Author: Robert Hyatt
Date: 07:18:27 12/13/99
Go up one level in this thread
On December 13, 1999 at 09:12:42, James T. Walker wrote: >On December 13, 1999 at 04:20:09, Dave Gomboc wrote: > >>On December 12, 1999 at 09:08:34, James T. Walker wrote: >> >>>On December 11, 1999 at 23:07:48, Robert Hyatt wrote: >>> >>>>On December 11, 1999 at 14:19:36, James T. Walker wrote: >>>> >>>>>On December 11, 1999 at 01:03:58, Robert Hyatt wrote: >>>>> >>>>>>On December 10, 1999 at 18:27:51, Len Eisner wrote: >>>>>> >>>>>>>On December 09, 1999 at 22:58:28, Robert Hyatt wrote: >>>>>>> >>>>>>>>On December 09, 1999 at 22:11:00, Len Eisner wrote: >>>>> >>>>><snip> >>>>> >>>>>>CRA _never_ used tournament time controls. They played game/60 time controls. >>>>>>It was a _huge_ controversy at the time, where everyone felt that the USCF did >>>>>>this to inflate the ratings a bit. This made the manufacturers happy since the >>>>>>CRA rating was always published on the outside of the packaging. I can >>>>>>guarantee you that the Mach III was _not_ a 2265 player at 40/2. I have one >>>>>>in my office. The mach IV was somewhat faster but was _not_ 2300+ at 40/2. >>>>>>They were good. But not that good. I learned to thrash my Mach III pretty >>>>>>regularly, so long as I avoided games so fast that tactics were overlooked by >>>>>>human frailty. :) >>>>>> >>>>>> >>>>>>Both were tactically not bad... but positionally they had problems, and the >>>>>>endgame was horrible compared to today's programs... No clue about outside >>>>>>passed pawns, or majorities... or king safety... Once you learned the >>>>>>Stonewall as white, you wouldn't lose against them again with white... >>>>> >>>>>Hello Bob, >>>>>I guess we will have to agree to disagree. My memory is not good these days and >>>>>when I left Japan I threw away 15 years of Chess Life magazines because of the >>>>>extra weight in my shipping allowance so I cannot prove what I remember. I >>>>>believe the Fidelity Mach 3 & Mach 4 machines were rated in a very large >>>>>tournament of 4 or 5 rounds. Fidelity provided the necessary number of machines >>>>>to make 40 games and thus get a rating (8 machines for 5 rounds?). They were >>>>>both awarded the USCF Master title for their performances of 2325 & 2265. This >>>>>was printed on the box as I remember and a Certificate came with each machine >>>>>showing it was the first micro to be awarded the Master title by the USCF. >>>>>This was tournament time controls in a real tournament I think, not a CRA >>>>>created "Test". >>>>>Jim Walker >>>> >>>>Here is what used to happen. A commercial company would enter _multiple_ >>>>machines in something like the US Open. If they entered 4, it is very likely >>>>that one would produce a TPR significantly above 'reality'. That TPR would >>>>then be prominently displayed on the box. USCF decided that this was helping >>>>commercial sales, and saw how _they_ could benefit. They stopped allowing >>>>commercial programs to enter USCF rated events and then using the rating or >>>>results for advertising. Instead, they started the CRA (Computer Rating Agency) >>>>to rate programs. For a fee. A significant fee. Then a commercial company >>>>submitted a program and USCF played it against many players to get a pretty >>>>reasonable rating. However, rather than rating at 40/2, they chose action chess >>>>(game/60). It caused a lot of complaints, but it was probably done to make the >>>>ratings _higher_ than they should have been. The manufacturers got ever- >>>>increasing ratings, USCF was raking in money hand over fist... >>>> >>>>I don't know whether the CRA exists any longer, but Crafty is a USCF member, >>>>and before it was allowed to join, I had to agree to several USCF requirements, >>>>one being absolutely no advertising based on results obtained in USCF events. >>> >>> >>>Hello Bob, >>>All of that sounds correct. The only thing is, I believe the Fidelity machines >>>(Mach3/4) were the last ones to play in the US Open before the CRA was invented. >>> I still believe they played the correct number of machines needed to get 40 >>>games each in the tournament and that's how they achieved their ratings. As I >>>remember almost all of the games were against experts/masters. I rember a write >>>up in one of the magazines, maybe Chess Life in which the author covered some of >>>the games and stated in his opinion the Mach 3 was lucky to achieve it's rating >>>of 2265 but he felt the 2325 earned by the Mach 4 was about right. :-) With a >>>2:1 speed difference it would seem that they both hit the mark pretty close.(60 >>>points difference for 2:1) >>>Jim Walker >> >>Yeah... my recollection was that the ratings for these two machines were based >>on 48 games. (I own a Mach III Master... the Designer 2265 version. It hasn't >>worked for years, though. Still, one day, I might try to fix it! :-) Anyway, >>it was quite a long time ago, and my recollection is pretty vague, so I could be >>completely off the mark. >> >>Dave > > >Hello Dave, >You are correct about the number of games. It was in fact 48 games and it is so >stated on the Certificate supplied with the computer. Here is the exact wording >on the certificate: >"Be it known to all, that the Fidelity MachIII Chess Challenger computer >obtained a certified rating of 2265 by competing in 48 tournament games against >rated players thereby achieving the classification of CHESS MASTER That doesn't mean it wasn't a CRA rating of course. That could easily have been 48 CRA games. It should easily be possible to go back to USCF and ask them about the history of this computer, to see if it actually played in several USCF events (to total up 48 games) or if it was rated by the CRA. I personally don't remember. I own one of these brown beasts. I do remember Larry Kaufman repeatedly saying it was well-overrated. And in 1992 or so told larry that we were playing it at a 60:1 time odds when testing Cray Blitz, and that we were winning all games. That would make Cray Blitz over 500 rating points better, not factoring in the fact that at 60:1 we were only using 1 cpu, that had we used 16, it would have been 60*16:1 time odds. We both agreed that 2265 was hardly realistic for the Mach III. I suspect that if you entered it in an event today, it would probably be a decent expert (2000+). It played "solid" but could be crushed with kingside attacks. And most simple endgames were beyond it totally.
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.