I feel if black gets to play a white stone second and it isnât heavily restricted on how stones should be placed itâs an advantage. If white puts a black stone in a sensible spot, black has the option to put a white stone on the first line.
If itâs the case youâre only allowed put opponents stones in specific places it doesnât sound too much different from what youâre saying about handicap go. Itâs limiting the openings in some sense.
1 Like
Almost Totally OT
Am I the only one who always reads this as âTrue Remorseâ? 
(And, TBH, I think this terribleâor a totally different game ⌠Go is difficult already, and I wouldnât want to have to think about âcorrectâ order of moves while pondering over my next move. IF at all, then I could bring myself to accept the Pie Rule or Komi Bidding)
4 Likes
with all due respect, i think this is the worst idea i have ever seen.
either side will just put their opponentâs stone in the worst place possible, ie 1-1 and 19-19.
if you want a âneutralâ first move, the pie rule (I) mentioned is much simpler and gets the job done.
5 Likes
You arenât trying hard enough if you think this is the worst idea youâve ever seen. Personally I think this idea beats normal handicap rules (one players gets nine stones places in influential points around the board? Seriously?)
Anyway, yes, I agree, the âYou move, I choose if I want to play that or another moveâ rule is probably the best possible for a single move solution, but itâs not something easy to test on the server because you canât switch colors in the middle of the game. At least not currently. Maybe they will implement it as an option. BTW, in the pie rule, you have to take into account which player makes the move; give an example of a 15k versus a 3 dan, and the 3 dan probably doesnât care where the stone is, if they put it or their opponent does, they see the game so differently it would take 10 moves before they would start to really have an opinion. See the handicap thing above⌠you can give the 15k nine stones and they still lose.
From that, I think Komi really is the best solution for even games, and requires tracking players and assigning Komi/Negative Komi as needed. The games Iâve played with this system have been both enjoyable for both players and competitive because the higher level player is probably unsure how close the game is unless they are exceptionally good at counting.
1 Like
Handicap go, while it of course does change the style of the games, isnât that different from some games of go. It kind of forces black with 5+ stones to play sanrensei more than likely, especially with 6 stones onward. So in principle you can use handicap games to practice influence based games, black should try make boxes and moyos with enough starting stones. It is something you have to learn though, how to use the stones you have effectively.
From that, I think Komi really is the best solution for even games, and requires tracking players and assigning Komi/Negative Komi as needed. The games Iâve played with this system have been both enjoyable for both players and competitive because the higher level player is probably unsure how close the game is unless they are exceptionally good at counting.
I would kind of see that as a slight negative to the modified komi system. You could say it doesnât change the style of game as much as handicap stones, but thats only if the players arenât strong enough to know how differently they should play with 6.5, 0.5 -6.5 etc komi. You could say itâs similar to the way a weaker player might not utilise the extra handicap stones they get effectively, because thereâs something to learn and work on before they work as intended. In the modified komi games, you have to be good at estimating the score of the game as you said so you can make sure youâre accounting for different komi to usual games.
you are comparing apple and oranges.
the pie rule, or the thue-morse sequence, or the tak rule you proposed, they are all trying to give a fair game between even strength players, eliminating the advantage of black 1st move;
while the handicap is a set up to narrow the margin between different strength players.
5 Likes
Thatâs a fairy tale. Iâve never met two player of equal strength. The difference in games is barely ever within komi and itâs the result of a cascading mistake. On the extremely rare occasions (one in a thousand or more?) where two player are actually equal, they generally are professionals or high Dan level players and they have an epic battle where no one can really be sure who is winning. Those games become legendary and we study them⌠because they are as rare as unicorns.
I donât think Go needs a mechanism for making games fair between equal opponents.
I think Go needs one for the vast majority of games played, where the level of difference is over âtwo stones.â And, IMHO, thatâs Reverse Komi.
