Author: Matthew Hull
Date: 20:09:43 11/22/05
Go up one level in this thread
On November 21, 2005 at 18:15:13, Peter Kappler wrote: >On November 21, 2005 at 17:43:08, Robert Hyatt wrote: > >>On November 21, 2005 at 12:15:28, Uri Blass wrote: >> >>>On November 20, 2005 at 21:19:21, Robert Hyatt wrote: >>> >>>>On November 20, 2005 at 20:56:42, Sherry Windsor wrote: >>>> >>>>>Dr Hyatt now admits reluctantly that computers are now GM strength, >>>>>unfortunately he is still stubbornly stating that they are not over 2600? To me >>>>>this looks very prejudicial. I think they are easily playing at the 2750 level. >>>> >>>>Define "computers". >>>> >>>>Do you mean a big multi-cpu opteron, or a single-cpu home system? >>>> >>>>There is a huge difference... >>>> >>>>I don't admit anything "reluctantly". In 1995 when this discussion started, >>>>computers were nowhere near GM strength. >>> >>>Deep thought at that time was maybe GM strength. >>> >> >>Poor wording on my part. "personal computer programs". As in mass-market. The >>hardware was simply too slow. Deep Thought was a different kind of animal... >> >> >> >>>Other programs were weaker than it at that time but they clearly had chances >>>against GM's >> >>a 2200 player has chances. I have seen them win occasional games from GM >>players in events like the US Open. >> >> >> >>> >>>I remember that Fritz3 on P90 drew against 3 GM's in a tournament near 1995(it >>>may be 1994 and I am not sure exactly) and it got the IM norm in that >>>tournament(it had bigger problems against weaker players who prepared more >>>against it and not against the GM's) >>> >>> They are now clearly playing at that >>>>level, thanks to great advances in hardware speed from 1995. >>> >>>I think that if we talk about 120/40 time control then about half of the >>>improvement from 1995 is from software. >>> >>>If you give programs of 1995 the hardware of today then it is not clear if they >>>can win against program of today with the hardware of 1995. >> >>No, but I'll bet most programs of 1995 would do about as well against humans on >>today's hardware as current programs. >> >> >> >>> >>> But they are not >>>>quite super-GM (2700+) yet, unless you talk about very pricey hardware. Not a >>>>$500 home computer. >>> >>>We do not have evidence to say that they are not(2700+) >> >>Nor do we have any evidence that says they are, either... >> > > >Bob, this is simply not true. Do I have to back and dig up the data for all the >Argentina man vs machine events? Remember that thread? Couldn't have been more >than 2-3 months ago. If these programs played on the GM circuit with only their built-in learning as a way to keep up, their performance ratings might not hold up over time. > >Unfortunately, the CCC search engine doesn't seem to have posts from the last >few months. > >-Peter
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.