Komi pie: will it ever become widely used?

[watchingo]

At my local club, we actually use "komi pie" almost every game.
Except we call it "auction komi."

First we do nigiri, then the winner chooses if they want to bid or choose the color.
It seems to work great for us and I find it enjoyable. To our knowledge, we haven't encountered any problems.
There is some fun in choosing a komi outside the norm like 2.5 or 10. Although, most the time we say 5.5 to 7.5.

Then again, we are all friends and play each other every week.
Maybe next time when there is a newcomer, I will try "auction komi" and see how it goes.

I also like the idea of "blind auction komi", i'll give that a shot as well.

[ez4u]

I agree with oren.

Komi was calculated statiscally from game results with players of same strength. ..

I agree this probably won't come to be widely used. But I don't think the stats argument really works that well. That's because the statistically calculated komi is a single number calculated from a group of professional players.

There exists the possibility that the "most fair" komi for two amateur 5k, for example, is a different number. At least with the komi pie, these two players could try to find this "fair" komi.

This seems confuse 'komi', to compensate for the advantage of playing first, with 'handicap', to compensate for a difference in strength.

[Kirby]

Kirby, do you think we should have endgame quizzes before the game starts, and award points to the winner?

No, but what if 6.5 or 7.5 isn't the ideal komi for amateurs? Then someone is automatically at a disadvantage when they start the game with a given color.

With the komi rule, the bidder has the option to keep things the same and bid 6.5 or 7.5. But there's also flexibility to fix the inaccuracy of the komi if it hypothetically exists.

[Kirby]

1. Knowledge of what a good komi is for you is part of your strength.

Huh? As I said, the correct komi for a game between me and you depends on what a good komi is for me and what a good komi is for you. The game is, after all, between the two of us. Making a choice on what the best komi is based on insufficient information (aka guessing) doesn't seem particularly more useful or fairer than having it fixed.

It may be difficult to know what a "good" komi is for amateurs. But at the very least, it can be as good as the current system with a bid of a "normal" komi.

[Kirby]

This seems confuse 'komi', to compensate for the advantage of playing first, with 'handicap', to compensate for a difference in strength.

Why? Having initiative by playing first is something that can affect the outcome of the game. It's not clear to me that this "initiative factor" is necessarily identical between pros and amateurs.

For example, let's say you take a 1000 5k games, and find out that, in 900 of them, black wins by playing first with a 7.5 komi. The komi doesn't seem correct for 5k - perhaps because "initiative" can mean a lot in 5k games.

It's unlikely to be exaggerated like this, but this is the general idea I am trying to get at.

[Bill Spight]

This seems confuse 'komi', to compensate for the advantage of playing first, with 'handicap', to compensate for a difference in strength.

Why? Having initiative by playing first is something that can affect the outcome of the game. It's not clear to me that this "initiative factor" is necessarily identical between pros and amateurs.

For example, let's say you take a 1000 5k games, and find out that, in 900 of them, black wins by playing first with a 7.5 komi. The komi doesn't seem correct for 5k - perhaps because "initiative" can mean a lot in 5k games.

It's unlikely to be exaggerated like this, but this is the general idea I am trying to get at.

In theory, if both players are sufficiently far from pro strength, komi should be less. IIRC, the results of European tournaments supports that idea. Komi for beginners should probably be around 3. Maybe komi for 4 kyus should be around 5. Similarly, the win rate for a one stone difference in skill should diminish as the skill level drops.

[Uberdude]

The "correct" komi could even differ between the same two players of equal skill depending on the type of game they play. It seems quite plausible to me that in a moyo game with large territories the fair komi is larger than in a low-scoring game of many small fragmented territories.

[ez4u]

This seems confuse 'komi', to compensate for the advantage of playing first, with 'handicap', to compensate for a difference in strength.

Why? Having initiative by playing first is something that can affect the outcome of the game. It's not clear to me that this "initiative factor" is necessarily identical between pros and amateurs.

For example, let's say you take a 1000 5k games, and find out that, in 900 of them, black wins by playing first with a 7.5 komi. The komi doesn't seem correct for 5k - perhaps because "initiative" can mean a lot in 5k games.

It's unlikely to be exaggerated like this, but this is the general idea I am trying to get at.

If I understand this idea correctly, you would be expecting to see different winning rates by dan/kyu level at existing komi, e.g. kyu-level players win more as black at 6.5 komi than high dans, etc. The information is there in the hundreds of thousands of amateur games stored on servers like KGS. As soon as you get through the analysis please let us know your results. Absent that you have no general basis for cutting the pie other than either: the existing work/experience on komi that points to something very close to 6.5 points or some entirely personal experience with the person sitting across from you. The second, I would still claim is a consideration of handicap. The best description of any general claims regarding differences for kyu levels in the absense of some sort of analysis would be "pie in the sky", IMHO.

[hyperpape]

Kirby, do you think we should have endgame quizzes before the game starts, and award points to the winner?

No, but what if 6.5 or 7.5 isn't the ideal komi for amateurs? Then someone is automatically at a disadvantage when they start the game with a given color.

And with komi pie, we have the enjoyable task of following the estimates on the ideal komi for our rank!

If it's a problem, do the analysis, determine the optimal komi, convince the AGA and the EGF to change their events' komi for amateurs. Komi pie just makes us do busywork.

[Kirby]

This seems confuse 'komi', to compensate for the advantage of playing first, with 'handicap', to compensate for a difference in strength.

Why? Having initiative by playing first is something that can affect the outcome of the game. It's not clear to me that this "initiative factor" is necessarily identical between pros and amateurs.

