Hello everyone I am trying to gather feedback for a new website I am building to track the results of board games! Obviously feedback from gamers would be tremendously appreciated as I try to build something that would work great for you (us).
I would say that such a site would need to be coupled with carefully chosen statistics in order to be useful. I can already write down the results quickly with various score sheet -type applications, typically on a phone, but to start using a specific dedicated web service I would want to get more analysis out of the plays. Your site seems to have some kind of player rankings, which is in the right direction, but most ways of presenting that information are uninteresting because the usual statistics (winning percentage, total number of wins, etc) do not mean much when the statistics are over a wide range of games (two-player games have much higher expected winning rate, the role of luck varies from game to game etc). A site that manages to provide an interesting ranking over the players would be useful, but I presume doing that would actually require developing a new variant of something like TrueSkill, a ranking tool that can handle multiple games with multiple levels of luck, while smoothing out the skill levels of players across similar types of games. I guess this is beyond what you can do for your hobby project, but nevertheless I would suggest seriously thinking about how to show the rankings when most players will only have played one round of each of the games.
A friend of mine has been keeping up similar
log of his own games for a decade or so. His statistics show what kind of distribution for player and game counts one might converge to over the years; he has played more than 650 different games against more than 350 different opponents, and most gamers I personally know would be having similar statistics. I notice you said "against relatively stable group", which means people like these (or me) might not be your real target audience, but it could be worthwhile to think whether you can provide something interesting for this kind of gamers too? I can't imagine wanting to use a service that did not stretch out to cover proper statistics of everything I play over the years. His stats also show nicely why things like "winning percentage" really mean nothing as the top list of players is populated solely by players who have played exactly one game, yet there are clearly differences in the player population (the winning percentages of players with 100+ games range between 12-33%, partly because the players tend to play different kinds of games but partly because the player skills really do vary a lot -- that list has both regular drunkards but also people with continental championship titles in boardgaming).
Another practical thing is seamless integration with BGG. I anyway log everything there, and I definitely would not want to log anything twice. In practice, I would need a system where I can type in the results of arbitrary game and the play-counter in BGG for that particular game would then be automatically incremented by one (at least for me, but possibly also for the other players). Also, smooth mobile access is crucial; I never play around desktop machines.
(Now that I think of it, I probably should make the TrueSkill extension mentioned in the first paragraph. I've had the basics figured out for quite a few years already, but never bothered to actually implement and test it. If anyone is willing and capable of creating a publication for ICML/NIPS on this, I'm willing to collaborate. The model is effectively just a combination of Gaussian processes, collaborative filtering and TrueSkill, so nothing an ordinary grad student in machine learning couldn't handle...)