Calculation of rank, difference in points after counting

Why is it unusual to take into account the “amount” of victory, i.e. the difference in points after counting, when calculating the rank in Go?


How do you take resignation into account?

I can think of 3 ways it could go

  1. Resignation is treated like a big loss. The loser now only resigns when the difference is very high, prolonging games within margin.
  2. Resignation counts like a small loss. The looser now will resign instead of passing to prevent a rating loss.
  3. The rule is rather complicated to prevent 1) and 2). In this case the rule is more complicated than needed to get good results.

What does the difference in points tell you? I don’t think it tells you anything about the relative strength of the players.
A stronger player can ensure a win by a small margin rather than risk a loss chasing a bigger point difference win.
Or a game between two DDKs might finish with a margin of a few points or a massive margin if a big group dies but either way it doesn’t tell you that the winner was much better in one game or the other.


Actually there is a go variant that takes the point difference into account for the game result:

1 Like

That seems interesting as a variant, though is there a reason to prefer the step-wise linear relationship to a linear relationship? 10 points + margin to the winner (max 50) and 100 - winner’s points to the loser would be simpler and encourage fighting for every last point, no?

(I would prefer some numbers that are nice and even in dozenal instead of decimal, but that’s immaterial to the basic concept)

Well, you can also ban resignation altogether so each game has a complete result. (ofc then you have the same problem with timeouts)
But I’ve considered something that fits along the lines of 3 for a team tournament where points matter (and also which boards you get wins on matters)

Not 100% true. It doesn’t tell you much under the version of the game where the .5 point that wins the game is all that matters, but if all players are going for the best result they can, it makes sense, and there are models to handle that, notably in the first sections of this paper by Mark Glickman.

The problem with the usual models, of course, is that it assumes if team A against B has average score difference Y1 and team B against team C has average score difference Y2 then A against C has average score difference Y1+Y2, which I believe doesn’t make sense, especially in a game with a maximum score difference. Fixing such a thing likely requires a lot more complications, and I’m not aware of a proper state-based system that even covers the basic case.

It is a thing worth considering in many cases, but many amateur go players want to at least mimic the professional world – if not aspire to be part of it. And the professional world only cares about the win, so naturally we replicate that in the ways that we play and the ratings systems we use.


I don’t think this is at all a bad thing; I think a prerequisite of any successful sport is the ability of viewers to go out after watching the professional game and play themselves. Soccer, Tennis, Table Tennis, DotA 2, SCII, and Go all fit this criteria and are very popular. I don’t think it’s the only factor, but I think it’s necessary. Virtually noone watching the professional games will have the ability to play that well, but the ability to go out and strive to emulate the plays they just inspired by is integral to being a successful spectator sport, I think.

I previously discussed this here:

Ultimately, caring about the margin of victory/defeat significantly changes the nature of the game. It would be a different game, essentially a go variant, and you would need to decide how to handle resigns and timeouts.


Yes, I 100% agree; though as a variant it does sound interesting, especially for a juubango-type format.

1 Like

not for all players, the strategy is still quite similar, although very likely with less counting and a different mindset to risk-taking.
But you are right that for a good number of players, especially those looking at imitating or becoming a part of pro play (like I described above), this would be a significant change to the incentives structure.

1 Like

That’s bad news for teaching, if you cannot try to win by a small margin anymore, and have to kill everything on 9 stones handicap.

you don’t typically play ranked in a teaching game…

But you teach to play ranked, no?

Not necessarily… I teach to pass on skills and experiences that I have and others desire… what they use that teaching for, if anything, is largely irrelevant


It’s a can of worms. Because this way player that lost could still rank up because of small margin of losing, couldn’t they. We’d be essentially redefining victory as winning by X points given Y rank difference.

That’s the same as incorporating material difference at the end of the game in chess.

And it would discourage risk taking which is bad for Western go, Western players are very soft as it is.

If you want to change anything, adopt the “captures worth double” rule, that would make the game way more interesting.

Point difference only matters when you play for money and OGS doesn’t want that.

I highly doubt that. It would discourage certain risks that you might take when you’re losing. It will encourage them in comparison when you’re winning. But all risks you take are then calculated by (if calculated at all) a cost-benefit analysis in direct points rather than win%s.

1 Like

I still prefer the idea that a win is a win, whatever how many points, if by resigning or by timing out.
I like more close, balanced games as big margin victories. In the latter case I would prefer players learn to resign as taking in account the size of the difference of scores.


Absolutely, I am only in favor of this as a variant, absolutely not as the norm.

1 Like

When go only have 2 possible results; win and loss, (lets not bother with jigos now), one part of the strategy is maintaining your lead to the finish without screwing up everything. “Winning a won game”

Avoiding stupid risks is part of important skills of the game, not a weakness as some tend to think it.


Beyond these points, this affects the game deeply enough to splinter the rankings between OGS and elsewhere even more than they already are. Let’s just…not.

1 Like