Author: Javier Gamero
Date: 16:27:49 12/28/04
Go up one level in this thread
On December 28, 2004 at 12:48:48, Keith Boone wrote: >Just got the Mephisto "master" table top computer for Christmas, and while it >plays very well in it's regular levels, the "fun" modes don't seem to work at >all. From the manual, referring to "Fun 1", which should be the absolute lowest >level of play: "...the computer will not only "hang" pieces (leave pieces >unguarded), but will also actually try to lose material. Even watching the >computer throw away it's Queen would not be uncommon on this level!" > >I have played several games against this machine at the fun levels, and find it >very difficult to beat. (I'm about a 1700 CFC player on my best day - on my >worst day, I might dip into the negative numbers!) Thinking I was having a bad >day, I tried playing Chess Tiger 15.1 on my Palm Zire 71 against this lowest fun >level on the Mephisto. I'm running Chess Tiger at "Trainer level 8" which is >estimated at 2050 USCF or 1840 FIDE (See: >http://perso.wanadoo.fr/ct_chess/faq.html#trainerelo) >The Mephisto, playing very quickly, is holding it's own in game after game >against the Palm, while set at Fun level 1. I haven't kept accurate score, but >there is certainly no evidence of the Mephisto dumbing itself down, and hanging >pieces or playing weakly. > >OK, so why should I care? Just one concern really - if the Mephisto is failing >to drop back into it's fun levels, does this mean it is broken? I have double >checked the method for switching levels, and I'm sure I have done it correctly. >I tried clearing all settings using the ACL button and started up again, but >with the same result. Any ideas about this? > >For reference, here is the Mephisto model I'm talking about: >http://www.chesshouse.com/chess/B10.htm I'm interested in some Saitek machines and I'll have to check the "weak" levels to see how hard they are. Fritz's handicapped ELO levels are way too stronger than signaled, BTW. Comparisons between USCF and FIDE are so differently used/cited than it looks as a puzzling headache. I have heard 50, 100 and even 200 points difference and some linear formulas. Tiger Palm's page implies a strange linear relation between USCF and FIDE: no diff at 1000 and increasing diff as ELO goes up. It is against any other formula I have seen and general assumption that diff is greater in weaker levels. So, I'm pretty sceptical about Tiger's scales, in this regard. I wonder what scale is for real, USCF or FIDE? Also I'd like to know if training levels are independent of processor speed. I did a quick test, choosing 10 players (10th, 20th, 30th,... of USCF official list) and I found a average between 50 and 100 points (with some people more or less deviated than that, of course). Even counting sampling errors, it is extremely unlikely that diff is 200 points in this level of player (USCF > 2400). A guy in a FIDE forum did a linear regression with 64 players in US Open and he found USCF = 0.986·FIDE + 85. It translates in a diff of 71 with FIDE=1000 and 58 with FIDE=2000, and 45 with FIDE=2500. I.e. one can assume a diff between 50 and 100, possibly best 50 than 100. I hope this can throw some partial info in this "confusing" stuff.
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.