Two Pawn Handicap

[mhull]

The reason analysis in books (pre-computer) was so much better than actual play is not primarily due to being able to move the pieces, but to spending hours rather than minutes on the key moves. Letting a 2800 player analyze for say one hour per move with a board for a game played over a week or so (with computer taking say one hour total) would indeed be an interesting handicap, but too expensive for us to sponsor.

Perhaps you've already considered that a 20:1 time ratio could be achieved even at standard time controls (ponder-off). So instead of an average 3 minutes per move as the human gets, Komodo gets 9 seconds. You could even pad this with idle time for the rest of the 3 minutes so the human has his usual ponder time on the opponent's clock. So Komodo thinks for 9 seconds, then waits to play its move until the 3 minute mark more or less.

[Laskos]

The reason analysis in books (pre-computer) was so much better than actual play is not primarily due to being able to move the pieces, but to spending hours rather than minutes on the key moves. Letting a 2800 player analyze for say one hour per move with a board for a game played over a week or so (with computer taking say one hour total) would indeed be an interesting handicap, but too expensive for us to sponsor.

Perhaps you've already considered that a 20:1 time ratio could be achieved even at standard time controls (ponder-off). So instead of an average 3 minutes per move as the human gets, Komodo gets 9 seconds. You could even pad this with idle time for the rest of the 3 minutes so the human has his usual ponder time on the opponent's clock. So Komodo thinks for 9 seconds, then waits to play its move until the 3 minute mark more or less.

I don't like this time handicap matching. Most people have no idea what a factor of 20 or 200 in time means, even specialists would disagree in terms of FIDE ratings. Most players understand what two pawns handicap is, even if at first it's maybe not that impressive in the eyes of an weak amateur player. Then, making a much weaker Komodo can be done creatively, for example setting king safety to such weird values that Komodo will go for the kill disregarding its pieces or even the king.

[chessico]

I'm just saying that it's a much smaller factor for 2800 players than for 2200 players.

Thanks for the hint.

... Or they find better moves deep on some line, that don't change the root move.

Some better moves may change the root move, others may not. Actually I am surprised to have a discussion about if moving the pieces, at least after the game, can lead to better moves. People may just have forgotten how it was to analyze a position on their own. Some twenty years ago only a tiny minority would have said the best way to do it is staring at the board for hours without touching any piece. Sounds absurd to me ... That does not mean that senseless hand work alone does the job, of course.

I might actually try it in some test games to see if it is only a distraction or can be productive. Although a stronger player would have better chances, of course.

[mhull]

The reason analysis in books (pre-computer) was so much better than actual play is not primarily due to being able to move the pieces, but to spending hours rather than minutes on the key moves. Letting a 2800 player analyze for say one hour per move with a board for a game played over a week or so (with computer taking say one hour total) would indeed be an interesting handicap, but too expensive for us to sponsor.

Perhaps you've already considered that a 20:1 time ratio could be achieved even at standard time controls (ponder-off). So instead of an average 3 minutes per move as the human gets, Komodo gets 9 seconds. You could even pad this with idle time for the rest of the 3 minutes so the human has his usual ponder time on the opponent's clock. So Komodo thinks for 9 seconds, then waits to play its move until the 3 minute mark more or less.

I don't like this time handicap matching. Most people have no idea what a factor of 20 or 200 in time means, even specialists would disagree in terms of FIDE ratings. Then, making a much weaker Komodo can be done creatively, for example setting king safety to such weird values that Komodo will go for the kill disregarding its pieces or even the king.

I wouldn't favor a weakening by a software lobotomy. This is why I suggest slowing down the AI by taking it back in time, as it were, to slower hardware. You can simulate this by introducing delay loops to slow-down the effective NPS or simply restrict the time to achieve a similar effect.

But I do see the attraction of chessmen handicaps since the AI gets to keep his "marbles" and his think time at more or less full speed.

[Jesse Gersenson]

That game should not have been played, I certainly count the sixth game as a resignation without playing. I did not like the idea of these handicap matches at all, I must admit, but if you make big claims in order to sell more copies of your (fine) program, then at least stick with the rules when things don't go as expected.

Yes, we conceded the 6th f2/c2 game. Neuman demonstrated, very methodically, how to win against f2/c2 odds.

"Stick with the rules"?? Which rules? All of the moves were legal and, except for some trash talking and casual banter during the game, it was just like a tournament atmosphere. The match aimed to test this unexplored handicap and test Komodo's strength against a human opponent at this handicap. The handicap proved too big and Komodo had zero chances for counterplay from game 3 onward. Rather than waste the opportunity to play a strong human we tried a different handicap in game 6 and would have tried something different in game 5 too but this wasn't practical.

Pawn and move is probably an even match against Neuman.

[chessico]

Yes, we conceded the 6th f2/c2 game. Neuman demonstrated, very methodically, how to win against f2/c2 odds.

