Go komi Rules


Go komi / Rules

Back to forum

Lionel Vidal    (2006-05-31)
Go komi / Rules

Hello to all,

As a brand user FICGS user, I first read the rules (of course!), and I am wondering about the komi value of half a point. AFAIK it is not very common, unlike, say, 5.5. Is there a specific reason for such a komi value?

Best wishes, Lionel

Lionel Vidal    (2006-05-31 19:35:58)
Go komi

Oops! sorry. I meant 'brand new user'
Best wishes, Lionel

Thibault de Vassal    (2006-05-31 20:01:30)
Weiqi komi

Hello and welcome Lionel :)

That's a very... good question !

Actually I thought it was the more 'organic' (like this game) way. But it seems to be a large debate, without a clear response.

I suggest all go players to read this article : http://senseis.xmp.net/?Komi

Feel free to give me your opinion about that. Thanks in advance.

Lionel Vidal    (2006-05-31 21:15:10)
Weiqi komi


Thanks for that link: it is indeed a good reading. I would suggest also to read the chapter on the rules of Go in 'The go player's almanac' (Bozulich 2001), that gives much more details.

By reading these, you can see that half a point komi is indeed quite uncommon :-)
But an important point is that it should probably depend on the chosen rule, or more precisely on the counting mode, area or territory.
BTW, what is the counting rules here in FICGS ?
Note that this rule question may be quite critical on some life-and-death situation and may change the result of a game! And some of the simplest cases (like bent four in the corner) can arise not too uncommonly.


Thibault de Vassal    (2006-06-01 04:10:56)
Komi, pass, area scoring, ko, seki....

Ok... I'll bring answers and modifications to the rules for all points that remain blurred, today and tomorrow !

The counting method will be Area Scoring (chinese scoring) : When alternate play has ended, each player's score is the number of his stones on the board plus the number of empty intersections surrounded only by his stones.

Now I have a question from a beginner (me) : Is it possible, even with a non entire number komi (ie. 0.5 or 5.5), to have a draw situation ? In other words, what happens if a game can't finish because in example ko rule can't apply ?!

Thanks a lot for your help.

Lionel Vidal    (2006-06-01 12:46:14)
Go rules


Even if a well chosen komi rules out scoring draws, you still have the problem of infinite repetition.
Traditionally, a game in which a triple ko or other infinitely repeating position arises is annulled, or treated as draw or replayed.
Some rules (New Zealnd, USA, SST) deal with these positions by stating that a player cannot make exactly the same position on the whole board twice. (Note that it is then not always easy without computer assistance to determine the legality of a move).
The chinese rule (at least the 1988 official one) also forbids reappearance of the same board position, but in some situation this is not enough to prevent a draw: in some cases neither player want to start a sequence and keep passing to avoid solving a situation at their disadvantage. (and to forbid passes triggers others worse problems...).

As you see, the situation is quite complex, and while rules of Go seem simple, their precise definition is not easy. In practice, you eventually have to rely on the sportmanship of the players or on a referee decision.

Personally I have played till now only under the japanese rule, and in case of problems (very rare as this rule is quite detailled, but then rather complex in its exceptions handling), a senior player says the truth, and, at least in Japan, this truth is undisputed and becomes the laws :-)... quite simple!


Thibault de Vassal    (2006-06-01 13:22:24)
Infinite repetition

Couldn't the problem of infinite repetition be solved by giving the full point to White (the 2nd player, not like chess) ? It could be a compensation added to komi, as it can force black to give such stones that provoke repetition !?

But can White force easily such a situation with many stones....... ?

Lionel Vidal    (2006-06-01 15:02:56)
Infinite repetition

Well, it might work, but that seems unfair to black IMO, because it creates discrepancies in the rules depending on you being sente or gote.
Admitedly, the komi can be seen as such discrepancy, but all it does is forcing sente to be bit more aggresive, the stategy and tactics, the feeling of the game being globally the same. What you propose would induce a kind of strategic play on the rules... not really Go anymore!
I realise I may not be clear: as an extreme example of rules discrepanies, just consider Renju. Here the rules are different for sente and gote and the strategy is indeed really different for both! While this is fine in Renju, because actually it became the heart of the game, I do not think Go needs it.
Besides, it would spoil somehow IMO the aesthetic feeling of the game flow.
Just my opinions of course :-)

Thibault de Vassal    (2006-06-01 21:25:04)

It is very clear Lionel.

