In the broadest sense, the issue at hand is: When is it reasonable to create a site for a sub-community, when that community's questions would be on-topic for an existing site?
In most cases, it is not reasonable to create a site for the sub-community. However, there are some rare cases where it is advantageous to do so. It doesn't make sense to have a C++ StackExchange or a Java StackExchange, but we do have a Unix & Linux StackExchange and an Ubuntu StackExchange, and they're both doing well.
We need a way to determine if a sub-community should have its own site, or be merged with the existing, broad-topic site. I believe we can determine this by looking at a few semi-subjective criteria:
- Unit of topic division
- Sub-topic overlap
- Community overlap/insularity
Unit of Topic Division
Communities where this question comes up seem to have fairly obvious units of division. They're typically easy to see in the tags. In Gaming.SE and board/card gaming, the division is between individual games. For StackOverflow, the division is probably programming languages.
Once we establish the topic division, we can look at the relationships between subtopics.
- Is it easy or difficult to separate these sub-topics?
- How likely is it that a question/answer on one sub-topic would relate to another sub-topic?
- Are there a lot of similar, broad concepts that apply to all or multiple sub-topics?
In the case of StackOverflow, programming languages are fairly easy to separate. It wouldn't be hard to determine what's on/off-topic for a Perl.SE or Ruby.SE. However, specific language questions on SO tend to relate well to other languages. Many constructs and ideas move easily between languages. In many cases, the syntax is similar, but even when it isn't, the underlying concepts are typically not unique to the language. There are many broad concepts that can be applied regardless of language. All of this points to a unified StackOverflow being a good thing, and individual programming language sites being overly fragmentary.
On the other hand, we can look at board and card gaming. Again, it's easy to separate topics by individual game. However, questions and answers pertaining to a given game are not very likely to relate usefully to other games. Poker strategy will help me very little in chess. Go Joseki will not improve my play in Settlers of Catan. Even when it comes to fairly broad topics like bluffing, strategy doesn't often transfer well between games. This does not build a strong case for having a unified Board and Card Games site instead of sub-sites for individual games.
Gaming.SE seems similar, in that sub-topics of individual games are easy to separate out. But I feel that electronic gaming is somewhat less diverse than card and board. There are some very unique games, but most fit neatly into a few genres, with a high level of overlap. There are effective tactics that can be applied to lots of first-person shooters, or real-time strategy games, or role-playing games. The execution may differ (as with programming languages), but concepts can be transfered.
Community Overlap and Insularity
Finally, we have to look at the overall community and the sub-communities that focus on a particular sub-topic.
- How much overlap is there between sub-communities?
- How does overlap differ between casual participants and experts/professionals?
- How does loyalty to the sub-group compare to loyalty to the larger group?
- Do sub-communities have the size and interest to support a site by themselves?
On StackOverflow, the overlap between subcommunities is extremely high. Most programmers have worked with multiple languages, and many use several languages on a daily basis. Experts and professionals often have experience and interest in a variety of languages, even if they claim one as their forte. And while there are small religious wars and some people who strictly refer to themselves as Perl hackers, Java masters, etc., folks mostly just call themselves programmers. There is a strong loyalty to the larger group, as compared to specific languages.
As an active participant on Gaming.SE, I think there is a similar situation over there. We may love individual games, genres, or platforms, but above all, we are gamers. Hardcore gamers (aside from super-pros) tend to play a larger variety of games than casuals. There is a lot of overlap between game cultures.
In both of these cases, there are sub-communities big enough to form their own sites, like Java or World of Warcraft, but none of the other points really indicate a need.
However, board games played at a professional level, such as chess and Go, have more insular communities. There are lots of people who take chess very seriously, but have no interest in other board or card games. I believe (but admittedly have no hard evidence) that there is even less crossover among serious and professional players. Casual players are probably more likely to dabble in chess as well as poker, Monopoly and Settlers of Catan.
There are many players of chess or Go that are only interested in that particular game. They do not consider themselves board game players first and foremost. They consider themselves chess and Go players. The sub-group often commands much higher loyalty than the broader board games group. These communities boast thousands of players, from local and school clubs, online play, professional rankings and many tournaments, with and without prize money. People who consider themselves strictly chess players are unlikely to participate in a website for all board games, where perhaps 5% (and probably a lot less) of the questions relate to their game of choice. These people are more likely to be the ones with significant expertise.
I believe these criteria highlight the most important aspects of the argument. Applying them, I think the comparison to StackOverflow is somewhat flawed. I find that Gaming.SE is probably less likely to spin off successful sites for individual games, because gamers as a whole have a strong identity, like programmers. The sub-domains of individual games also relate strongly to one another.
On the other hand, I think card and board games are much more likely to succeed as individual sites addressing a sub-domain. Chess and Go seem like the strongest candidates for success to me. They have large, loyal communities, but have little crossover with other board games, especially at higher levels of play.