Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / What would be the most natural handicap for GM against Rybka
- - By ChessMate (***) Date 2007-07-14 13:39 Edited 2007-07-14 13:58
1.  Time handicap is just a nonsense as Shortening the time for rybka could be nullified  by increasing the processor speed.
In reality this (ridiculous handicap ) decreases the human pondering time and so reduces the human chances .
2.  Material handicap is just not a chess.
3.  Playing only with dark colour is natural though it only reflects one side of the engine and man capabilities.
4.  No opening book and no table bases are acceptable but  programmers may arguement that they could have been already added algorithms for  their engines if  EGTB and opening books are not available.

When talking about such handicaped matches  people are missing the basic question .
   " why we need the handicap matches  ?"
Just because on equal bases our engines are too strong to be matched by any Human GM .
So by un equal  time  or material odds the engines rating should be decreased and thus the gap between GM and engines.
But we are miising the point that people want GM and Rybka matches just because Rybka is the top engine and have crushing 3030+ elo , if u decrease the rybka's strength by these unnatural means in order to decrase the gap between Engine and GM  then  why not arrange  the GM matches only against low rated programs  and have  natural and more enteraining games.
It is really disappointing for me that still we don't have any Proper Rybka vs GM match.
Don't know why the organinser are more interested in organising handicaped matches .
I am sure all the chess followers just want to see Rybka playing at its full strengt against a grandmaster  and not these bullshit handicaped matches.
So the most natural way to give the GM a benefit  is to increase the time controls may be little longer than tournament time controls and only one game a day.
Parent - - By Felix Kling (Gold) Date 2007-07-14 13:48
Another possibility would be to play a normal match without handicaps and to say that it is enough to score let's say 2,5/8 for the GM to draw the match, 3/8 would be enough to win it.
Parent - By garyf919 (**) Date 2007-07-14 13:55
Here's another one. Have a 10 game match at normal tournament controls and rules.
No handicaps whatever, but if Rybka loses one game at any time, the match is over
and Rybka loses. Rybka would only be allowed to win or draw its games. I think some
GM's would think they have a chance to win one game out of ten.
Parent - - By Jim Walker (***) Date 2007-07-14 13:58
The only problem with things like 3/8 = win is that GMs would become draw masters and not play fighting chess.  We would end up with "father" type games. 
By the way, handicap games are not bullshit.  They have been played for centuries.  Two years ago I was campaining hard for GMs to play the computers because when the computers pass them up it will no longer be interesting to watch them get beat by computers.  Once they pass you up you will never catch up again.  They passed me in 1981.
Jim
Parent - By garyf919 (**) Date 2007-07-14 14:08
I just think people want to see Rybka playing at her full strength
against a quality GM. Don't get me wrong. Elhvest and Benjamin
are quality GM's, but not top 5 or 10. It seems to me that we will
never again see an epic match such as Kasparov-Deep Blue. I think
this is what enthusiasts really want. I know, it's all about the
money and sponsors, but maybe a better effort could be made to
get a big name sponsor (Intel, IBM, Microsoft, or combo of these, etc.)
Parent - By theoak (**) Date 2007-07-14 15:30
They passed me in 1997, though I have reasons to believe that this was done with "external" help. :) :) :)
Parent - - By Lukas Cimiotti (Bronze) Date 2007-07-14 16:22
I think, the best way to give a GM a chance against Rybka would be to use a very slow Computer, but allow Tablebases (the effect is small) and opening book.
Parent - - By JhorAVi (***) Date 2007-07-14 17:10
I disagree about TableBases and full Opening Book because Rybka isn't thinking when this things are on.. It's cheating! Unless the GM is also allowed to refer to opening and endgame books.. If it happens, then they are no longer playing chess!

I suggest:

1. Rybka to use the fastest hardware (we all want Rybka to play this way).

2. GM has no time limit with a maximum of whole day with time breaks (good compensation to rule 1).

3. Ponder On for Rybka (and GM :) )

4. No TableBase for Rybka (because TableBase isn't chess)

5. 3 Move Opening Book for Rybka (Because Opening book isn't chess too!)
Parent - - By Lukas Cimiotti (Bronze) Date 2007-07-14 17:31
I disagree
Tablebases are  chess - Vas relies on them, else he'd write some code to improove Rybka┬┤s abilities in endgames like kbnk.
Opening books are chess at its best - btw. GM uses them, too. Many openings are just stored in his brain.
Parent - By JhorAVi (***) Date 2007-07-14 18:57
An Opening Book isn't Rybka..  I got the feeling that during the opening phase Rybka is just sleeping and it's the FRITZ INTERFACE who's playing it..    when it's done, only then Rybka wakes up and start playing chess.
Parent - By George Tsavdaris (****) Date 2007-07-16 00:38

>I disagree about TableBases and full Opening Book because Rybka isn't thinking when this things are on.. It's cheating!


There is one playing entity:
-Rybka without opening book and without endgame tablebases.
And there is another playing entity:
-Rybka with opening book and with endgame tablebases.

The second playing entity is stronger than the first(because of the book mainly).
You suggest that the second playing entity is not Rybka. I agree. It is Rybka + opening book + tablebases. OK i agree.

I'm more interested in seeing a match between a GM against the strongest possible Chess entity, than against a weaker Chess entity.
There humans will prove their value. Against the strongest! Not the weaker....
Why you don't want this?

>An Opening Book isn't Rybka.. 