In another hand, each tournament rules and generally each situation influence the strategy at chess (so other games). And FICGS chess wch rules are special ones in the knockout tournament that should avoid draws. Actually, only a "one game match" can have no influence on 'the game'. (not perfectly true, as the player's strength is another factor)

Rules are flexible, particularly for the game of Go, so I think we can use even uncommom ones, if it is balanced enough (= there's still a challenge). Do you have an idea about this rule avoiding repetition, how many stones or komi it could be worth ?

Another question : Are there situations that look like zugzwang in Go (where the best move could be 'passing') ?

Lionel Vidal    (2006-06-01 22:17:30)
Go rules

I don't quite understand what kind of problems with draws remains with FICGS (sic!) rules: as passing is not allowed, if you add the non repetition of the same whole board position and a non integer komi, I do not see how a draw is still possible.

BUT... this solution does actually not solve anything as I don't think you can forbid passing (as a matter of fact, I checked the official japanese, chinese, new-zealand, AGA (USA) and SST rules: pass is allowed and needed)
The main reason is, IMO, that you need a legal way to end the game (double pass). And yes there are situations where the best move for BOTH players is NOT to move at all in the area: the simplest case I can think of is thousand-year kos, which in the case of japanese rule usually end in seki.
Note that a single pass (that is the game goes on after it) can change the difference in scores in area mode: the AGA rule introduces the concept of pass stone to compensate and insists on white making the last move (if necessary with an additional pass and pass stone) to ensure that the total number of stones played by the two players are equal!
(BTW this is one of the reasons, admitedly far behind familiarity, why I prefer the japanese rule in face to face go)

To sum things up (!!), while I agree that FICGS could develop its own set of rules, I feel that the subject is too complex and error-prone, and has been long, and still is, disputed by highly competent authorities : why not use the result of their work?
I would add that the point of all this is rather moot if you consider that situation like triple ko and alii are indeed rather rare: why not stricly stick to, say, the official chinese rule, and replay the game by referee decision in the rare cases where neither playing side will yield?

oh, but I could also check what they do in the kiseido server ?!?
oh, and do take what I say with great caution: I don't feel and I am certainly not competent enough on the subject! Any other advice over there? :-)

Lionel Vidal    (2006-06-01 22:21:52)

It seems that the postings are sometimes not in chronological order ? (see the last two posts)

Thibault de Vassal    (2006-06-02 12:24:54)
Weiqi : FICGS rules

Thanks for your explanations Lionel. (forum bug is fixed)

I've changed the rules. I would like some opinions before I announce it :

First, now you can pass, just entering 'pass'... Special rules in FICGS are : Suicide of more than one stone is authorized, and infinite repetition means a win (full point) for White. Both players must play until one resign, both players pass (then call referee) or game is adjudicated. Scoring method is area scoring with chinese counting. Eyes in seki count as territory.

Thanks in advance !

Lionel Vidal    (2006-06-02 16:28:11)
Go rules

It's me again :-)

What is the point of the special cases you chose? Why not simply follow the chinese rule? I reread it yesterday and compared to what you say:
- reappearance of the same board position is forbidden (note that should be easy to check by computer with hash keys associated to positions)
- Seki is not really a special case in chinese rule (it is only in territory scoring): you count stones and enclosed vacant points; others vacant points are share equally.
- Winner is determined by comparing one's score to 180 1/2 (half number of points of the board). - Komi: 2 3/4 points are deducted from black's score and added to white's. - After both sides have agreed to end the game (that is after a double pass), if any unsettled positions remain on the board, both sides' stones are treated as alive (that is neat and solve most drawing problems) - Basically a player that makes an illegal move loses his turn (i.e. in effect passes): that includes repeating the same position (why should white win in such a case?).

That sounds much cleaner IMO.
The only possible draw may be some very complex round robin kos, where the position keeps changing, but I guess we can forget it (and it should eventually been resoved by double pass anyway, even if one side is unhappy: see the preceding neat point).
BTW you can probably find the full text on the Web (I have only a paper version from the 1988 official rules of Chinese Weiqi Association).

