This is not a thing that needs to be left to theorycraft! If there are, in fact, natural styles people tend to fall into, this could be discovered by a statistical survey analyzed to find natural clusters in the answers. Rather than try to guess, it is entirely possible to ask people and find out what the natural groupings are!
Indeed, WotC did this back in 1999. To quote Ryan Dancey, "We did research like this at Wizards in 99/Y2K and what we found were very clear segments ...
Who wants to find a few statisticians and run a kickstarter?
So, I have done quite a bit of clustering for a variety of data in my profession (statistics and machine learning) and I disbelieve Ryan Dancey. I looked at as much of the base data as he presented and honestly, I could see little that said there were “very clear segments”.
In 20+ years of examining data, I have
never found “very clear“ clusters in any actual people-based data. In fact, recently I was in charge of team looking to compare clustering algorithms and we spent 3 person-months collecting data from any source we could find for which there were good clusters, so we could compare algorithms. The only ones that really had strong clustering were physical systems. No people data.
So when I read threads like this, where people suggest there are “natural clusters” or “six clear genres” I am very, very skeptical. In general, here I is what I observe in most clustering analyses:
- There’s usually one big, or maybe two really big clusters. The rest are mostly small. Very commonly the biggest group is an undifferentiated mass of people who don’t have any strong factors.
- The strongest clustering factor is usually a measure of size/strength/commitment/usage/intensity
- If the optimal number of clusters is N, then N-1 and N+1 will be only slightly less good and you could easily use them (unless N=1 which is not uncommon)
- Either one or two factors almost completely determine the clustering, or it’s very hard to characterize the resulting clusters. Usually the former.
It doesn’t seem like gaming is a very unusual activity, so if I were to guess at the results of a clustering activity, I’d guess it would result in something like:
- The biggest cluster is people who are happy to play anything and don’t really mind what style it is
- The most important differentiator is how often people play
- You could merge the “OSR cluster” and the “gamers age 50+” groups, or the “Nordic LARP” cluster and the “Story Now” cluster (or whatever you came up with) and it wouldn’t make much difference.
- No-one would agree on obvious names for the clusters unless there are a couple of obvious big factors, in which case we’d likely end up with “new D&D players”, “old D&D players”, “new non-D&D players”, “old non-D&D players” or something similar as our clusters.
So, for me, I would not attempt any form of clustering and instead simply look to identify a few basic factors that differentiate. I would not expect any clear-cut divisions, but instead a continuum in which imposing cut-offs to define groups is pretty arbitrary. Fundamentally, I think that trying to define genres is doomed to failure and rather than saying “Jill is a Nordic LARP gamer” it makes more sense to say “Jill strongly likes playing anything, substantially prefers games that stress character development, and mildly dislikes systems with long rule books”.