Slashdot Mirror
Archiving the History of Virtual Worlds
eldavojohn writes "Some members of the University of Texas are trying to create a repository to store the histories of online virtual worlds. They hope that game makers will take advantage of this repository as they define standards of how to save interactions not only between players and the virtual worlds but also other players. How many times have I destroyed you in a duel? Let's check the records!"
It looks like they are discussing recording the primary in-game events (they list the WoW plague outbreak and the death of Morpheus). This makes it sound like they really just want a nifty little wiki dedicated to each game. When they start talking about interactions between players, significance starts becoming very important. Are we talking statistics? Chat logs?
With real world history, we have the benefit of a (somewhat) objective viewpoint from which to determine how much the world has truly been impacted. With these games, and I say this carefully, who cares?
The statistics are important - how many people stopped playing after Morpheus died or the outbreak made them think the game was unfair. But do they represent history in a virtual world, where death is mutable and guilds form and die in weeks instead of years?
Take it to the limit, everybody to the limit, come on, everybody fhqwhgads.
The ability to store 'demo' captures has been in Unreal since UT1 in the late 90s. I discovered this when I ran a server for deathmatch games and decided to keep a log of every game played. The format is a timestamped delta encoding of every movement, entity and actor in the level. Of course it occupies quite a lot of file space.
Many years later I stumbled upon a few gigabytes of this data on a backup disk. Watching through old games from different player viewpoints was very entertaining. Then it occured to me that certain players were behaving strangely. In my spare time I started thinking about how to analyse this data and eventually discovered it was possible to tell with a *high level of certainty* that some players were cheating all along.
Now I don't play online computer games much any more, but I still see the same problems and chuckle at games like WoW trying to detect and defeat cheats with client solutions. This will never work. But, given enough server side data you can easily see that some players are way outside the statistical norms for certain actions. Distributions show a typical curve from the worst to the best players, and then a separate, clearly identifiable peak of weird data bound to about 5% of players. My reckoning is that these are the cheats.
For example, one of the oldest cheats is a wall hack that allows you to see other players that should not be visible. What is the chance of a player being able to regularly track another within a few degrees of their location without this knowledge? When you run the sums on enough old games the cheats stick out a mile.