And one a personal note, I didnât propose the Tak rule, I said it existed and suggested how it might be implemented in Go, Iâm not proposing it be used, only offering up how other turn based games have delt with first move advantage. Why is that a âpersonal noteâ? Because you called it the stupidest thing you ever heard, which is uncredible and implied I was, by inference, stupid for explaining it.
Enjoy.
Wait⌠This doesnât work out. There are vastly more players who are kyuâs than there are who are professionals, and youâre saying that âusuallyâ a game between kyuâs there must be one of them who can be certain to win?
If youâve met 100 go players, youâre saying each of them should be given a different (reverse) komi, since none of them are of equal strength. So from the last 100 players here on OGS who had the same rank estimation, you would give all of them a different number of komi.
Komi is not meant to make the result of a single game as close to 0 as possible, but itâs meant to average the results over many games to go to 0. Against an equal opponent youâre supposed to be equally likely to lose with 10 points as to win with 10 points difference. The game result itself could be anything of course.
Nobody used the word stupid until you did, so your inference is flawed, Iâm afraidâŚ
I do agree with playing your opponents move does not change the game to be any more fair nor help level out the playing field. Itâs just another move for you, with good and bad places to put it, the good being the move thatâs bad for your opponent and vice versa.
3 Likes
Yes. Because one will know Joseki and the other will know how life and death shapes or one will understand direction of play and the other will be very aggressive. In these match-ups between kyuâs itâs most often the case that one will almost always win against the other because they possess a skill in the game the other lacks or canât counter. Worse, this changes with time as players either develop new skills or become lax and donât continue to study the game.
Thatâs currently unrealistic as a goal, but yes, in an ideal situation there would be a real number for komi for any two specific players based on their actual chance to win. Remember that rank is already being subdivided on OGS with ELO. In the future it might be possible to have an estimate or value for two players relative âKomiâ, specifically for their situations and playing styles. That, however, is overly optimistic for the near future and the Reverse Komi chart is likely close enough for making games interestingly close enough. That doesnât mean we shouldnât consider the idea that individuals will always have idiosyncratic plays styles that match up against other players differently and that all rating systems currently available ignore this because itâs either impossible to keep track of those variables for lack of data sets or because the amount of games required to gather that data is not available.
Yep, youâre right. Iâm the one projecting stupidity to the concept of presenting the worst idea andysif had ever seen. Mea Culpa.
Summary:
I think the most interesting games are the ones where both players are invested in attempting to win, whatever the win condition happens to be for them, and that they feel they have a reasonable chance of achieving that. I also feel that there should be maximum freedom of choice when choosing strategies and tactics, particularly for the higher level player who comes to the game with more tools. In an attempt to create that situation the suggestion of pie rules, Tak rules, and double move rules all feel a bit flat. They donât seem to address the most common situations. The huge range of skills (general and specific) across Go (as compared to other games), itâs very likely that in any given game, even among closely ranked players, one will most certainly have a winning advantage regardless of a trivial first move advantage, IMHO. Iâm almost certain that the vast majority of games played on OGS, where players tend to play players many levels different from their own, will almost always result in games ending in a predictable pattern.
If a player knows joseki and his opponent doesnât, then they are most certainly not nearly the same rank. Donât go tell me that all your opponents either knew a joseki and therefore won the game, or you knew some direction of play and therefore you won (or something similar). Sure, a few games end quickly if someone pulls off a trick play or gets punished for overplaying, but the vast majority of (say) 10k players knows about the same number of joseki, is equally good (or perhaps more fittingly, bad) at direction of play, life and death, etc. My experience certainly is not that usually I lose my games because my opponent knew something I didnât or win my games I know something my opponent didnât.
What are you on about? 6.0001 komi or 6.99 komi or anything in between yields exactly the same outcome. You canât score a non-integer number of points in go. Itâs a discrete game.
