Author: Majd Al-Ansari
Date: 04:44:08 10/09/05
Go up one level in this thread
On October 08, 2005 at 14:26:42, chandler yergin wrote: >On October 07, 2005 at 11:52:04, Roger D Davis wrote: > >>On October 06, 2005 at 19:51:47, Terry McCracken wrote: >> >>>On October 06, 2005 at 19:35:23, Roger D Davis wrote: >>> >>>>On October 06, 2005 at 11:53:05, Terry McCracken wrote: >>>> >>>>>He may be dismissing computer chess too lightly, but I've watched players on ICC >>>>>who were NOT GMs and obtain winning positions against these "silicon brutes", >>>>>and often they're losses are on time. The games were 15/0 or small time >>>>>increments. These favour machines, still I've seen them burn but escape due to >>>>>the bell. >>>>> >>>>>There are people here who do in fact beat programs, and we know this to be >>>>>true. >>>>> >>>>>Anand never took the matches between Kramnik and Fritz too seriously or Kasparov >>>>>matches with Deep Junior or Deep Fritz. If you really look at those games, you >>>>>can see both Kramnik and Kasparov dominating these beasts, but for what ever >>>>>reason they messed up in even and also won positions, more than once. >>>>>So those matches don't mean as much as you think. Sure the machines were strong, >>>>>but in no way better than either of these grandmasters. >>>>> >>>>>Hydra is the only _real_ exception, and even here, GM Nichols with a computer >>>>>was beating it at corr. GM Topolov had it beat and let it slip to a draw. >>>>>There's a stronger ver. now but I suspect a top GM on a good day who plays >>>>>computers often, could win a game, even a match, but I suspect after GM Adams >>>>>poor performance we might not see such an event. What a shame. >>>>> >>>>>Machines are NOT completely dominating the top humans or very experienced >>>>>computer players, at least not yet. Say what you will, but the losses are often >>>>>due to oversights that make the machines look better than the actually are. >>>>> >>>>>That's my two cents. >>>>> >>>>>Terry >>>> >>>>Maybe humans should be allowed to take back an oversight when they play >>>>computers. At least 1 oversight per game. More than that and they deserve to >>>>lose. >>>> >>>>Roger >>> >>>You are joking yes? Well that would really favour the grandmasters;-) >>>I also think their egos wouldn't let them accept these terms. >>> >>>Terry >> >>Well...people are saying that computers just can't play at a GM level because >>the GM loses interest and messes up the game. So give them 1 take back. They'll >>still lose. Not all losses are due to dramatic errors like overlooking a >>combination. > >I think you missed Terry's Point: >"Sure the machines were strong, >but in no way better than either of these grandmasters." > >"Hydra is the only _real_ exception, and even here, GM Nichols with a computer >was beating it at corr." >In fact he won both games. >"GM Topolov had it beat and let it slip to a draw." >Remember, Opening Books are based on games played.. played by Grandmasters, >the Top 99.9 % in the world. So, in reality, in the Opening, they are playing >against themselves. A small mistake in the Opening by humans, therefore has >serious effects in the middle and endgame. > >> >>Then it becomes much harder to argue that computers aren't GM strength. If you >>bend over backwards to accommodate human players and they still lose, well... >> >>Roger > Ratings are an indication of 'performance' not true strength. >Look at Topolov's performance Rating 3158 ! >Show me a Computer that comes close! >Hmmm? Anyone? Well, Hydra's methodical destruction of Michael Adams must give it a rating that is just as impressive as Topolov's performance. I am afraid that even the mighty Topolov will have no chance against Hydra. With regards to the correspondence games that Hydra lost. You have to realize that Hydra was not "on" all the time between moves. On the other hand you can be assured that the world champion correspondence player would have several extremely powerful PC's working overtime, and he most likely would have forced some lines on some computers for analysis. A professional correspondent player such as Nichols would have forced the most promising lines using several of his favourite engines. He would probably choose the results of different engines analysis for different positions. That kind of effort would be the equivalent of a super computer far stronger than Hydra in calculating terms.
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.