For example, let's say you take a 1000 5k games, and find out that, in 900 of them, black wins by playing first with a 7.5 komi. The komi doesn't seem correct for 5k - perhaps because "initiative" can mean a lot in 5k games.

It's unlikely to be exaggerated like this, but this is the general idea I am trying to get at.

If I understand this idea correctly, you would be expecting to see different winning rates by dan/kyu level at existing komi, e.g. kyu-level players win more as black at 6.5 komi than high dans, etc. The information is there in the hundreds of thousands of amateur games stored on servers like KGS. As soon as you get through the analysis please let us know your results. Absent that you have no general basis for cutting the pie other than either: the existing work/experience on komi that points to something very close to 6.5 points or some entirely personal experience with the person sitting across from you. The second, I would still claim is a consideration of handicap. The best description of any general claims regarding differences for kyu levels in the absense of some sort of analysis would be "pie in the sky", IMHO.

Well, that's the point, really - absense of information. The existing komi is there because of analysis of pro games as I understand it. Until we do an analysis of amateur games, we can't know what is correct - and the existing 7.5 is pie in the sky Hence, a flexible system like the pie rule at least does not restrict us to the pro-analysis-defined komi.

Like Bill said, maybe a lower komi is better like 3 or 4 or 5 for some ranks. Maybe if the pie rule were in place, I could bid this.

[Kirby]

Kirby, do you think we should have endgame quizzes before the game starts, and award points to the winner?

No, but what if 6.5 or 7.5 isn't the ideal komi for amateurs? Then someone is automatically at a disadvantage when they start the game with a given color.

And with komi pie, we have the enjoyable task of following the estimates on the ideal komi for our rank!

If it's a problem, do the analysis, determine the optimal komi, convince the AGA and the EGF to change their events' komi for amateurs. Komi pie just makes us do busywork.

No, you are free to bid a komi of 6.5 or 7.5 if you'd like, just to keep things as they are. But for those that are interested in different komi - they have e flexibility to do so.

What I'm really saying is:
1. We don't know what "good" komi is for amateurs.
2. The pie rule is more flexible and allows those that are interested to use a different number.

[Kirby]

Absent that you have no general basis for cutting the pie other than either: the existing work/experience on komi that points to something very close to 6.5 points or some entirely personal experience with the person sitting across from you. The second, I would still claim is a consideration of handicap. The best description of any general claims regarding differences for kyu levels in the absense of some sort of analysis would be "pie in the sky", IMHO.

Let me put it this way. With the quote above, it looks like the claim is that the burden of proof is on me to say that the correct komi is not 6.5 or 7.5.

I don't see it that way, because I do not claim that the correct komi is not 6.5 or 7.5.

Rather, I claim that we have not proven that 6.5 or 7.5 is a good komi for amateur players since this analysis was over pro games. With the pie rule, NO claim is made about correct komi by the system.

Correspondingly, the system itself cannot be wrong - it defers judgement to the players.

So perhaps the correct komi for, say, 5k is 6.5. But I feel the burden of proof is upon this claim. Using the pie rule, you are free to go with komi as you'd like, proven or not.

Don't think another komi is fair? Bid 6.5. If you're not the bidder, choose the side accordingly. At least as good as the current system, is it not?

[Bill Spight]

The "correct" komi could even differ between the same two players of equal skill depending on the type of game they play. It seems quite plausible to me that in a moyo game with large territories the fair komi is larger than in a low-scoring game of many small fragmented territories.

I guess that your idea is that the variance of the results is larger with moyo games than tight territorial games? But if the results are normally distributed, then just changing the variance will not affect the median result, which is what determines komi in theory.

OTOH, if not playing a moyo game reduces the median result for Black, Black should play for moyo.

[oren]

At least as good as the current system, is it not?

I want to sit down and play go. If I wanted an extra bidding phase, I'd play bridge.

In friendly go over the board, I don't even discuss komi. Online, I would expect 6.5 nowadays. I don't really want an extra phase to game starting.

I've used a lot of 'proper handicap' on DGS which has a lot of ranges for komi which can be interesting, but I don't mind since it's not an extra phase to handle.

[ez4u]

Absent that you have no general basis for cutting the pie other than either: the existing work/experience on komi that points to something very close to 6.5 points or some entirely personal experience with the person sitting across from you. The second, I would still claim is a consideration of handicap. The best description of any general claims regarding differences for kyu levels in the absense of some sort of analysis would be "pie in the sky", IMHO.

Let me put it this way. With the quote above, it looks like the claim is that the burden of proof is on me to say that the correct komi is not 6.5 or 7.5.

I don't see it that way, because I do not claim that the correct komi is not 6.5 or 7.5.

Rather, I claim that we have not proven that 6.5 or 7.5 is a good komi for amateur players since this analysis was over pro games. With the pie rule, NO claim is made about correct komi by the system.

Correspondingly, the system itself cannot be wrong - it defers judgement to the players.

So perhaps the correct komi for, say, 5k is 6.5. But I feel the burden of proof is upon this claim. Using the pie rule, you are free to go with komi as you'd like, proven or not.

Don't think another komi is fair? Bid 6.5. If you're not the bidder, choose the side accordingly. At least as good as the current system, is it not?

I am not sure who you think the 'we' is that haven't proved something. I have played thousands of games with existing komi and do not feel that it favors Black or White. So 'I' have 'proven' that 6.5 is fine for amateur play. I do not claim that I can play accurately enough that small differences in komi make a difference anyway.

And yes the burden of proof is on you. Why? Because Go players are just like everyone else. On average we are complacent, lazy you-know-whats who don't like to change our ways. Nothing new there, right?