Discussing format changes for the OGS July Team League

This thread is going to make reference to previous discussion in OGS team league proposal – see also OGS June Team League

This is a place to propose and discuss changes to the format of the OGS Team League, if and when we play another edition next month.

  1. Players / team. My initial idea was to have a mixed league of three-player and five-player teams, in which the larger teams would only field three players in match-ups against the smaller ones. What actually happened is that we began with a proposed universal three-player format and, with increased registration, developed it into a four-player one.
  • The main options: a. 3 players b. 4 players c. 5 players d. 3 / 4-player mix e. 3 / 5-player mix f. 4 / 5-player mix g. 3 / 4 / 5-player mix. I doubt we’ll have enough registrants to consider six-player teams.
  1. Round duration. In the first vote, there was a narrow victory of two weeks (9 votes) over three weeks (8 votes). With the increase of registrants in the second-stage registration, I decided that two weeks was an inappropriately short time and shifted it to three. However, I think that two weeks does have to be discussed again. I personally find three weeks to be an awkward compromise between a fortnight and a month, so I favour a month. One week would certainly be too short, and I suspect round times over a month would be unwieldy. Note that the original poll specified playing only three games.
  • The main options: a. two weeks b. three weeks c. a month
  1. Game length. In the poll, 30m main time (7 votes) convincely triumphed over 45m (3 votes) and 60m (2 votes). Everyone seems fairly happy with 30m, and I wouldn’t like main time to be any shorter. What I’d like to do is rerun the poll with only 30m and 45m, so as to avoid splitting the > 30m vote. In the overtime poll, 5 x 30s won with 7 votes and longer time controls collectively received 5. Similarly, I want to compress those options into a single longer option, which for the sake of argument can be 10 x 30s.
  • The main options: a. 30m + 5 x 30s b. 30m + 10 x 30s c. 45m + 10 x 30s
  1. Ruleset. My suggestion is that both Japanese and Chinese rules, and indeed all rulesets, will be permitted; but that either player should have the right to insist on playing under Japanese rules. I wouldn’t like mandatory Chinese rules, but if the people demand it…
  • The main options: a. mandatory Japanese b. default Japanese, optional Chinese & others c. default Chinese, optional Japanese & others
  1. Komi in handicap games. The current rule is that in handicap games White will receive either 0 or 0.5 komi, with both being valid. Should we standardise this?
  • The main options: a. 0, b. 0.5 c. either
  1. Handicap. This is perhaps the most controversial element of the rules. A dense discussion on the subject took place, after which I took an executive decision to use what I call H = ΔR / 2, which is to say that one stone is given for every two ranks. We simply needed to come up with a rule so that the league could begin. All handicap options are still on the table – anything from a stone / rank to reverse komi to no handicap at all – but the proposed system must be simple and practical. If I can suggest a few basic ideas:
  • Some options: a. 1 stone / rank b. 1 stone / 2 ranks, c. reverse komi of around 6 or 7 points / rank d. no handicap

PS. My judgement for this month is that free placement is optional, but that White has the right to insist on standardised Japanese placement. I’d like free placement to always be available, which is something for discussion.

  1. Rank balancing. This is another of the more difficult aspects. It can’t be denied that some of the player matchups are unbalanced to the tune of four ranks or sometimes much more. I’d ideally like to try to avoid these unbalanced matchups, but they’re a function of our current level of participation, especially that it seems to have been hard to recruit DDKs. Some suggestions for guidelines have been offered, but I think most people would agree that it’s desirable for the top board to be occupied by dans, or at least 1- or 2k players; and the bottom board by DDKs, for at least 8- or 9ks.

We’ve got a week in which to hold conversation on these points – I’ll probably poll on the 20th and close on the 22nd, proceeding to registration.

What do you all think?

2 Likes

Although I’m enjoying the theme, I’ll politely bow out for the next one if last table has such big rank difference again. But I’ll follow and cheer all the same. :slight_smile:

About handicap, I personally find it much easier, fulfilling and also more educating to play with reverse komi. I think people get carried away with finding the absolutely exact number that would be appropriate, though, instead of focusing on a playable system, so we can do the thing and have fun. I don’t find it helpful or considerate, I find it outside the scope of the discussion.

There was a nice discussion that led somewhere here Go (Baduk, Weiqi) Players on Facebook : What would appropriate negative Komi be if I (10k) play a 25k who does not want to take Handicap

tl;dr 1
even if not absolutely accurate, this chart could be useful

tl;dr 2
“Here in western Japan, we represent a step of one dan or one kyu as one handicap stone on the board or six points or some combination of the two. This means that the case in the post the 15 kyu difference would yield a 15 times 6 or a 90 point handicap if no black stones were placed on the board.”

If it’s stones and not reverse komi, I guess the standard 1 rank=1 stone would be sufficient, but what do I know.

1 Like
  1. Players / team.

4 players is good for me. With fewer players, ranks would be too spread out in each category. Or we could allow 4/5-player mix.

  1. Round duration: I like 1 month.

  2. Game length. I am happy with 30 min + 5x30s. As a player I’d prefer longer times, but as a spectator 30 min + 5x30 is better.

