An Experiment with 575,000 Subjects

te_fb_cover_plain_3I worked on a team from Serbia and Holland* that ran a money supply experiment in a game with 575,000 players (study news release). This was the football management game Top 11, developed by Nordeus. The experiment was about changing the initial virtual currency endowments of new players, and it ran for a month. We found that if you give people more virtual currency when they join a game, their purchases of virtual currency later go up.

This seems odd at first. If you give me carrots, won’t my demand for carrots go down, not up? Carrots are not money, however. Economists who theorize about money have pointed out that the quantity of money in circulation can increase the demand for money, through the simple mechanism of inflation. If money is dropped into an economy and results in a rapid bidding-up of prices, then it costs more money to get the same good. So people want more money.

This is what happened in the game. In T11, winning and losing depends on how good your virtual footballers are. You can buy new athletes through an auction system, where you bid your virtual currency, Tokens. Now if new players of T11 are given more Tokens, they will take those Tokens right to to the auction house and start bidding on athletes. This will quickly cause the price of those footballers to rise. And this means that anyone who wants to compete for good athletes has to get more currency. So they go to the Token store and buy.

None of this is revolutionary or even interesting from the standpoint of monetary theory and monetary policy. But it is interesting, and maybe revolutionary, to see yet more evidence that social patterns in virtual environments are the same as the ones we would see in a real environment under similar conditions. In other words, putting societies in virtual environments changes nothing about the fundamentals of social behavior.

This is the fourth major study on this topic that I’ve been involved with, and each one has come to the same conclusion. It does not appear that virtual societies act differently from real societies. There’s lots of weird behavior, yes, but it all seems to be caused by the parameters of the virtual world. It does not seem to be caused by fundamental changes in what societies do or how they operate.

Why does this matter? Two reasons.

  1. Our lives are going virtual. But social behavior in the virtual is not a completely new puzzle. Humans will be humans, online or off. The wisdom we have accumulated about social behavior still applies, even in fantasy environments.
  2. We can use virtual environments to study the fundamentals of social behavior. We can build virtual worlds as test-beds for new ideas and policies.

*My thanks go out to the good people at Nordeus (Nenad Zivic, Igor Andelkovic, and Milovan Dekic), the developers of Top 11, for including me on this experiment. Thanks also to Tolga Ozden for being a great research partner.