Thibault de Vassal    (2006-06-02 17:48:34)
Go rules

Hello Lionel. I just read the 1988 version of the official rules of the Chinese Weiqi Association.

The point here is to play with the most interesting & fair rules, not 'official' ones or others if it could be improved...

Note that FICGS chess rules have a peculiarity : 50 moves rules isn't applied if the mate can be forced. FICGS chess world championship rules are not (of course) the rules used by FIDE. I spent much time thinking about rules which are IMO the best thing in this server and I think most players will appreciate these points.

I think avoiding draws in Go is interesting because energy consuming could be too different in some games and lead to unfair situations in tournaments.

Hash keys don't solve all problems, 'superko' situations could remain as draw, furthermore these special rules could avoid any ambiguity. It is clear, it brokes 'symmetry' and I feel it is fair enough.

Then, rules exist to be enforced ! :) .. More seriously, I'm not convinced these new rules don't make sense, even if it needs adjustments. Still inquiring, but unless I find (or you convince me :)) a solid argument in another way, I think I'll apply them.

Lionel Vidal    (2006-06-02 17:58:20)
Go rules

That is ok for me: you rule :-)

And anyway, the domain of application of these special rules is indeed quite narrow!

Just a question to better understand your choices: what is the point of seki special scoring rule?

Thibault de Vassal    (2006-06-02 18:07:39)
Seki rule

About the seki rule, nothing special : actually it is Chinese rules.


Lionel Vidal    (2006-06-02 18:12:34)
Hash keys and draws

Just to be sure I was clear: the aim of hash keys is just to enforce, I should really say to program, the rule of non repetition of positions; i.e. kos or super kos are irrelevant: if the resulting position occured before, then a move is not legal.
Then I am not sure how a draw could happen: let's say there is a superko running somewhere; either both players eventually pass, letting the situation unsettled, with the unsettled stones declared alive.... or they play the kos and sooner or later (ok, later in case of superko!) the position will repeat, hence a forced pass, or be solved...
Do you have an example of situation that will not eventually lead to a repetition?

Thibault de Vassal    (2006-06-02 20:39:25)
FICGS : Chinese rules !

You convinced me :)

(sorry, I didn't understand well some concepts before)

Good news : I have implemented the non-repetition rules with hash keys...

Now the rules are exactly CHINESE RULES with a 6.5 komi, which is the value which tends to be used in tournaments.

Thank you Lionel :-)

Thibault de Vassal    (2006-06-02 20:50:09)
Positional superko

The superko rule says that it is forbidden to repeat a previous board position. There are two versions of this: 'positional superko', where it is forbidden to repeat a board position, and 'situational superko', where it is only forbidden to repeat a board position with the same player to move.

FICGS now uses the positional superko rule.

Lionel Vidal    (2006-06-03 09:50:52)
Chinese rules

After reading the new rules (nice :-)), I have two suggestions:
- maybe you should cleary state that repetition of position leads to a loss (or does it? it is the way I understand the wording, but either way is possible), which is not the case in chinese rules (AFAIK at least in some versions, you just loose your turn).
- maybe you should announce somewhere (not in the rules of course) whether the already running games have to follow the new rules. Knowing the applied komi is important, even in fuseki.

Now let's try them :-)

Thibault de Vassal    (2006-06-03 14:05:02)
Repetition of position

Repetition of position doesn't lead to a loss, as it's now impossible to make a move that provokes such a situation. (message would be 'Incorrect move'). So the pure Chinese rules apply.

The change of rules have been announced in the news (page 'My messages'). Everyone is supposed to read it, as it's the first page appearing when you log in.


Thibault de Vassal    (2006-06-06 17:39:44)
7.5 komi + Superko

Hello to all.

In accordance with Chinese rules, the last update of the FICGS rules of Go states a 7.5 komi (3.75 by chinese counting)


The last ambiguity may concern the superko rule, as now FICGS graphical interface forbids any board repetition (not only previous move). See this page on BRITGO site, at the very bottom : Positional superko (PSK) means a play may not recreate a previous board position from the game, referring to the position just after the play and consequent removals.

However, forbidding any board repetition is the only way to prevent draw games.