Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Poll: Result of Benjamin Draw odds match?
Poll Result of Benjamin match? (Closed)
Joel wins 8-0 2 4%
Joel wins 7-1 1 2%
Joel wins 6-2 1 2%
Joel wins 5-3 7 15%
Tie at 4-4 5 11%
Rybka wins 5-3 12 26%
Rybka wins 6-2 14 30%
Rybka wins 7-1 2 4%
Rybka wins 8-0 2 4%
- - By turbojuice1122 (Gold) Date 2007-12-19 03:45
Time for one of my increasingly rare polls, but this should be fun and focus discussion: what do you think will be the result of the match in January 2008 between Rybka and Joel Benjamin?  Pick one of the choices.  If you are wondering why there aren't any choices for scores involving half points, then go directly to jail.
Parent - - By turbojuice1122 (Gold) Date 2007-12-19 03:47
I'm currently leaning toward a 5-3 Rybka win, but this could change, of course--but that's the choice I picked.
Parent - - By Graham Banks (*****) Date 2007-12-19 04:10
I'll go for a 5-3 Rybka win also.   :-)
Parent - - By Permanent Brain (*****) Date 2007-12-19 16:00
I cannot imagine that Rybka can win the match under these conditions... a GM having White always, and can go far a draw. Also, long time control. - He should succeed at least in 5 from 8 attempts.
Parent - - By dareapa (**) Date 2007-12-19 16:26
It seems as if rybka will have her fins full as u suggest, but no one knows what this "new" rybka is capable of.  If she is stronger in certain positions or not.  I believe that matches that occur between updates will prove essential to the strength of rybka 3.0, so I hope J.B does very well. Improvements in rybka is what the "rybka fan" is really interested in. If rybka destroys J.B, will rybka have the incentive to improve?
Parent - - By turbojuice1122 (Gold) Date 2007-12-19 16:37
I think the incentive will be there and roughly the same regardless of the outcome of the match with Joel Benjamin, but the difference is that if Rybka loses, there will be a sort of basis for improvement, a sort of measuring stick by which to more accurately measure progress.  Of course, Joel doesn't have unlimited time in which to play matches, but Rybka could be improved based on decisions made in the games.  The only way I see Joel winning the match is with Father-like play; if this works, then newer versions can be tested against the sorts of moves made in those games.
Parent - - By dareapa (**) Date 2007-12-19 18:02
Is this Father-like play avoidable by rybka's book?  Does J.B. have his own style to draw or win? 
Parent - By turbojuice1122 (Gold) Date 2007-12-19 21:29
I don't think we know this now (though Larry might); we'll have to wait for January...
Parent - - By Banned for Life (Gold) Date 2007-12-19 21:41
There are too many ways to enter locked up positions to totally avoid them by book alone. Clearly Larry will be looking for openings that tend toward open games, but if the Rbyka variant that shows up for the match can be seduced into a locked up position through the use of sacrificed material, I think Rybka will probably lose.

Alan
Parent - By turbojuice1122 (Gold) Date 2007-12-19 21:56
If this is the case, and if Joel has already practiced a bit, it would be quite amusing if he began each game by moving a different pawn.
Parent - By Carl Bicknell (*****) Date 2007-12-19 16:35
There have been many, many examples of strong GM's playing white wanting draws off world champions who had black. In many cases they didn't get them.

I don't think being white = easy draw at all.
Parent - - By NATIONAL12 (Gold) Date 2007-12-19 23:07
i have forcasted same result as you,but if someone asked me to put down money i would run a mile.
Parent - - By turbojuice1122 (Gold) Date 2007-12-20 00:02
I probably would, too, especially since I can do that in under 4:40 :-)
Parent - By NATIONAL12 (Gold) Date 2007-12-20 00:12
i am a lot older than you,it would take me 10 mins.
Parent - - By BB (****) Date 2007-12-19 03:51
Time for one of my increasingly rare polls,

I was going to make this poll, but then decided to wait until 2 weeks before the match start (maybe to maximise hype). As usual, my choice (none of the above, as the January 6th late game turns out not to  affect the overall result and gets called off due to a conflict with the NFL playoffs) doesn't appear :(
Parent - By turbojuice1122 (Gold) Date 2007-12-19 13:12
They would probably play all eight games because Joel gets more money for each point he scores.
Parent - By skulibj (*) Date 2007-12-19 13:13
Tie. I am really excited about this match and will watch closely. I hope it will bring fireworks.
Parent - - By Uly (Gold) Date 2007-12-19 16:21
This time I have no prediction, but that the final score will be something like 7-1 or 8-0 (But I don't know who wins.)

Because either Rybka has an unbeatable strategy that "owns" the GM, or Rybka is "owned" by the GM due to bad strategy (Because the GM finds a way to win and uses it repeatedly.)
Parent - - By Hetman (*****) Date 2007-12-19 18:16
The poll will show the winner. Match is not neccessary :-).
Rgds
Hetman
Parent - - By Uly (Gold) Date 2007-12-19 19:31
If so then I challenge Rybka to a normal match without handicap and I expect to beat her. I only need to convince enough members to vote for me.
Parent - - By saxon (**) Date 2007-12-19 21:09
My guess is tie (4:4)
Parent - - By Uly (Gold) Date 2007-12-20 14:38
The only way I'm seeing a tie is if Rybka wins the first four games and exactly on the fifth, Joel finds the winning strategy and beats Rybka on the rest of the games. That seems highly unlikely.

I can't just imagine a "Joel wins, Rybka wins, Joel wins, Rybka wins" like result, unless Rybka is changing her strategy during the match.
Parent - - By turbojuice1122 (Gold) Date 2007-12-20 14:39
Or unless Larry has several different Rybka versions at his disposal that he will try during the match.
Parent - - By Uly (Gold) Date 2007-12-20 14:57
I would not consider this fair, Joel isn't switching brains, after all.
Parent - - By turbojuice1122 (Gold) Date 2007-12-20 16:39
Switching game approaches is the same thing--that's one of the things that gives some grandmasters an advantage over computers when they keep playing the same programs.
Parent - By Uly (Gold) Date 2007-12-20 19:16
But Rybka is a program, and doesn't care about how the opponent plays, if Joel faces a very different Rybka each game, I don't think he has a chance.

(But it seems there will only be slight book and contempt changes, so everything will be fair.)
Parent - - By lkaufman (*****) Date 2007-12-20 15:53
Obviously I will change the opening book during the match to avoid a repeat draw or loss. Of course this is fair; Joel can read opening books and/or review his games with computer help between rounds. I can also change settings like the contempt factor between games to make more variety. I doubt that we would make program changes between games unless a serious bug is found, though we are allowed to do so.
Parent - By Uly (Gold) Date 2007-12-20 15:57
Changing the opening book seems fair, and changing the contempt should be fine too, unless it changes Rybka's style drastically.
Parent - - By dareapa (**) Date 2007-12-20 16:08
Will this version of Rybka be the midway point to version 3.0.  So how much stronger from 2.3.2a...? 
Parent - By lkaufman (*****) Date 2007-12-20 20:48
We can call it midway, but we can't predict how much progress we will make between now and final release. As for the gains so far, we don't have any significant testing against other programs to measure it by, but if we say fifty points I think that won't be too far off from the truth in either direction.
Parent - - By Graham Banks (*****) Date 2007-12-20 20:13
Fiddling with the contempt factor is aways risky.
Parent - By lkaufman (*****) Date 2007-12-20 20:52
I don't think so. We had to forfeit a game last time because I set contempt on a version not ready for it, but once it is in use I don't see any problem with a modest change between games if we see that it was too high or too low.
Parent - - By Uly (Gold) Date 2008-01-05 23:58

> I can't just imagine a "Joel wins, Rybka wins, Joel wins, Rybka wins" like result, unless Rybka is changing her strategy during the match.


But on the first 5 games, it was "Rybka wins, Joel wins, Rybka wins, Joel wins, Rybka wins."

It seems I was very wrong.
Parent - - By turbojuice1122 (Gold) Date 2008-01-06 00:45
But you said the key thing: "unless Rybka is changing her strategy during the match".  This has been important.
Parent - - By lkaufman (*****) Date 2008-01-06 03:50
I wouldn't say Rybka is changing her strategy during the match. We are just changing contempt settings and making tweaks to try to reduce the chance of certain types of draws.
Parent - - By turbojuice1122 (Gold) Date 2008-01-06 04:43
Yes, that's exactly what I was talking about in my response--Rybka "is changing" in response to the state and history of the match.
Parent - - By Uri Blass (*****) Date 2008-01-06 07:04
Of course rybka is changing but even without change
I see no reason to think that what happened is impossible in the first place.

There is no simple strategy for a draw so the idea that if benjamin draws one game he can draw later every game is an idea with no basis.

Even if we have the same rybka with the same book in all the games you need to remember that rybka is noy deterministic because of
parallel search.

Uri
Parent - By turbojuice1122 (Gold) Date 2008-01-06 14:01
It's not completely deterministic, but the general strategies employed are fairly deterministic, and if a loss occurs due to an apparent error in something specific, then it would be unwise not to try to change that if there is good reason to believe that changing it will not hurt its performance.
Parent - By Hetman (*****) Date 2007-12-19 21:23
You have my vote Vytron :-)
Rgds
Hetman
Parent - - By turbojuice1122 (Gold) Date 2008-01-06 00:46
It's looking like we have a fairly decent chance that my 5-3 vote will be correct :-).  In fact, we have an extremely high chance that one of the two highest vote-getters will be correct.
Parent - By lkaufman (*****) Date 2008-01-06 05:44
All of our polls have been pretty accurate so far.
Parent - - By turbojuice1122 (Gold) Date 2008-01-07 00:48
Well, it looks like those who voted "6-2 Rybka" "won". :-)
Parent - By Felix Kling (Gold) Date 2008-01-07 05:24
I've won! I've won! I've won!

Felix :)
Up Topic Rybka Support & Discussion / Rybka Discussion / Poll: Result of Benjamin Draw odds match?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill