Some handicap results and conclusions.

[lkaufman]

In an other subject. I see on chess.com Komodochess is rated at bullet without playing game. I think a nice idea would be to see the elo rating at bullet or blitz of komodo with some handicap. Player could play against it with knight handicap or rook handicap. I don't think any player would win against knight at bullet, but who knows, and they can play it at blitz.

I had the same idea. I think the problem is that as far as I know, chess.com has not yet implemented automatic play by an engine other than one on their own computers. I suppose they could set up Komodo on one of their computers to play handicap games (not even sure about that), but probably they don't have very powerful computers available for this purpose. We know from our own experience that on my big machine Komodo (almost 2 years ago) could break even giving knight odds in blitz (3 min plus 2 sec) to GMs Lenderman and Mikhalevski. Another point is that a knight odds book would be more or less necessary to avoid repeat openings; I could do that of course.

[Lyudmil Tsvetkov]

Plz Larry, let us know about the test result.

The doubled pawn in the shelter idea tested poorly; we'll try the queen-king tropism idea next. I think we tried it long ago and it failed, but the devil is in the details. It is a sound idea.

Larry, just to ask you, if you don't mind, if you mind, I will just stop:

how do you define doubled pawns, and how do you define shelter?

certainly, you checked only for doubled pawns on the 2nd and 3rd ranks on the file where the king is as well as the 2 adjacent files?

because SF defines its shelter as going as high as the 6th rank, and considering doubled weakness for any rank above 3 is simply bad.

did you try penalising only g3/b3 doubled shelter pawn?

I don't want to get into detail, but I'll say that the way we tried was reasonably close to what you would do. As for the king queen tropism idea, it also tested a bit negative, in this case due to the roughly 1% slowdown it entailed; on a fixed depth basis it was slightly plus. This is the case with many ideas; they are helpful if you ignore the cost of computation, but net losers in reality. That's why many of your ideas are not practical. But this is not your problem, it's up to us to judge whether the cost of an idea is low enough to try it.

right, sometimes trying an idea will depend on its speed performance, and this performance in turn probably by the way it is implemented.

for example, concerning king tropism, king-queen proximity.

I know of 3 ways trying to do this:

- giving bonus for pieces attacking the king side, the way SF does it, which is not bad, but you have to figure out what particular squares those pieces attack, how far from the king
- calculating tropism in terms of the square distance between piece and enemy king(they have some precomputed tables based on square indices), which I think is far from optimal, as the distance itself is far less important than the presence of the piece/queen on the side where the enemy king is
- giving bonus for the presence of piece/queen on the side where the enemy king is, the way I like to do perceive it

now, in case one opts for the 3rd option, the problem is how to implement it.
and, seemingly, there are many ways again:

- giving bonus for presence of queen on the side where the enemy king is, which might be fine, but will clash with direct queen shelter attacks, so one might choose to specify bonus is valid only if queen not attacking the king shelter
- giving penalty for the queen, when not on the side where the enemy king is, which makes more sense practically, as in many cases such a queen will not attack the shelter, and so you avoid redundancies
- same queen penalty, but specifying the penalty is given only when the enemy queen is on the opposite side, as in many cases the penalty might fail, because both queens are on the queen side, with one version penalising it and another not, and zero game effect

[Lyudmil Tsvetkov]

In an other subject. I see on chess.com Komodochess is rated at bullet without playing game. I think a nice idea would be to see the elo rating at bullet or blitz of komodo with some handicap. Player could play against it with knight handicap or rook handicap. I don't think any player would win against knight at bullet, but who knows, and they can play it at blitz.

I had the same idea. I think the problem is that as far as I know, chess.com has not yet implemented automatic play by an engine other than one on their own computers. I suppose they could set up Komodo on one of their computers to play handicap games (not even sure about that), but probably they don't have very powerful computers available for this purpose. We know from our own experience that on my big machine Komodo (almost 2 years ago) could break even giving knight odds in blitz (3 min plus 2 sec) to GMs Lenderman and Mikhalevski. Another point is that a knight odds book would be more or less necessary to avoid repeat openings; I could do that of course.

another option is to open an entire enterprise, based on human-Komodo, human-engine games.

Engine is constantly available for competition, when a game is played, there are some stakes, if engine wins, human pays small maount, if draw, some specific conditions apply, if human wins, human gets some nice amount.

this could be tried with different handicaps, and I am certain interest will be very high, if stakes and odds are reasonable.

one can make a looot of money, after all, by beating Komodo.

