Sunday 20 September 2009

Raph Koster talks on Discrete Maths in Games

It was a very pleasant surprise to get to know that Raph Koster has held a talk on mathematics behind games at GDC Austin. I've long have had great respect for Raph's work and thoughts, but I wouldn't have expected someone with a background in creative writing to be going that deep in graph theory. I guess I underestimated him.

The reduction that he makes to show that behind the sensitively appealing hull, games are actually formal systems, is a point he has already made. It is also coherent with my understanding (thus the name of this blog).

I am not sure if Raph went that much further, but since he mentions NP-completeness, I'd add that NP-complete problems are actually interesting to the human brain, because they cannot be "gamed" (in Raph's terminology, as in his theory of fun).

One other interesting topic is what kind of mathematics is actually appealing to people. I guess there is no single answer to that, but definitely discrete sets, with preferably manageable sizes are something that the human mind can conveniently handle. To me this is an endless topic, but now I'd just like to refer to a cultural difference I've noticed during my education. While in Germany at university you will be taught how to reduce new problems to ones that have been already solved (optimization), in Bulgaria (and possibly in other post-communist countries in East Europe) people actually explore creative ways to tackle each particular problem (combinatorics). One example of this could be given with 3D geometry. Whereas in Germany e.g. volume would be calculated with multiple integration, in Bulgaria a subject called stereometry exists.

3 comments:

  1. The talk was all about ten specific NP-complete problems and how they are used in game design. :)

    ReplyDelete
  2. Thanks for the clarification!

    Just wanted to add that Raph has also published the slides: http://www.raphkoster.com/2009/09/22/gdca-games-are-math-slides-posted/

    ReplyDelete
  3. You could actually test the NP-complete is interesting hypothesis here: http://www.planarity.net/game.php
    Thoughts about the experience are welcome...

    ReplyDelete