Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fruit 2 and endgame play

Author: Dann Corbit

Date: 19:37:48 01/13/05

Go up one level in this thread


On January 12, 2005 at 21:29:45, Uri Blass wrote:

>On January 12, 2005 at 21:10:25, Dann Corbit wrote:
>
>>On January 12, 2005 at 20:30:28, Uri Blass wrote:
>>
>>>On January 12, 2005 at 20:23:57, Dann Corbit wrote:
>>>
>>>>On January 12, 2005 at 19:40:26, chandler yergin wrote:
>>>>
>>>>>On January 12, 2005 at 18:55:04, Dann Corbit wrote:
>>>>>
>>>>>>On January 12, 2005 at 18:49:06, chandler yergin wrote:
>>>>>>
>>>>>>>On January 12, 2005 at 18:45:47, Dann Corbit wrote:
>>>>>>>
>>>>>>>>On January 12, 2005 at 17:49:17, chandler yergin wrote:
>>>>>>>>
>>>>>>>>>On January 12, 2005 at 17:39:55, Uri Blass wrote:
>>>>>>>>>
>>>>>>>>>>On January 12, 2005 at 13:31:16, chandler yergin wrote:
>>>>>>>>>>
>>>>>>>>>>>On January 12, 2005 at 10:54:26, Anthony Cozzie wrote:
>>>>>>>>>>>
>>>>>>>>>>>>On January 12, 2005 at 02:33:38, Jouni Uski wrote:
>>>>>>>>>>>>
>>>>>>>>>>>>>In my (private) endgame testsuite Fruit scored better than some programs
>>>>>>>>>>>>>with tablebase support (e.g. Junior8 and Crafty).
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>>Quite stunning - it seems,
>>>>>>>>>>>>>that excellent search depth compensates TBs!
>>>>>>>>>>>
>>>>>>>>>>> Your opinion.. Provide evidence!
>>>>>>>>>>>
>>>>>>>>>>> And my suite has some 5/6 piece
>>>>>>>>>>>>>positions were TB access is definitely advantage.
>>>>>>>>>>>>>
>>>>>>>>>>>>>Jouni
>>>>>>>>>>>>
>>>>>>>>>>>>IMO the 5-piece tablebases are just not that interesting and really not worth
>>>>>>>>>>>>that much in terms of elo.
>>>>>>>>>>>
>>>>>>>>>>>What are the Current ELO Ratings for Top Programs, including yours?
>>>>>>>>>>>
>>>>>>>>>>>THey represents exact play, and all positions possible are immediatly shown.
>>>>>>>>>>>What more can you expect?
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>> A little endgame knowledge can cover most of the
>>>>>>>>>>>>positions and be a lot faster too.
>>>>>>>>>>>
>>>>>>>>>>>Absolute NONSENSE!
>>>>>>>>>>
>>>>>>>>>>Not nonsense.
>>>>>>>>>>
>>>>>>>>>>In most of the position of 5 pieces or less than it computers can find the right
>>>>>>>>>>move with no tablebases.
>>>>>>>>>>
>>>>>>>>>>Uri
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>Nonsense Uri!
>>>>>>>>>They "May" find it... Ha Ha.. in how long!
>>>>>>>>>Stop the Crap!
>>>>>>>>
>>>>>>>>Suppose that a tablebase takes 2 days to create today.
>>>>>>>>Next year it will take one day to create it.
>>>>>>>>The following year, it will take 12 hours.
>>>>>>>>Year 3: 6 hours
>>>>>>>>Year 4: 3 hours
>>>>>>>>Year 5: 90 minutes
>>>>>>>>Year 6: 45 minutes
>>>>>>>>Year 7: 22.5 minutes
>>>>>>>>Year 8: 11.25 minutes
>>>>>>>>Year 9: 5.625 minutes
>>>>>>>>Year 10: 2.8125 minutes
>>>>>>>>Year 11: 1.40625 minutes
>>>>>>>>Year 12: 42.1875 seconds
>>>>>>>>Year 13: 21.09375 seconds
>>>>>>>>Year 14: 10.546875 seconds
>>>>>>>>Year 15: 5.2734375 seconds
>>>>>>>>Year 16: 2.63671875 seconds
>>>>>>>>Year 17: 1.318359375 seconds
>>>>>>>>Year 18: 0.6591796875 of a second
>>>>>>>>Year 19: 0.32958984375 of a second
>>>>>>>>Year 20: 0.164794921875 of a second.
>>>>>>>>
>>>>>>>>What that means is that perfect information will be generated in a fraction of a
>>>>>>>>second, if that is what is desired.
>>>>>>>>
>>>>>>>>This is a conservative estimate, since compute power seems to be growing
>>>>>>>>superexponentially, rather than just exponentially.
>>>>>>>