[JJJ]

In an other subject. I see on chess.com Komodochess is rated at bullet without playing game. I think a nice idea would be to see the elo rating at bullet or blitz of komodo with some handicap. Player could play against it with knight handicap or rook handicap. I don't think any player would win against knight at bullet, but who knows, and they can play it at blitz.

I had the same idea. I think the problem is that as far as I know, chess.com has not yet implemented automatic play by an engine other than one on their own computers. I suppose they could set up Komodo on one of their computers to play handicap games (not even sure about that), but probably they don't have very powerful computers available for this purpose. We know from our own experience that on my big machine Komodo (almost 2 years ago) could break even giving knight odds in blitz (3 min plus 2 sec) to GMs Lenderman and Mikhalevski. Another point is that a knight odds book would be more or less necessary to avoid repeat openings; I could do that of course.

another option is to open an entire enterprise, based on human-Komodo, human-engine games.

Engine is constantly available for competition, when a game is played, there are some stakes, if engine wins, human pays small maount, if draw, some specific conditions apply, if human wins, human gets some nice amount.

this could be tried with different handicaps, and I am certain interest will be very high, if stakes and odds are reasonable.

one can make a looot of money, after all, by beating Komodo.

It would be tough to know if the players had cheat or not.

[lkaufman]

I had a new idea for a handicap for players a bit too strong for knight odds but not strong enough for the smaller handicaps we have used. It is knight for tempo odds; White (Komodo) opens 1e4, Black "takes" the knight off of b1 (or g1 as specified before the game) but this counts as his move, so it is now White's turn. Actually I didn't invent this, Philidor sometimes played this way, I just never analyzed it before. It looks sufficiently harder than normal knight odds to be worth a try. Can anyone suggest a better name for it than "knight for tempo"? We could use "Philidor's knight odds" but no one would know what that means.

[JJJ]

Knight for tempo sounds really good to me and would be nice to watch. I don't know if it would chance the outcome, but worth the shot I guess.

[Lyudmil Tsvetkov]

I had a new idea for a handicap for players a bit too strong for knight odds but not strong enough for the smaller handicaps we have used. It is knight for tempo odds; White (Komodo) opens 1e4, Black "takes" the knight off of b1 (or g1 as specified before the game) but this counts as his move, so it is now White's turn. Actually I didn't invent this, Philidor sometimes played this way, I just never analyzed it before. It looks sufficiently harder than normal knight odds to be worth a try. Can anyone suggest a better name for it than "knight for tempo"? We could use "Philidor's knight odds" but no one would know what that means.

this one is really much more reasonable than pure knight odds.

I am unable to win all games vs SF, as with pure knight odds, at least for the time being I have not accustomed to it.

and play is interesting, I would say, humans should be very careful with engine initiative.

[Lyudmil Tsvetkov]

to find the optimum, you might want to test knight odds with 1,2,3,4 etc. spare engine moves.

[Teemu Pudas]

Plz Larry, let us know about the test result.

The doubled pawn in the shelter idea tested poorly; we'll try the queen-king tropism idea next. I think we tried it long ago and it failed, but the devil is in the details. It is a sound idea.

Larry, just to ask you, if you don't mind, if you mind, I will just stop:

how do you define doubled pawns, and how do you define shelter?

certainly, you checked only for doubled pawns on the 2nd and 3rd ranks on the file where the king is as well as the 2 adjacent files?

because SF defines its shelter as going as high as the 6th rank, and considering doubled weakness for any rank above 3 is simply bad.

did you try penalising only g3/b3 doubled shelter pawn?

I don't want to get into detail, but I'll say that the way we tried was reasonably close to what you would do. As for the king queen tropism idea, it also tested a bit negative, in this case due to the roughly 1% slowdown it entailed; on a fixed depth basis it was slightly plus. This is the case with many ideas; they are helpful if you ignore the cost of computation, but net losers in reality. That's why many of your ideas are not practical. But this is not your problem, it's up to us to judge whether the cost of an idea is low enough to try it.

Suppose a halving of speed costs 70 Elo. Then a 1% slowdown costs about 1 Elo. Now, if you have 100 ideas that are worth 1 Elo each at equal NPS but slow down the engine by 1% (compared to its current speed), you can make the engine 30 Elo stronger by implementing them all (if the ideas are truly orthogonal).

[Lyudmil Tsvetkov]

to find the optimum, you might want to test knight odds with 1,2,3,4 etc. spare engine moves.

just 2 tempos are still doable, more or less easily, of course, after some preparation.

I don't know however how black wins that one without tremendous difficultites against a top engine:

[d]rnbqkbnr/pppppppp/8/8/3PP3/8/PPP2PPP/R1BQKBNR w KQkq - 0 1

maybe a good position for a match against a top GM.