Traditional ratings use material advantage to align grades, in that they measure one's
ability to overcome a n-move advantage. Japanese rules standardised the placement of these first moves when they were two or more.
As most are aware of, aligning win
confidences to material advantage is tricky, as the same winning confidence difference requires less material advantage. 90% confidence in the dan range may only require a few handincap stones, but for a beginner it requires many. So rating systems such as the one used by the EGF take this into account, but it still seems to involve a little judgment of arbitrary values.
So over the past few days, I've wondered about two things relating to this. The first is that there maybe should be a division between grades generated from material advantage, usually derived through over-the-board play, and confidence grades, as used in online servers, and achievement dans, given to amateurs in China and South Korea and professionals in all organisations with professional players. I'll call them
Abililty kyu-dans,
Confidence kyu-dans, and
Diploma dans, because I can't think of any good names.
The second is that in purely confidence-based ratings, it may be reasonable to use large confidence bands for grades among weaker players. Mamumamu0413 uses 75%, and that is among pros. I thought about 90% for amateurs, which just doubles the width of the bands (the 75% value is based on powers of three, and 90% on powers of nine, which is three squared). The reason is because I am not aware of martial arts with more than a dozen kyu grades—I only know of them having at most just over half. As most know, ranks for weaker players are unstable, so it doesn't make a lot of sense to worry about small differences, and changing grades would still occur quickly due to it being easier to improve. But one may also avoid online-go anxiety online if your grade is barely at stake for any game. Maybe it will also encourage games between players of different strengths. It may even be more simple and relatable for newcomers.
The third is that it may be good to derive confidence rating ratings from random play via bots. For example, random play could be rated at
0, and the bot that wins against random play by 90% will have a rating of
1. One could even go into negative ratings by programming bots to be worse than random. It's probaly best to use multiple bots; Dennis Hassabis said that Alphago tends to do better against versions of itself than other players of similar strength, which may also be true of others bots. Something
approaching an absolute rating table might be achieved—of course, not outright, for reasons reasons including the fact that bots are not quite an accurate way to represent human play, but this text is already too long.
After this, we could experiment with different handicap stones and apply then en masse to the bots at different ratings to generate a table of handicaps for each difference in confidence.
edit: another silly idea I hadn't time to write, but if 90% win ratios are used from 1 dan, one might have a few kyus—say
1 to
5as brown band, red band, blue band, green band and yellow band with the sixth kyu being white belt (bands instead of belts, what with igo being a mindsport were one uses the head and often the hands. This is getting sillier by the sentence).
Except, six kyu would be a white band with a black stripe. My memory recalls that in some martial arts, 1 dan is represented by a black belt with a white stripe. The black belt of one set of rankings is the white belt of another*. This double sets becomes a triple set with the go bands. So a a beginner may start anywhere within the Junior kyus, from a band solid white at 12 kyu to those white with and middle striped up to six. Then the colours repeat themselves in solid mode kyu's five to one, then at
1 dan you have a black band with a white stripe, changing the colours of the stripes up
6 dan.
By the level of
7 dan with a solid black belt, it should be too much to use winning confidences of 90% as a form of demarcation if they should be used at all, so ending it here seems best. it may be worth noting that if seven dan aligned with the EGF's seven dan, one dan may be near many Japanese club's one dan.
*
Life begins at 40, the end of the beginning and the beginning of the end, rebirth at—ah, the concept should be clear by now.
, she is forced to fill an eye and leave one group with only one liberty.
), then whoever has the most stones on the board wins. (Ignoring komi)