Handicap poll

[lkaufman]

Material handicaps were the clear preference, with support divided evenly between pawn and Exchange. So I think initially we'll aim for material handicap games with titled players at perhaps 45' +15". My current thinking is that for Elite GMs (2700+) we'll give f7 pawn or the Exchange, perhaps alternating. For average GMs two pawns (f2 and c2). For average IMs two pawns and first move (so f7 and c7). For FMs knight odds, Komodo winning draws. But nothing is set in stone, and we may have to modify our plans based on experience. We have already modified Komodo to avoid simplification in such contests, and we are refining that.

Aren't you afraid that a 2700 GM can outplay Komodo if he tries to study this kind of handicap ( f7 pawn ) with a strong engine like Stockfish ?

Yes, this is a real concern at that particular handicap, since Black has very limited options to avoid immediate defeat. But I think there will still be decent chances for Komodo, since although White will definitely get a won game out of the opening, it will not be an easily won game, just an extra pawn and a great position.

[carldaman]

Material handicaps were the clear preference, with support divided evenly between pawn and Exchange. So I think initially we'll aim for material handicap games with titled players at perhaps 45' +15". My current thinking is that for Elite GMs (2700+) we'll give f7 pawn or the Exchange, perhaps alternating. For average GMs two pawns (f2 and c2). For average IMs two pawns and first move (so f7 and c7). For FMs knight odds, Komodo winning draws. But nothing is set in stone, and we may have to modify our plans based on experience. We have already modified Komodo to avoid simplification in such contests, and we are refining that.

Aren't you afraid that a 2700 GM can outplay Komodo if he tries to study this kind of handicap ( f7 pawn ) with a strong engine like Stockfish ?

I was going to suggest the same potential problem. Komodo may be up against deep preparation instead of natural play.

This is why I thought a handicap by way of an unexpected gambit would have been the far better solution, preventing specific home preparation by the human player.

I also didn't like the way the gambit option was presented in Larry's poll (as a "blunder", which is frankly a turnoff to me and possibly other spectators). If you present it as a gambit (or gambits) more people will 'buy' into it.

CL

[lkaufman]

I'll go for the exchange plus move, coupled with the right to retract a move if it's before the computer has replied.

Exchange plus move feels about right to me for an average GM, too much for a 2650 I think. The right to retract a move before the computer has replied is probably worth another 50 elo or so; having the right to retract after seeing the engine move (but before your next move) is a much bigger advantage, maybe enough for Carlsen if he also gets White in normal chess.

[lkaufman]

Material handicaps were the clear preference, with support divided evenly between pawn and Exchange. So I think initially we'll aim for material handicap games with titled players at perhaps 45' +15". My current thinking is that for Elite GMs (2700+) we'll give f7 pawn or the Exchange, perhaps alternating. For average GMs two pawns (f2 and c2). For average IMs two pawns and first move (so f7 and c7). For FMs knight odds, Komodo winning draws. But nothing is set in stone, and we may have to modify our plans based on experience. We have already modified Komodo to avoid simplification in such contests, and we are refining that.

Aren't you afraid that a 2700 GM can outplay Komodo if he tries to study this kind of handicap ( f7 pawn ) with a strong engine like Stockfish ?

I was going to suggest the same potential problem. Komodo may be up against deep preparation instead of natural play.

This is why I thought a handicap by way of an unexpected gambit would have been the far better solution, preventing specific home preparation by the human player.

I also didn't like the way the gambit option was presented in Larry's poll (as a "blunder", which is frankly a turnoff to me and possibly other spectators). If you present it as a gambit (or gambits) more people will 'buy' into it.


I thought of that, but the problem is that if we want positions that Komodo scores as at least a full pawn down, we have to do things like 1.e4 b5?? or 1.e4 f5??. Unless there is some clear compensation for a pawn sacrifice, calling it a "gambit" is unreasonable; it is just a blunder. If we use real but bad gambits, such as 1.d4 e5? for example, Black will only be down something like half or 2/3 of a pawn, which is probably not enough of a handicap for any human except perhaps Magnus.

CL