4 Likes
Thatâs interesting, you feel that when you lose against an âevenâ opponent, itâs not because of a mistake you made that your opponent took advantage of? What, then, is the reason you lost those games? Reversed, why did you win games against âevenâ opponents?
This is the Reverse Komi table. As you can see, if both players are ranked at 10k the Komi is considered to be -6.5 but if you change to 10k versus 11k, the Komi becomes -0.5. 10k versus 12k?.. +5.5. This suggests gradients between ranks that are not yet acknowledged. You could have two players, accurately ranked, who had a Komi of +3.5 between two specific 10k and 11k players.
Komi could be mapped to ELO rather than rank. Gradation to the whole stones^ of Komi is possible, for example, on OGS, but I donât think anyone has studied it.
Please note here Iâm NOT suggesting Komi of whole stone values, like a Komi of 4.0⌠the benefit of having the .5 is that you donât end up with tied games and thatâs important, IMHO. Rather Iâm suggesting that between the ranks of 10k and 11k there is a gradient of ranks which is obvious from the amount of Komi covered between these ranks in the study. These could be mapped to ELO rather than to Kyu levels. âŠď¸
I do not feel that it usually is because of a mistake, but because I made several of them, more than my opponent (who also makes mistakes, I mean, everyone does). What is usually not the reason is that Iâm more skilled in a certain aspect of the game. Moreover, often it is carelessness or laziness in reading thatâs the culprit of the mistakes. With most opponents of the same rank, I would be very hesitant to bet anything substantial on the next game having the same result.
You talked about real numbers, which I took as a suggestion to give people komi like 6.23, or -Ď/2. You meant integer komi with an added half point to decide draws, which is a different idea.
Sure, I agree on that. But still then, out of 100 players ranked 10k, I would estimate about 10 of them will end up playing with the same komi against each other; hence being âequalâ in strength, simply because there are only so many values for the komi that you could reasonably expect to pop up (almost all of them between 0.5 and 11.5, I bet, since thatâs what their rank is supposed to be implying). Itâs outrageous to suggest the chance of playing ar equal strength opponent is as rare as a unicorn.
2 Likes
At 11k every move is a mistake
9 Likes
You make mistakes because you lack the skills to make the proper move for the situation. Making the right move or moves in a sequence is a skill, a skill a higher level player would possess. You mentioned careless or lazy reading, thatâs a skill. Youâve made this point for me. In Go, you lose because you lack the ability to make better moves, the only way to make better moves is with skills you donât yet possess.
You may come from a background of mathematics which loads the words âreal numbersâ with a meaning I never intended. Perhaps, in this case, the phrase âactual valuesâ would have been better but now knowing your background, or that of other readers, it would have been hard to think this far ahead. I wasnât comparing Komi of real, imaginary, integers, etc⌠Just specific numbers.
Even with players who would statistically be consider equal in ELO, which is more specific a number than rank, itâs very common for players, particularly at kyu levels, to have a weakness which loses them games. They might even be very good at hiding this weakness, for example perhaps Alice is very bad at end game so she has learned to play very good at openings (or even because she is very good at opening, she hasnât needed to learn how to play endgames.) Our hypothetical Alice has strengths and weaknesses and will be easier or hard to play against for any given Bob. But itâs very likely that for any given Bob, that the results of play will be similar or the same. If Bob is very good at end game, but is good enough to weather Aliceâs openings, then Bob will consistently beat Alice, even if Bob canât beat Charlie, because Charlie loves to playing fighting style, which is Bobâs weakness.
But you can, if you wish, simply look at your own game records, without cherry picking, and see that games played against similar or the same level player, when you click on that player, you will find it rare that you have a 50/50 split in game results with them. Rather, youâre much more likely to have something like 80%, 90%, 100% splits.
Finding a player with whom you actually split games 50/50 is rare because it requires a balancing of skills between both players and those skills have to impart roughly equal chances to win the game. You might even seek out these players because these games are by nature more exciting or interesting. So you may have a history filled with specific âunicornsâ who you enjoy playing. The vast majority of people on this server, however, are either much better or much worse than you.