  3. Ruleset. Default Japanese, optional Chinese & others .

  4. Komi in handicap games. 0.5 is standard.

  5. Handicap. The simplest is H = ΔR -1 (maximum 9 stones): a 5k plays even against a 6k, gives 1 stone to a 7k,…, gives 9 stones to a 15k and weaker.

  6. Rank balancing. I wrote my opinion here: OGS June Team League - #118 by jlt

2 Likes

I only just realised that next month is July, not August… :D

2 Likes

H = ΔR -1 (maximum 9 stones)

Here’s another point of discussion.

Personally, I prefer uncapped handicap, but I hope that the boards won’t be unbalanced enough for this to become a practical concern.

Also, if we have uncapped handicap but mandatory standard placement then we’ll have to invent new standard placements for 10+ stones, like my 13-stone and 16-stone formations.

2 Likes
  1. No real preference
  2. A month seems fine (and it sounds cool to have a “July League”, “August League” etc.)
  3. 30m seems fine and is standard. I won’t cry if it’s 45m.
  4. Chinese default, Japanese optional
  5. 0.5 seems standard
  6. Difficult… I agree with the “1 handicap/rank” rule, as I feel your initial formula is hopeless for the weaker player (maybe I’m wrong though). On the nature of handicap, could we possibly imagine a mix of reverse komi and stones ? Like reverse komi up to -20, then stones (or the reverse) ? I feel like it would help balance the awkwardness of extreme handicaps.
  7. Not much to say, this is a practical issue more than a theoretical one.
2 Likes

Too much work, I am good with whatever you guys decide. :grinning:

I like Japanese ruleset only because it is the default. Again, I am lazy and whichever is fine.

Uncapped is fine for me too.

If instead of handicap stones we choose reverse komi, Black should get 14(ΔR -1)-6.5 komi if ΔR is at least 2. So

  • If players are of equal rank or if the rank difference is 1: even game with 6.5 komi for White.
  • If rank difference = 2: Black gets 7.5 points komi.
  • If rank difference = 3: Black gets 21.5 komi.
  • If rank difference = 4: Black gets 35.5 komi.
  • If rank difference = 10: Black gets 119.5 komi

    This may be capped or not.
1 Like

I’d be interested in joining this shindig. I don’t have a team and not sure if I’d be able to form one. Anyways, I’ll note myself as a free agent in case there is a team looking for a replacement.

3 Likes

I have never understood your formula, jlt, and I don’t think I’m going to understand it now.

No matter how many times you say that a difference of one rank is fourteen points, I know that is not correct. A 5k does not give an 8k fourteen more points than he gives a 7k.

I find Gia’s proposed system of a simple six points / rank much more practical. Complicated formulae are not suitable for application…

My formula is not complicated, I could make a table like in Gia’s post. My point is that 6 points per rank is too little. For instance in Gia’s table, a 1k is supposed to give 48.5 komi to a 10k, this is definitely too small. In my table I would put 105.5 komi, which would be more realistic.

To convince yourself about this, open Katago, let Black play 9 times and White pass 9 times. Then Katago expects Black to win by 126 points.

2 Likes

On the handicap, can we adjust based upon the results of the June one?

I don’t have confidence at all to play with only 2 stones against 4-5d as a 1kyu, but we will see. we have three 1kyu and 3 4-5 d. If the results overwhelmingly favor one side, we can adjust i guess.

OK, but could you please try to rearrange or approximate

komi = 14(ΔR -1)-6.5

into the form

komi = NΔR

or

komi = NΔR+/–X

or

komi = NΔ(R–X)

You have a lot of operations going on. Could we keep them down to three or ideally two?

I’d appreciate something everyone can understand at a glance, so we can be sure of what your proposal is.

Please be aware that not everyone in the league has the same level of mathematical literacy.

What I’d personally be in favour of is allowing Black to choose his preferred handicap mode for the league.

In the registration period, every player would specify whether they wished their handicap as either

  1. Set handicap stones

  2. Freely placed handicap stones

  3. Reverse komi

and that information would be appended to them in the relevant places, and would apply to all their games in that month’s edition.

I see organisational issues in mixing stones with reverse komi in a single handicap, so I’d suggest we hold off on that idea for now.

I suggest:

  • If ΔR = 0 or 1: play even game (with 6.5 komi for White).
  • If ΔR is between 2 and 11, Black gets 14 ΔR - 20.5 komi.
  • If ΔR is 11 or larger, Black gets 133.5 points komi.

3 Likes

But we should set a ruleset to look serious. No major tournament uses various rulesets.

We can think about the first formula as well, obviously, and poll it.

Don’t worry, everything will be voted on on the 20th–22nd :D (or perhaps before if discussion tails off).

I was going to open voting tomorrow, but I’m not sure how interested people are… better to do it later, perhaps.

How do people feel about handicap minus one? i.e. in a game with rank difference 6 the lower rated player gets 5 stones.

I feel that the half handicap is generous to the higher rated player. Out of curiosity, has any of the handicap games resulted in favour of the lower rated player?

Yes I (2k) lost a game without komi against ngothonghy (4k) but in our game, “handicap minus one” was equal to half handicap.

1 Like