Author: Robert Hyatt
Date: 08:17:47 06/08/98
Go up one level in this thread
On June 08, 1998 at 09:45:18, Dan Homan wrote: >On June 08, 1998 at 07:10:37, Vincent Diepeveen wrote: > >> >>On June 05, 1998 at 20:25:15, Will Singleton wrote: >> >>> >>>>On June 05, 1998 at 11:14:24, Dan Homan wrote: >>>> >>>>I'll log EXchess onto FICS tonight. I should have an account >>>>set up on ICC soon. (I've set it up, but I am just waiting for the >>>>admin to register it as a free computer account.) I was waiting >>>>until I could afford a fast machine, but I figure that will be >>>>a long way away.... >>>> >>>> - Dan >>>> >>> >>>Yes, ICC does have slightly more action. The program is hardly ever >>>idle, plays almost continuously. Good competition too, though the >>>ratings are inflated relative to fics. >>> >>>Will >> >>Are they really inflated, considering that at icc the average >>timecontrol of >>a game is way faster? > >Definitely inflated, by at least 100 pts. My program plays on >both servers with essentially the same time controls (almost >always 5 0, some 3 5) and is about 100 pts stronger on ICC. This >is after only a weekend of play on ICC, however, and I had a bug in >my icc (hacked robofics) interface which caused me to hang and >lose many games to < 1500 strength players. I think the descrepancy >will be larger when I play more games. Several other programs that >I know seem to have +200 (or more) pts on ICC relative to FICS using >the same hardware, but I do not know about time controls. > >It is very plausable that one server may be inflated/deflated with >respect to another... for many reasons. One might be that FICS and >ICC use slightly different rating systems - the FICS system takes >into account how many games you have played recently when deciding >how much your rating should change as a result of a given game. >I guess that it really doesn't matter the reason though, because >all that is meaningful is the rating difference between players on >the same server. > > - Dan > >>Greetings, >>Vincent not to mention that FICS has a much higher "cheater" percentage based on the thousands of games I've played on both. IE *many* FICS folks are using computers and not reporting it... so that everyone is really stronger than their ratings show... and this keeps everyone at a lower upper bound. Currently, Crafty is rated at 2819 blitz and 2900 bullet on ICC, 2650 in both lightning and blitz on chess.net where Crafty plays about 90% of its games against a couple of GM's, the remainder against other computers, and on FICS (which is down right now it seems) Crafty has about the same rating as on chess.net. (I should point out that the 2819 blitz is a very low rating for Crafty of recent, and is a result of lots of changes, a few of which caused hangs and flags recently). It reached a low of 2600 a few days ago before I found and fixed the (hopefully) last SMP hang bug.) However, this overlooks one important point... relative position. which is what rating are all about, *not* the absolute value of a rating, but how two players compare. Which is why I continually remind everyone to not take the SSDF 2550+ numbers as "Elo" (FIDE) ratings, but as "numbers" only that indicate how two programs will likely do against each other, having nothing to do with how those programs will do against a different pool of players (such as FIDE GM players.) So don't get to imaginative about the ratings on various servers. They can be too big or too small. IE I know several 2400 GM players, and a couple of 2800 IM players. And a few 2900 programs. And logic tells me that is "upside down" except for the fact that many games are very fast.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.