On last, interesting piece, here⌠perhaps we should move to a ranking system that has over 200 levels. 
Failing to predict that a reader will assume you are using a term correctly and read it in that light is not indicatory of any error or lack of charity on the readerâs part, nor of a lack of knowledge on your part as to the background of the reader, but of a mistake in terminology. His background does not matter, that is not a valid excuse for using the wrong word.
Yes, one can construct non-transitive loops of players. Your point? Some people will have a harder time beating certain other players.
In my games, I doubt thereâs a single other account against which I have a statistically significant record, and even less so one equal to me in rank, which is where youâd start to expect 1:1 win ratios. 1:1 ratios (if we assume we can find opponents with a statistically significant heads up record with oneself) are not rare due to the lack of a âbalancing of skillsâ, but due to most of them being just plain weaker or just plain stronger than oneself.
1 Like
Of course I make mistakes because of lack of skill, thatâs not my point. My point is that I donât lose games solely because of lack of skill. If I play the same opponent 5 times in one week, and the results are win, win, loss, loss, win (seems very reasonable to me), then youâre claiming that my opponent suddenly gained skill after two games, and then I did after four games.
Assuming itâs not our first games ever, but that weâve both been playing for years, I refuse to address the reason for winning those games as difference in skill.
Well, you could claim that your âreal numbersâ means something different, but you canât expect me to know what it means if it doesnât mean the thing it commonly expresses. If you didnât meant the mathematical âreal numbersâ, Iâm not even sure how to interpret ârealâ to make sense.
As a side note, I believe the concept of real numbers is usually taught in high school, so you donât need a degree in mathematics to know that a real number is the wrong word for what you meant.
So your idea is that Alice wins from Bob wins from Charlie wins from Alice. This is a nice argument for not using a linear rating system, but I donât see what it has to do with assigning different komi. If Iâm correct, you want to assign komi based on a finer rating system than kyu/dan ranking, letâs say ELO. Then you canât do that such that Alice, Bob and Charlie can all play equally matched games, since ELO is linear.
So this has nothing to do with this discussion, as far as I can see.
If I look at my game records, there are very few players Iâve played multiple times, and even less of them who have the same rank, so I canât really judge if this is true based on empirical data. Iâm not sure where you base your idea that it is rare on, but if you have data backing up your claim, I would love to see it.
Especially since, although I canât judge the claim using my own game history, I can judge it based on my background in mathematics (but again, nothing beyond high school is necessary for this). Suppose I played so much that I have a significant number of people who are all the same rank as me who I play with regularly (and usually go to scoring), and with most of them I have a winning average either around 15% or around 85%.
Then I would need the average result of the score for each player to be around 1.5 standard deviations away from 0 (thatâs what it means that my winning average is 15% / 85%). Letâs make the conservative assumption that this standard deviation is about 6 points (meaning that for the games with a certain player, two thirds of the scores fall within a range of 12 points). I believe that this is conservative, since I usually see games being resigned (hopefully this means that the disparity is even larger), or win margins that are far greater (not uncommonly >40 points). Of course this goes down as players get stronger, since an average pro does not make 30 point mistakes in a usual game, while at 20k it happens multiple times.
So, using this standard deviation, that means that for each player I win on average by more than 9 points, or lose by more than 9 points. But then I should not be the same rank as these players, since we do know based on empirical data that a rank difference is about 6 stones in komi (from your reverse komi table).
So this simply cannot be the case, as long as my estimate of standard deviation is not too high.
1 Like
The answer is âNo.â
6 Likes
Okay, Iâm done with this.
Thanks for summary smurph.
1 Like
hi Iâm new here. My name is Andrew and Iâm in 5th grade and I really want to be a grand master at the game of go. Can someone explain this thread to me in how Iâd understand it?
1 Like