Draws would still not be that common even at the highest level and it might be more balanced? I read that in New Zealand a komi of 7 is used.

I would favor making any single digit difference a draw.

Go is such a complicated game, that if you can't win by at least 10 points, you are not clearly better.
( This may not be true at the pro level, but for the other 99.9% of us, it works )

_________________
Help make L19 more organized. Make an index: https://lifein19x19.com/viewtopic.php?f=14&t=5207

I would say that it depends on the style of the game. For games with lots of big life and death of large dragons, big point swings may happen easily. When it's more just both players collecting points here and there, small differences can be common - and sometimes hard to overcome when the other guy is better than you.

_________________
be immersed

Here is a citation of my related theorem:

"Definition 40 [standard area komi]
Standard area komi has the form K := 2z - 0.5 for some whole number z.

Presupposition
We use standard area komi.
Theorem 140 [area scores and winners]
If the board parity and seki parity are unequal, the possible area scores are ..., -4.5, -2.5, -0.5 (smallest white win), 1.5 (smallest black win), 3.5, 5.5,... points. Increasing the komi by 1 does not change the winner.
If the board parity and seki parity are equal, the possible area scores are ..., -5.5, -3.5, -1.5 (smallest white win), 0.5 (smallest black win), 2.5, 4.5,... points. Decreasing the komi by 1 does not change the winner." [21]

Click Here To Show Diagram Code

[go]$$B Komi = 7, Score = 0
$$ ----------------
$$ | . . . . . . .
$$ | . . . X . . .|
$$ | . . . . . . .|
$$ | X X X X X X X|
$$ | O O O O O O O|
$$ | . . . O . . .|
$$ | . . . . . . .|
$$ ----------------[/go]

The board parity is odd, the seki parity is even, they are unequal as usual on the 19x19 board. With the standard area komi 7.5, the possible scores would be -4.5, -2.5, -0.5 (smallest white win), 1.5 (smallest black win), 3.5, 5.5,... points so in the example -0.5 (smallest white win). Decreasing the komi by 0.5 to have the komi 7 changes the score to 0 (jigo).

I frequently play under area scoring with 7 komi and jigos occur regularly, as they should. If the komi were 8, jigos would be rare because then an odd seki parity is needed.

Tournament scheduling would have to change to make room for possible draws. For example you can no longer have a final scheduled the day after the last semi-final because you might need another game in the semi. Draws would be rare, but draws being possible at all would have too much impact on scheduling imo. I don't think that's worth it, even if it's slightly more balanced. Go is hard enough that the better player should win anyway, despite one player starting with a 0.3 point advantage. Just look at SJS' win streak.

Komi is 6 in a lot of Slovenian tournaments including the national championship. The reasons being something along the lines of "it's become a tradition" and "it's something different/interesting", I wasn't around when they decided this so I don't know what the original reason was. I've never been a fan but it doesn't really make that much of a difference, even though the occasional jigo does happen (we don't have any knock-out tournaments).

At the end of the day, I think I'm in favour of whatever the correct komi turns out to be, if AI decides it's an integer, I'm fine with that and if 0,5 is needed that's fine too.

1/2 points don't really cause problems for any competent pairing program. If you want to use integer komi I don't see any problem.

_________________
North Lecale

Indeed. Only tournament formats that have some form of elimination require tie-breakers.

I'd prefer integer komi in all other cases.

Under area scoring, 7 komi plus button is best and probably no less balanced than komi 7 without button. It's equivalent to territory scoring with 6.5 komi.

Different kinds of buttons have been suggested so you must specify your type. Equivalence (of what kind?) should be proved.

6.5 komi is fairest for territory scoring, and these together should alway be used in elimination tournaments or swiss tournaments with as many or less games than what a supergroup or top group have if there was a knockout between them.

7 komi is fairest for area scoring, and these should be used for all non-elimination tournaments with more rounds than would be in an elimination tournament.

Let Black and White be perfect players trying to maximize score. Then the komi should be (Black's points)-(White's points), which is an integer, probably 7.

This is is the usual explanation given for fractional komi. But I have never found it completely convincing. Chess and checkers (i.e., draughts) games often end in draws, and yet tournaments seem able to cope with them. What is different about go?

A story, or rumour, that I have heard is that scheduling was not the only reason for fractional komi. Perhaps not even the main reason. Supposedly, back in its early days, the Nihon Kiin was worried that players might collude or conspire to engineer a draw, in order to split the game fee evenly between themselves. Whether there was any evidence of this happening, or if instead they just wanted to prevent this problem from developing in the first place, was not clear in the story. In any case, the Nihon Kiin introduced fractional komi to prevent draws, and thereby prevent this potential collusion. The effect was that the players would instead each try their best to win, since a draw was impossible.

It sounds plausible. But I have never known if there is any truth to it.