>>>>>>>Key Word:
>>>>>>>"Seems"
>>>>>>>
>>>>>>>There is a Practical Limit!
>>>>>>>There is not enough time even to the end of the World to "Brute Force"
>>>>>>>all possible combinations... Until you do, there is NO Proof of anything!
>>>>>>>Now if you can't understand that.. I'm sorry!
>>>>>>
>>>>>>The computer does not have to calculate chess perfectly to the end.  It only has
>>>>>>to outcalculate its opponent.
>>>>>
>>>>>Reminds me of the story wehn Einstein & Nils Bohr
>>>>>were hiking in the woods...
>>>>>They came accross a huge grizzly bear..
>>>>>Nils bent down and started going through his backpack, pulled out , and started
>>>>>putting on his Tennis shoes..
>>>>>"VAT are you doing Nils?".. said Einstein,
>>>>>"You can't outrun that Bear!"
>>>>>
>>>>>Nils replied.. "I only have to outrun YOU, Dr. Einstein!"
>>>>>
>>>>>
>>>>>  Very soon, humans will have no chance against
>>>>>>computers.
>>>>>
>>>>>The Opening Tree.. the Opening Book is a DataBase of Positions PLAYED!
>>>>>
>>>>>Humans determine Theory! Programs "Store & Retrieve" information..
>>>>>
>>>>>In a  Game of Human vs Computers...
>>>>>Humans are playing what Top GM's have played before!
>>>>>Computers have NO Intelligence! You know that!
>>>>>ONLY when by "Advanced Chess...Humans USING Computers to advance Theory
>>>>>will the ELO Rating of Humans & Computers advance..
>>>>>Humans.. have to PLay the games FIRST!
>>>>>NOW.. The Formula for ELO Ratings is such that for GM'S, they Win or Lose
>>>>>very FEW POINTS for a loss among their Peers!
>>>>>Therefore.. ELO Ratings will NOT JUMP! For either Humans OR Computers.
>>>>
>>>>Intelligence is not needed to play the game of chess.  Only compute power and
>>>>storage.  It is just like any other game with a deterministic tree.
>>>>
>>>>It will be a few years longer until GO is solved, but in 20 years, the best GO
>>>>players will also be slaughtered by a computer.
>>>
>>>We do not know it.
>>
>>True.  It is the most likely outcome, however.
>>
>>>We even do not know that chess will be solved and it is only a possibility
>>>
>>>I also think that the difference between solving chess and solving go is big and
>>>solving chess is clearly easier problem and machine that is 1000000 times faster
>>>and have 1000000 time more memory than the machine that is needed to solve chess
>>>will be unable to solve go.
>>
>>Then, in log2(1000000) years, we will solve the other after the first.
>>Less than 20 years is not such a long time to wait if you are a curious fellow.
>
>I said that log2(100000) years will not be enough if you assume doubling every
>year.
>
>Number of positions in go is probably more than 2^361(simple upper bound is
>3^361)
>
>suppose for the discussion that it is 2^361
>
>number of positions in chess is probably less than 2^160
>
>2^361/2^160=2^201 so you may need 201 years if you assume doubling every year.
>Of course it is possible that the improvement will be faster than doubling every
>year but at least it is not clear that go will be solved few years after
>chess and I think that it is not clear that chess will be solved.
>
>I can imagine upper bound for technology and maybe more than 1000000 times
>faster than today is simply impossible practically.

Not long ago, this message would have seemed impossible.  The internet had not
been invented yet.

I think it a mistake to pronounce anything as impossible.  Improbable is the
best we can do.



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.