rxe
Junior Member
Posts: 61
|
Post by rxe on Jun 8, 2014 3:24:29 GMT -8
Would anyone have any major objections to adding a handful of test players to Tiltyard? I have some spare cpu cycles on a dedicated machine, so that won't be too much of an issue. I guess they're pros and cons to this. Not over thought this, nor really want to debate this endlessly - but looking for ayes or naes. On the pros side: I can see it keeping the rating system a little more primed (kind of sucks when you a very good player but everyone is a little better, thus your scores plummet). It would be a little nicer to new comers when there are only 3-4 players on Tiltyard and 2 of those are sancho and greenshell! On the cons: May be frustrating for the top players who want to play each other (but that is more a scheduling issue). They might just play each other all the time. In terms of a player, I was thinking about a combo of the SampleMonteCarloGamer and SampleSearchLightGamer player, with a decent propnet. So not MCTS. alternatively could have a random player, but that is only interesting in multiplayer games (which we already do) IMHO. Also question: can a player, in theory, play more than one game at a time on Tiltyard? in other words does the game server handle it if a player responds to an 'info' message during playing a game with 'available'? Cheers, Richard
|
|
|
Post by Sam Schreiber on Jun 8, 2014 7:32:00 GMT -8
I have no objections to running sample players on Tiltyard. I don't do it myself because I don't have the spare compute resources.
I think Tiltyard will only allow a player to play in one match at a time right now; ultimately my intention is to allow players to play in multiple matches simultaneously if they report being able to do so (by continuing to report being available even while playing a match) so if you'd like to do this, we should talk about the best way to get it working.
|
|
rxe
Junior Member
Posts: 61
|
Post by rxe on Jun 8, 2014 9:31:54 GMT -8
I've put up three players then; ggtest1, ggtest2 and ggtest3. Guess if anyone speaks up against the idea I'll take them down.
Sam - let's not worry about multiple simultaneous matches then. I don't have many more open ports, so if we want to add more than 3, then we'd have to do something like that. Thanks.
|
|
|
Post by Steve Draper on Jun 8, 2014 10:01:52 GMT -8
I'm happy to see this, but it might be a good idea (non-urgently) to weight the match selection so that players of similar ratings play one another more often (not to the exclusion of rating-mismatches, but just a bias). Similarly (related point) it might be nice to weight the game selection such that 'more discriminating' games are selected more often than less discriminating ones (discriminating meaning no strong-win/draw bias)
|
|
rxe
Junior Member
Posts: 61
|
Post by rxe on Jun 8, 2014 10:43:35 GMT -8
I will just run one test player for a few days and see how it goes.
|
|