Two Pawn Handicap

[thekingman]

One thing demonstrated by this match is that whatever the handicap, repetition of the same plan should not be allowed. The first two games were hard-fought and exciting battles, then after Neumann found a winning plan in game 3, he simply repeated it to easily take games 4 and 5.

This suggests that the handicap should change from game to game. If you have c + f pawn in game 1, switch to any other pair of pawns for game 2, etc. This will bring variety into the games, making them more interesting from a competitive point of view and reducing the ability of the human to repeat the same winning plan in multiple games. After all, why have a multiple game match if they're all going to be fundamentally the same? This would also mean that if a certain two-pawn handicap (like c+f) is indeed too much, it will still only decide one game and not the entire match.

Several other non-pawn handicaps strike me as potentially useful (and more chess-like) for a match against a human. As discussed in another thread, one such idea is initiative odds, where the human gets a certain number of extra moves (of his own choice or predetermined to prevent repetition) or the computer starts with severely compromised king safety (ie, king starts on d5). I also like the idea mentioned above of human gets move and draw odds.

Another handicap I haven't seen mentioned before is castling rights. On its own, it wouldn't be enough, of course, but if you think f7 pawn and move isn't quite enough, you could also remove the computer's castling rights. I have not done testing to estimate the Elo cost of this additional handicap.

There are a lot of reasonable handicaps that could be given (like just about anything mentioned in this thread except time odds, which is silly), but the most important thing is to try and make each game in such a handicap match different, ideally by slightly changing the details (but not type) of the handicap itself in each game.

[lkaufman]

One thing demonstrated by this match is that whatever the handicap, repetition of the same plan should not be allowed. The first two games were hard-fought and exciting battles, then after Neumann found a winning plan in game 3, he simply repeated it to easily take games 4 and 5.

This suggests that the handicap should change from game to game. If you have c + f pawn in game 1, switch to any other pair of pawns for game 2, etc. This will bring variety into the games, making them more interesting from a competitive point of view and reducing the ability of the human to repeat the same winning plan in multiple games. After all, why have a multiple game match if they're all going to be fundamentally the same? This would also mean that if a certain two-pawn handicap (like c+f) is indeed too much, it will still only decide one game and not the entire match.

Several other non-pawn handicaps strike me as potentially useful (and more chess-like) for a match against a human. As discussed in another thread, one such idea is initiative odds, where the human gets a certain number of extra moves (of his own choice or predetermined to prevent repetition) or the computer starts with severely compromised king safety (ie, king starts on d5). I also like the idea mentioned above of human gets move and draw odds.

Another handicap I haven't seen mentioned before is castling rights. On its own, it wouldn't be enough, of course, but if you think f7 pawn and move isn't quite enough, you could also remove the computer's castling rights. I have not done testing to estimate the Elo cost of this additional handicap.

There are a lot of reasonable handicaps that could be given (like just about anything mentioned in this thread except time odds, which is silly), but the most important thing is to try and make each game in such a handicap match different, ideally by slightly changing the details (but not type) of the handicap itself in each game.

Yes, all of the above have been mentioned, including castling rights. I think a reasonable handicap for a fairly strong GM is to start every game with 1.e4 f6? 2.d4 Kf7?, which amounts to three moves plus castling rights. Yes, variety is important. We can't easily avoid any repeat handicaps, but we should try. In the case of f7 handicap, it's not too important to avoid repeating, because there are at least four playable replies to 1.e4 which will all lead to different games.

[Jhoravi]

A handicap where the Grandmaster as White plays 4 initial pawn moves in the opening is also interesting. Like c4,d4,e4,f4

[lkaufman]

A handicap where the Grandmaster as White plays 4 initial pawn moves in the opening is also interesting. Like c4,d4,e4,f4

Well, that's a four move handicap where his choice is restricted. I think giving him free choice which four moves to make (in his own half of the board) makes more sense. Why limit him to possibly suboptimal moves?