I may be wrong, but it looks like the final result, correctly including the conceded game, was not even posted "officially". When I looked for it, all I could find is some remarks from Larry, and the very important fact that Komodo had actually won a game, that was not even on the schedule. The GM had no chance to prepare for it, just saying. So no result, but some random success that has nothing to do with the announced event. That's what made me a little uneasy ...
People may buy this "for fun and in order to learn something"-stuff, but let's be honest and admit it is a marketing event, of course. It did not go as expected, and I would find it much better style to give this fact some room to be processed by the potential customers.
The basic "problem" (of course, there is not really one) is that Komodo cannot develop a strategy to deal with any kind of handicap situation. That's because it never does develop a strategy. The stupid humans like Carlsen, Kasparov etc. would have much better chances to beat an opponent like Neuman under the given circumstances, because they can develop a strategy. So the whole event actually highlights the general problems of computer chess instead of showing how much engines have improved. While a player whose Elo is only 200 Elo higher can potentially win in such a situation, the engine needs an absurd rating advantage because of its inflexibility to adapt to the situation.

[Jhoravi]

The odds game normally starts with a Loosing game for Komodo but Komodo played PERFECT MOVES! If you play Perfect Moves in a loosing position it will eventually lead you to the Truth which is Defeat.

My point is that, normal strategy in handling equal position doesn't apply in lost positions. Komodo or any other engine are don't designed to adept in such condition.

[lkaufman]

The reason analysis in books (pre-computer) was so much better than actual play is not primarily due to being able to move the pieces, but to spending hours rather than minutes on the key moves. Letting a 2800 player analyze for say one hour per move with a board for a game played over a week or so (with computer taking say one hour total) would indeed be an interesting handicap, but too expensive for us to sponsor.

Perhaps you've already considered that a 20:1 time ratio could be achieved even at standard time controls (ponder-off). So instead of an average 3 minutes per move as the human gets, Komodo gets 9 seconds. You could even pad this with idle time for the rest of the 3 minutes so the human has his usual ponder time on the opponent's clock. So Komodo thinks for 9 seconds, then waits to play its move until the 3 minute mark more or less.

A more aopealing way to accomplish that goal is simply to play on one core with equal or maybe half time. For bigger time odds we would run Komodo on a cellphone. But these ideas don't really appeal to me much, because running on one core (or simulating doing so the way you describe) is just a test of how we would do on ancient hardware, while the cellphone idea is a bit more relevant but not something we would spend much money on, as the cellphone version is too cheap to be a significant revenue source.

[lkaufman]

It seems most of us underestimated the grandmaster in this match or at least underestimated the handicap. I think it's pretty obvious now that the handicap of f2 and c2 pawns was just too much for anyone or anything to give to a grandmaster in a serious game. Yet Komodo won fairly easily playing Black and giving the f7 pawn only, which is the worst pawn for Black to remove.
I think two pawn handicap is still playable against an ordinary (around 2500) grandmaster, if we are a bit more conservative about the choice of the pawns. As it was, White was not only two pawns down, but his king was weakened and his pawns were split up into three groups. Moreover he only had two of the four pawns that can control central squares. I chose this handicap out of deference to the tradition of giving the "f" pawn as a handicap, but it's just too difficult, especially if repeated game after game while the grandmaster learns each time.
When Kasparov gave two pawns to Terrence Chapman (said to be 2150 level) in a match, he removed the "a" pawn plus one other varying pawn. This is what we should have done too, although I think it was a bit unfair to play one game with both edge pawns removed, which is probably no more than the f7 handicap. I think Komodo can still offer two pawns to a grandmaster, if one is the "a" pawn and the other rotates between "b", "c", "d", and "e". These feel more like "just" a two pawn handicap with no added positional advantages on top.
Comments, anyone?

I prefer 90+30 time control and only one pawn handicap.
If you do not use 90+30 you cannot claim that you beat GM's with pawn handicap in normal chess and GM's do not use faster time control than 90+30 in normal chess.

Once the f7 pawn is removed from the starting position adjusting the time conditions won't ever get you to fide's definition of 'normal' chess (I assume you mean "standard" chess). Further, we disagree what constitutes 'normal' chess. You contend it's defined by FIDE; I contend it's defined by the two people playing the game, less you have instances such as, "That's not how we normally play, around here."

Morphy played Anderson and was using about an hour per game, Anderson was using about 8 hours. Was that normal chess? Check online games and figure out what percent of those online games are played at the 'normal' time controls you're claiming as normal. On the contrary, check how long games played in the noise of bar rooms take, how long do two players sit at the board in the santity of their living rooms, or the nooks of cafes?

Larry, we should offer Terrence Chapman knight odds. Also, as a guideline, the odds should be adjusted if there are two or three losses in a row by either side.

I don't know if Chapman even plays chess now, but it's an interesting suggestion. He might be too close to FM strength for knight odds though. Actually I'm expecting a visitor from England with a FIDE rating near 2100 who expressed interest in a knight odds short match around the end of the month, so he might be right about the proper level for this.

[JJJ]

I don't like any time handicap or hardware handicap for Komodo. Even half pot control time is not nice to me.

I just like pawn handicap. But I still think the pawn should be random, because the worst black pawn will be really easy to draw for a 2700 or 2800 GM, and in the long run, an easy win also.

If you want to play versus 2500 GM average, 2 pawn handicap is nice, if they re not the worst possible. ab pawn are nice, bc pawn are nice also, and I m sure you can find others.