Komi needs correction on 9 x 9


#21

Your comments actually agree with two of my points. I completely agree that DDK players (or perhaps we should say high DDK players) do not have the skill to take advantage of the first move. That is the core of my main argument. However, it can be applied up the line as well, I think, though no one knows where it may stop. In other words, pros have more ability to take advantage of the first move than do SDKs, etc.

Yes, I’ve previously seen the argument that DDK games hinge more on blunders than on komi. In a sense that supports what I’m saying: that pro-derived komi can’t reasonably be applied to DDK games. What is more important, is that it is a poor argument for saddling someone with what is in effect a handicap from the beginning. If we know they can’t take advantage of the first move, then we are knowingly handicapping them.


#22

Two points on this conclusion.

Firstly, “know they can’t take advantage” is very strong language. I would say it is less likely that they will reap its full benefit. Certainly, there is still some advantage to going first.

Secondly, as far as “we are knowingly handicapping them” goes, again, it is only 6.5 points. Very few games have results this close. Furthermore, DDK ranks are much less precise than higher ranks as it is so easy to learn simple parts of the game in as little as a few hours that can boost at times a player by several ranks. Due to this, I find ACTUAL DDK handicap games in which stones are awarded to a player to be of more concern than the seemingly inconsequential komi.

Even so, I wouldn’t advocate changing this. SImply put, the idea of a “perfectly even DDK game” is a joke and impossible to code for, there is simply too much volatility and as you say, decided more so by blunders than anything else. Trying to cater to them at this level will leave them unprepared when they get to higher levels. Better for them to have consistent rules throughout, and to just acknowledge their presence when it doesn’t matter so that when they are skilled enough for it to be significant, they already know and adjust for them naturally within their play.


#23

I’m not advocating for any change, as I previously stated. Also, the quote, “perfectly even DDK game,” does not come from me—I’ve never said or imagined such an absurdity.

I don’t really disagree with most of what what you say in this reply, because you are speaking mainly in practical terms, while my interests are more theoretical. “Reap the full benefit” is a more precise phrasing—thank you. I am, however, disturbed by the “it is only” argument, which is weak in almost all contexts (and may even fall into one of the many categories of fallacy—I can’t remember): for example, “it was only a punch in the arm.”

It would be interesting to see numbers on how close games are, at DDK as well as up the line. However, a komi that is too large can have psychological effects that haven’t been considered in this thread. I have often heard players speak of the need for black to play more aggressively in 9x9 because of the large komi, and personally speaking, I feel this way too. Of course, we may well be wrong, but the psychological influence is real, nevertheless.


#24

It’s hard to get and analyze all the games correctly. After all, indeed strong players can evaluate the score during the game and instead of peacefully losing by four points start doing crazy stuff. Plus many players just resign even if it’s just four-five points, so we wouldn’t be able to count that.

But I wonder which color people prefer to play. I haven’t played 9x9 in a long time, but I feel much more confident with Black. When I play as White I feel like I am just defending trying to live under Black’s attacks. And if I’m Black I’m controlling how the game goes.


#26

most people have a losing percentage as black on all board sizes because of being assigned black as the weaker player combined with the fact that no one seems to want to use handicap on ogs


#27

If Komi is set right, you’d think the “assign black to weaker player” would be nonsense, wouldn’t you?

It seems to have other undesireable effects too, like getting used to playing first if you’re playing a stronger player. Wouldn’t we all be best served to get an even mix?


#28

I think one reasonable way to solve this issue is to find the statistic of win rate of 9x9 on this site, If white or black has win rate significantly above 50% then we adjust the komi. Then we monitor the win rate and adjust again if needed.


#29

@q4w56 if you had read the rest of this topic you would have found several arguments for why it isn’t quite that simple.


#30

Yes, i’m sure the entirety of the go world’s professionals are wrong about what komi should be, and some data from one amateur server will show everyone the light


#31

@BHydden The thread is too long to follow, can you point out what are the arguments against deciding komi by win rate of games by equal ranks players on 9x9?


#32

I would have to read through the thread myself to find all the examples mentioned but i believe the most recent was that often the weaker player gets black so you would expect them to lose some arbitrarily higher percentage of games. Another that comes to mind is that anyone who plays on multiple servers might be thrown by OGS being the only one that has a different value to the world standard. I believe there are others mentioned also.


#33

What is the world standard then? As far as I know from past pro 9x9 tourneys, it has always been 5.5 and that hasn’t changed since either.


#34

As a DDk, and a “weaker” player I totally think 5.5 is way too high. I have noticed that I am dealt the black role almost every time, but what am I supposed to do with that? It doesnt help the weaker player very much if he is offered the first move if he doesnt know how to utilize it.

I personally find the 5.5 points really hard to overcome.
Thank you.


#35

Are you advocating for a rank-specific komi?

Wouldn’t that be what handicap is for?


#36

If you look at some old posts in this thread and in other threads, you will find that I have expressed similar sentiments in the past. So I strongly sympathize with your viewpoint. The first move is not worth komi for weaker DDKs, or maybe for most DDKs, because they don’t know how to make the best use of the first-move advantage. Komi was derived from analysis of pro games, not DDK games. I think it is clear that there is probably a sliding scale of appropriate komi based on one’s rank. Trouble is, no one know what that scale is. It would be a nice research project for a graduate student in statistics. As you get stronger, you may find yourself starting to think that komi is too small, or so I am told—at the moment I am still on your side of the dividing line, though I think I am feeling the ground beginning to shift beneath me.


#37

From my experience on the 9x9 board 5.5 seems fine to me for the time being, my program suggests that the correct komi should be 5 or 7 for perfect play. https://www.suomigo.net/wiki/Fuseki9x9 with over 130k games from goquest database using komi 7.


#38

There’s a world of difference between 5 and 7! Which would you chose?


#39

Didn’t they state above that the handicap feature is virtually never used?
New players, who are mostly westerners, I can imagine, will ignore the handicap system while learning the basics because its too exotic.
It’s really outlandish. We come from a totally different school of thought than Easterners. So, perhaps in Asia, they can say: “No, you have an advantage, you just don’t know how to use it” or “go do your homework instead” but people learn in a different way over here. Learning the basics and throwing in the “handicap” system is extremely difficult.

Mostly, I feel that this is a cultural thing, in the East they blame the weaker player :grinning: as we blame the object. :grinning:
I speculate that with time igo will evolve (in the west) to reduce komi for new players until they can become stronger, and the handicap system will be forgotten about, as it should :grinning: I mean, show it to a new player. They will get extremely annoyed! Its so weird. Okey, wall of text, end of story. :slightly_smiling_face:


#40

Hard to say at the moment current statistics typical point at 7 komi, but I would say that white does have the advantage in the majority of positions I have analyzed over the years. Currently I am seeing if that advantage is enough to stand true for all openings as I continue my research.


#41

My condolences on the earth quake.
Jokes aside; the Elo system, I understand tells how strong a player is? Well it’s just that 5.5 is a mountain to climb if I play others of my level.