Author: blass uri
Date: 02:15:29 01/09/00
Go up one level in this thread
On January 09, 2000 at 05:00:24, stuart taylor wrote: >On January 08, 2000 at 22:43:01, Robert Hyatt wrote: > >>On January 08, 2000 at 20:10:06, stuart taylor wrote: >> >>>On January 08, 2000 at 16:57:04, Robert Hyatt wrote: >>> >>>>On January 08, 2000 at 12:02:19, Graham Laight wrote: >>>> >>>>>On January 08, 2000 at 09:56:37, Robert Hyatt wrote: >>>>> >>>>>>Computers do _some_ tactics great. But I have seen GM players take them to the >>>>>>woodshed on tactics as well, as at times, the computer simply doesn't/can't go >>>>>>deep enough to see the _real_ answer, and the 'phantom answer' it sees can be >>>>>>wrong. >>>>>> >>>>>>But the problem is that the positional holes are significant enough that it is >>>>>>possible to exploit them without much risk, because many programs don't struggle >>>>>>to keep the game position in a state that favors the computer. While the GMs >>>>>>can definitely steer the game into positions that do not favor the machine, if >>>>>>it passively allows this to happen. >>>>>> >>>>>>Again, this comes from watching on ICC. If you'd like to see how bad your >>>>>>favorite program can play there, try this: log on, and accept _every_ match >>>>>>request from humans rated (say) 2500 and up. Even if they want to play 50 games >>>>>>in a row. And watch what happens after a while when they find a weakness they >>>>>>can pick on repeatedly. The weakness can be anything, from a feature that is >>>>>>not evaluated, to a book hole. And the humans _talk_. So when one finds a >>>>>>weakness, you can expect it to be hit on by several players... >>>>>> >>>>>>And you are correct that 2700 players will occasionally make a mistake that >>>>>>a computer wouldn't... but on balance, the computer has a lot more holes in >>>>>>its armor than that 2700 player. And when you move the time control up to >>>>>>40/2hrs, the tactical errors by the GMs go way down, while the positional >>>>>>errors by the computer are unchanged... >>>>> >>>>>This is good, convincing argument, but the computers are probably getting close >>>>>to being able to fight back. >>>>> >>>>>Firstly, look at DB v GK. Granted, GK didn't have enough time to discover all >>>>>the weaknesses, and the program was probably changed between games, but look at >>>>>the way the program survived in positions when all the GMs would have bet their >>>>>houses on it being beaten! >>>>> >>>>>And secondly, ongoing improvements in search selection and evaluation (+ faster >>>>>computers with more processors) will, in some marginal positions, result in >>>>>computers choosing a move from which it can survive, rather than one from which >>>>>it must die. How soon this happens depends on how close to being GMs the >>>>>computers are (if they're not already). >>>>> >>>>>-g >>>> >>>> >>>>I would never argue with any of the above. Faster hardware helps. It doesn't >>>>make a program invincible, but it _always_ helps. The question is, when will >>>>the hardware be fast enough to break the 2500 barrier? And then the 2600 >>>>barrier? It will happen. But maybe not real soon... >>> >>> >>>I would like to know what Robert Hyatt means by 2500 barrier. Surely the >>>results of top software is already well over that stage? Maybe the >>>knowledge level isn't. But there are other strengths which assure that >>>computers of today are always well over 2500 if not 2600. >>>Thank you! >>>S.Taylor >> >> >>What evidence are you talking about? The Rebel challenge match has Rebel well >>under 2500 at present.. 2466 or so last time I saw Enrique's numbers. 2466 is >>not "well over 2500" in my math book. :) > > >Obviously I'm speaking about ssdf and selective search types of rating. >I didn't realize they were THAT far behind other correct testings. And on >the likes of k6-2 450mhz. S.Taylor You cannot trust these rating because the rating system is wrong. The assumption about the expected result of higher rated players against lower rated players are wrong because this assumption does not consider the fact that part of the players are more stable and win more lower rated players(difference 200 elo) and lose more against higer rated players(difference 200 elo). The only right thing is that usually higer rated players get more than 50% against lower rated players but if the higher rated players do not play a lot of games against humans with similiar rating then you cannot trust the rating even for this. Uri
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.