man vs machine

[james uselton]

Ok---you have the strongest chess program, and a state of the art computer. How much handicap would the leading Grandmasters need---to have a chance of winning?

[michiguel]

Ok---you have the strongest chess program, and a state of the art computer. How much handicap would the leading Grandmasters need---to have a chance of winning?

110 Volts.

Miguel

[zullil]

Ok---you have the strongest chess program, and a state of the art computer. How much handicap would the leading Grandmasters need---to have a chance of winning?

How good a chance?

I really have no feeling for this. That said, I wonder if pawn+move would make the GM's probability of winning greater than the program's.

[Milos]

I really have no feeling for this. That said, I wonder if pawn+move would make the GM's probability of winning greater than the program's.

Against top program on i7 even minor piece might not be enough.

[Mike S.]

Against top program on i7 even minor piece might not be enough.

I would have made the same assumption, but then Rybka, on 8 cpu cores, played a match at knight odds against a FM (2284 Elo) and lost all four games:

http://rybkaforum.net/cgi-bin/rybkaforu ... l?tid=4249

Thinking time was 30m+30s only. which IMO made it more difficult for the human opponent (compared to 40/2h or similar).

By that I learned that if we talk about rather big odds, strong human players must not be underestimated vs. comps.

One pawn only is a much different question. There were some Rybka vs. IM/GM matches with this and other odds:

http://www.rybkachess.com/index.php?auswahl=Events

[gerold]

I really have no feeling for this. That said, I wonder if pawn+move would make the GM's probability of winning greater than the program's.

Against top program on i7 even minor piece might not be enough.

I piece may be enough. There would be a lot of draw gamesIMO.

[kurt]

Ok---you have the strongest chess program, and a state of the art computer. How much handicap would the leading Grandmasters need---to have a chance of winning?

Well; I venture to speculate that a high rated human if given a chance to consult with his prepared opening book lines with adequate time allocation should hold the presently best engine running on a fast computer to a draw.
Humans need to stay within familiar positions previously analyzed in great depth to overcome the engine search depth with pattern recognition where a normal 6 ply depth calculation is adequate to match 20ply or more of brute force position evaluations.
It seems only fair to make available the same opening library to humans when facing silicon monsters which use opening book lines to guide them into a sound position from which they can outplay humans.
But in speed chess humans only chance to hold is if the engine happens to select an opening line which allows guidance by perception where the degree of difficulties makes shallow depth search sufficient to match deep search because of the absent of imitate threats where the engine has multiple choices leading to no significant evaluation gains.
Yes I know this is a hotly debated topic. But my opinion stems from playing a lot of correspondence and OTB master level chess myself. IMO opening theory has already reached a level to an equivalent 3000 ELO rating and if allowed to consult should hold any engine to a draw.

[Mike S.]

FM John Meyer with 2284 Elo won all four knight odds games against Rybka on an octo, in a 30+30 match.

No draws

[kurt]

FM John Meyer with 2284 Elo won all four knight odds games against Rybka on an octo, in a 30+30 match.

No draws

This shows that engines can get outbooked by humans.
It also highlights the importance of book guidance for any engines.

[Mike S.]

I don't think this had anything to do with "outbooked". Rybka used a small book for these knight odds games (which obviously is required to avoid doubles). So, Meyer couldn't have predicted Rybka's early opening moves because these weren't engine moves.

He won at the board.