CPU of the computer is not Rybka too, but you allow it to help Rybka....
Parent - - By Dominik Landau (*) Date 2007-07-14 16:48 Edited 2007-07-14 16:56
For me the best configuration for a Match Rybka vs GM is the following:

1.) long thinking times like 2 hours / 40 moves to avoid zeitnot (time pressure)
2.) the GM can use fritz or shredder on a slow pc (< 500 MHZ) to avoid tactical blunder (like mate in one by Kramnik or instant losses)

There are no other handicaps necessary.

On the  other side i like pawn handicaps , because they give an symmetric imbalance. Thats good for programs since - on average - the game is going to have more action.
An yes - A pawn handicap IS Chess ;-)

Handicaps that i do NOT like are

1) to use a slow pc for Rybka
2) not to allow tablebases (because vas could have integratet endgame knowledge into his engine in that way that the difference in playing strength is less than 10 elo - but he did not)
3) giving rybka less time to think
4.) not to allow opening books (because vas could have integratet opening knowledge into his engine. I think he did not very much)
5.) playing only with dark colour .
Parent - - By ChessMate (***) Date 2007-07-14 22:17

> On the  other side i like pawn handicaps , because they give an symmetric imbalance. Thats good for programs since - on average - the game is going to have more action.
> An yes - A pawn handicap IS Chess ;-)
>


No  , In chess there are 32 pieces at the initial position . :-)
Parent - By Dominik Landau (*) Date 2007-07-15 23:04
true.

let's play chess not from the initial position :-)
Parent - - By MoKy (**) Date 2007-07-15 20:30
My suggestions how to balance a Man vs Rybka match:

a) one 90'+90" or 40/120' game per day
b) second board for analysis or Tablet PC without engine
c) max 4MB for Rybka's book and 100MB for EG positions

If this will not help, last chance is to build strong team of 6-10 GMs.

Mario
Parent - By Dominik Landau (*) Date 2007-07-15 23:06

> If this will not help, last chance is to build strong team of 6-10 GMs.


in that case i would prefer 6-10 different games: GM(n) vs Rybka
Parent - - By SR (****) Date 2007-07-14 23:09
One possibility an 8-10 game "sudden death" match with long time for both parties  (40 moves in 2.5 hours like 20 odd years ago).

Human wins if a win is produced.
Human losses, if computer wins 3 games in a row
The match is drawn otherwise  

This handicap can be adjusted, so human only losses if computer wins 5 games in a row etc.  Or could be made harder for the human by requiring the human to score 1.5 point in two games at any stage in the match.

Parent - - By ChessMate (***) Date 2007-07-15 17:45 Edited 2007-07-15 17:50

> One possibility an 8-10 game "sudden death" match with long time for both parties  (40 moves in 2.5 hours like 20 odd years ago).
>
> Human wins if a win is produced.
> Human losses, if computer wins 3 games in a row
> The match is drawn otherwise  
>
> This handicap can be adjusted, so human only losses if computer wins 5 games in a row etc.  Or could be made harder for the human by requiring the human to score 1.5 point in two games at any stage in the match.


These kind of handicaps could effect the out come of the match but have nothing to do with the outcome of the games.
More important  are the entertaining games under legal or natural rules not the match scores.
Parent - - By SR (****) Date 2007-07-15 18:24
Not quite sure what you are trying to say. My "sudden death" suggestion is a perfectly legitimate way of introducing a handicap, without distorting the game. My suggestion will encourage  the human to take certain risks, that I think will create more exciting games than most more ordinary setups where the human tries to avoid taking risks.
Parent - - By ChessMate (***) Date 2007-07-15 18:55 Edited 2007-07-15 18:58

> My suggestion will encourage  the human to take certain risks, that I think will create more exciting games than most more ordinary setups where the human tries to avoid taking risks.


Taking risk against (constant 3040+ player ) !!
The thing we need is a very strong GM who have atleast 8 hours to play a decent game against such a crushing chess entity.
Parent - By SR (****) Date 2007-07-16 09:17 Edited 2007-07-16 09:19
I think you focus too much on rating. My blitz rating against Rybka 2.3.2 is 2900+ and I am frequently are able to draw. In longer time controls my chances are even better.
Twenty odd years ago my ELO was 2300, but I think I am playing somewhat better today since I learned a lot about chess from the programs. I have specialized in playing computers (I have not played any over the board games in the last 10 years) and a few years ago I could thrash most programs.

So far I have NOT been able to win a single game (at any time controls) against 2.3.2, while I did occasionally win against some of the earlier versions of Rybka. In general I play openings where Rybka seems (at least in my mind) to misjudge the resulting middle game. My PC is just a single processor PC and I am not sure what would happen if I played on a Quad.
My suspicion is that some of my opening lines would not work anymore.

My sudden death handicap is a serious suggestion than I think will create interesting match condition. If I was to play with those conditions against Rybka on my 3200Mhz single processor PC, I would have reasonable chances drawing the match, and I just discovered that I can "trick" Rybka into favorable position when I play black. If I win, Rybka will change its preferences for the opening, but then it would be too late! I would have won the "sudden death" match...
Parent - By plicocf (***) Date 2007-07-16 16:38
I think that the last match against GM Elhvest was ok, maybe
just more time for the human.

Paulo Soares
Up Topic Rybka Support & Discussion / Rybka Discussion / What would be the most natural handicap for GM against Rybka

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill