Scientific method can be roughly described as a procedure of searching in experimental results for patterns that will finally lead to a theory. This theory will be used to describe the behavior of all similar systems to the ones under experiment.
The reason that natural sciences (mostly physics, biology, chemistry and astronomy) have been considered as the cornerstone of science is very simple; we can easily conduct experiments and repeat them again as many times as we like. In simple words; we have plenty experimental data.
Until recently this was the disadvantage of the social and behavioral sciences (Political science, Psychology, Sociology ,Demography, Economics etc); it is very difficult, expensive and time-consuming to do large scale research. But it seems that this is changing lately, since behavioral scientist’s attention is being focused on the large amount of data that people’s activity in the Internet.
virtual worlds, virtual experiments
The most valuable of this data is considered to be the on-line gaming data ; with the rise of the Internet gaming there are a lot of servers out there that they have enormous amount of user behavioral data. In such a game every single move of the player is recorded and if the game simulates aspects of real life or real life itself (e.g. Second Life) then searching in the data is really interesting.
Focus on the experimental potential of virtual worlds went public with the famous incident of the World of Warcraft pandemia in 2007 (http://www.time.com/time/health/article/0,8599,1655109,00.html). Almost in the same time scientific papers were published, which focus on the issue (http://www.sciencemag.org/cgi/content/abstract/317/5837/472). Lately the same year a game appeared specifically for that purpose (http://www.digitaltrends.com/computing/virtual-world-as-social-experiment/), although it seems that the idea is a lot older (http://www.sciencemag.org/cgi/eletters/317/5837/472#10232)
Judging from the example of the game “EverQuest”, economists were the first interested on people’s on line behavior (http://www.slate.com/id/2078053), but it seems that the issue is finally getting more wider focus lately (http://arstechnica.com/science/news/2009/02/aaas-60tb-of-behavioral-data-the-everquest-2-server-logs.ars)
can we get valid results?
But the main question is, if we can trust the results of such experiments. The key point to answer that is if we can really trust people’s on-line behavior. Virtual worlds are not real despite the quality of the technology used.
The only way to find out is to validate the results against the real world. The answer of trust will be negative if the theories, that will be produced, would fail in the real world. The tricky part is that if they do not fail, we don’t have a definite answer for the success of such methods, because people learn to copy on-line behavior back to the real world. The worlds are not so separate as we think them to be!
What is more interesting, is that applying the theories -which were produced by “virtual experiments”- to the real world we create a feedback from virtual to real, which means that we will experience the results of our virtual behavior in real world! For example; users enter virtual worlds in their spare time when they are more relaxed and want to entertain themselves, so they are not afraid of changing, exploring and acquiring new roles and experiences. Maybe some experimental economist will base a theory on that behavior, then don’t be surprised that your boss believes that laying you off, will help you achieve happiness in the long run! (In such a case feel sorry for the hard realists ones!)