_________________
And the go-fever which is more real than many doctors’ diseases, waked and raged...
- Rudyard Kipling, "The Light That Failed" (1891)

First, it is simply a matter of preference for either integer or non-integer scores.

Now, let us ignore this preference and consider amateur tournaments. Tournament organisers are either capable or not capable of organising tournaments with predictable schedules; overtime rules make this easier or harder. Let me now presume skillful organisers and reasonable time settings for the ordinary games. Next, any tiebreak games can have ordinary or blitz thinking times, but in amateur tournaments with typical weekend schedules and players having booked accomodation and tickets, only blitz / rapid are feasible for any tiebreak games.

Chess or checkers can use 5' SD for blitz tiebreaking and it will not be a problem. Go, however, wants at least 10' SD blitz but serious (championship) tournaments, such as the German Championship with 20' SD, prefer more and two KO tiebreaking rounds can be necessary. Together with some organisation (determining the results and pairings etc.) and rest, this amounts to circa 1 hour per KO round even with reasonable organisation. If the ordinary tournament was supposed to allow players taking a train at about 18:00, then they need to take some around 20:00. For national tickets, this is a major difference because train (or bus) connections during the late evening and night are limited. In Germany, train tickets at reasonable prices must often be booked in advanced. Given the possibility of a tiebreaking KO and the huge price differences from train to train, one must book a train around 20:00 or later. In practice, this means book a train at 20:00 to 23:00 and wait for 2 to 5 hours at the station just because there is the possibility of a tiebreak KO. Instead with tied winners, at most circa 2 hours waiting was the norm. When also tickets could be bought at the station for ordinary prices, 1 hour waiting was the norm - nowadays, such tickets are way too expensive (imagine, say, €250 instead of €70) for quite a few players, especially those without expensive discount ticket cards. Typical buses are often available at around 3 am to my city and can take up ca. 12h of travelling so are even much less suitable.

For one, tournament structure. Go tournaments, as far as I know, often have elimination brackets (e.g. https://homepages.cwi.nl/~aeb/go/games/ ... index.html) whereas chess tournaments pretty much never do. The frequency of draws in chess make this kind of structure impossible; you'd have a lot of tie-breaks every round, and you have to wait for them to finish before starting the next round.

If we want to introduce the possibility of draws in go, elimination tournament structures would need to change. Afaik round robin and swiss style tournaments, which are the norm in chess, are already common in go as well. They would become the norm. That seems fine, I guess.

Another thought I had is that in round robin and swiss formats, with draws being rare, getting half a point is worse than if draws are common. It would feel almost like a loss for both players. That half point is more of a tie-breaker than anything else; it might as well be 0.1 points because you're probably not going to draw twice in one event anyway. But I'm not sure if this is a good thought. I haven't been able to explain to myself how that makes the tournament less fair or worse in any way. It just feels wrong.

I suppose it depends on which region and which go sub-population you're focusing on.
Elimination brackets in pro tournaments do seem to be more common nowadays than they were a couple of decades ago.
But most (amateur) EGF tournaments use McMahon with some additional tie-break scores (such as SOS and SODOS), which is perfectly capable of dealing with draws. And some EGF tournaments have indeed used integer komi in the past, like the London Open in the 90s (6 komi under Japanese rules IIRC).


I remember that in one of those London Open tournaments in the 90s, there was a controversy that 2 top players from Russia allegedly colluded to make a draw in the last round, which maximised their combined prize money. Perhaps a careful structuring of prize money might work to avoid such issues?

For no real purpose. Direct comparison, drawing lots, letting them play 5' SD or previous ratings would be the honest tiebreakers. However, money prizes can be shared easily so there is no need to break ties. I have never understood this determination to split hairs among equal performers and have sometimes suffered dearly when having beaten the tiebreak winner with the same MMS or Swiss score.

When it comes to determining the winnner of a McMahon tournament, a McMahon tournament would usually have a top group with a size that is suitable for the number of rounds (IIUC), so there is likely a unique player with the highest number of wins in the top group and tie-breakers aren't needed (at least for 1st place).

I don't know how often integer komi would thwart this, but I guess organisers could could still ignore tie-breakers and share prize money in such cases, if there is no time for an additional play-off to break ties and there isn't really a strong requirement to have a unique winner.

We may wish. In practice, many German or European McMahon tournaments have 5 rounds to fit well into a weekend and the top group is chosen so that 6 rounds would be the most suitable (or in some very small tournaments, 4 rounds.) Many people, including me, like 5 rounds (although I would also like 6). It is just that the other conditions and tournament director / pairings manager do not necessarily enable a unique top scorer then.

I mean the standard button, which is the following, as far as I know:

The button is a special token which is placed next to the board at the start of the game. On your turn, if neither player has taken the button yet, you may not pass, but you may take the button instead of making a board play. At the end of the game, a half point is added to the score of the player who has taken the button. When checking for repetitions, two identical board positions are still considered different if the button had already been taken in one of them but not in the other.