technology with which we can do real, controlled, scientific experiments in the social sciences. If you don’t believe me, take a look at World of Warcraft. World of Warcraft is a whole bunch of more-or-less-identical worlds—each server, and in principle it would be possible for somebody playing the game to put a character on each of a hundred servers, on fifty of them do X and on fifty of them do Y, and then do statistics on those results in those little worlds. So, virtual reality is actually beginning to give us the ability to do something rather like the controlled experiment.1
I do not doubt for a moment that it would be possible to set up such an experiment through which you could accurately measure how players, on the aggregate, behave in-game and what effects these behaviors have. This sort of experiment would be great to help us figure out which sets of quests players should complete to reach the maximum level in the least amount of time, or which profession can earn gold more quickly. But could experiments conducted in Azeroth give us any reliable insights about human society on planet Earth? That is, do tests in the video-game world have any external validity?
In order for findings from studying in-game behavior to be applicable to the real world, behavior of players in video games would need to be nearly identical to the behavior of humans in their daily lives. There are many reasons why, however, in-game behavior can be expected to be different from that in real life. Firstly, the consequences in video games are not nearly as dire. If you die, you can always respawn or restart. If you ruin your reputation, you can always create another personality under another pseudonym. If you lose all your money, you can always find a way to earn it back without having to worry about having to go hungry or losing your house. Secondly, participation in video games is entirely voluntary. If someone dislikes the structure or governance of a game, they can simply stop playing. This is untrue of State, society and economy.
As a result, we see in-game behavior very different from that which takes place in the real world, even from the same actors. When I am playing World of Warcraft, I am not at all the parsimonious, scrupulous person I am in my real-world activities. I have spent half my savings on a rideable yak that I knew I certainly did not need and probably would never use. I have started petty arguments with group members with whom I knew I would have to interact regularly. I have gotten my character killed by intentionally attracting a big boss before the rest of the group was prepared, just for the sake of laughter. These are the types of behaviors—ones we would be shocked to hear about in the real world, but come to expect in virtual worlds—that would be measured by studies of in-game economies and societies.
In World of Warcraft, I am wont to attack any members of the opposing faction I see, even if my character is of a much lower level and has little chance of surviving the altercation. But would you accept from studying this fact that I would also seize an opportunity to attack a back-alley passer-by wearing an Obama/Romney 2012 or Oakland Raiders t-shirt? And is it peculiar to anyone that every ten-year-old with whom I have ever played Call of Duty has reportedly had sexual relations with my mother, yet I have received no such reports from any ten-year-olds in face-to-face interaction?
While it may be possible to construct certain simple in-game experiments where the similar circumstance would produce similar outcomes in the real world (for example, if all the players of a certain game had their money doubled overnight, we could expect inflation of prices in the trade between the players), a great number, if not the vast majority, of these experiments would be compromised by behavior that results from in-game incentives never being as stringent and severe as those in people’s real lives. Though hardcore gamers can be found in every locality, no one approaches feeding their Neopets in the same way they do feeding their real, biological body.
The attractiveness of using technological simulations to test causes and effects of social phenomena is in answering complex questions, but because video games do not produce comprehensive virtual lives that mirror the incentives of human life in the physical world, the predictive power of experiments in the video-game world is likely not very strong. At least with the current state of things, insights gained from the real world seem more useful in explaining certain video-game phenomena than vice versa. If experimentation in video games can only offer insights we already know from studying the real world, then such experimentation is of no use. Empiricism in the social sciences could, imaginably, have its day as technology advances even further, but that day is not yet here.
Mises Christ! 