LGCLadder/Discussion

Sub-page of LGCLadder

Table of contents

Proposed rules

  1. Players ascend the ladder by beating a player higher than him- or herself. No player may challenge someone more than 3 rungs higher unless those three players are not present. Only then can that player challenge the fourth higher player. And then, if they challenger wins, s/he can only ascend a maximum of 3 rungs.

Ken: Do we need this rule if we are using the ELO system?

  1. When the challenger wins, s/he ascends to the rung immediately above his/her opponent.

Ken: Shouldn't this be determined by ELO rating?

  1. A player can only challenge a person higher than him- or herself on the ladder.
  2. Two players may only play each other in a ladder match once per meeting. No rematches.
  3. Games between two AGA members will be rated.
  4. Intitial seeding of the ladder will be based on either a players KGS, DGS, or AGA rank. Disputes will be resolved by flipping a coin. After the initial seed, the parties can hash it out on the goban.

Ken: Will the staggered entry work better?

  1. Participation in the ladder is voluntary.

Ken: Oh no, it is manadatory baby! Well okay, I guess it can be voluntary.

  1. A player missing three consecutive regular Saturday meetings will drop one rung per meeting until they return or they are at the bottom, whichever comes first. Once at the bottom, if they miss one more meeting, then they are off the ladder.

Ken: Do we need this rule if it is rating based? I think we need a way to cull non-active people off the ladder. Especially if they are strong and high on the ladder. Maybe you lose X ELO poinsts for missing three meetups in a row.

   Or what about a monthly fee ($1) to stay on the ladder?  This would help pay for periodic awards for advancement.

Ken: I like this and would go for it. I am not sure about the other members though. We should ask I guess. Maybe it is a $1 a game or $20 a year or something.

  1. Once the initial ladder is established, new players may challenge anyone on the ladder, and if they win, then they take the slot immediately above their opponent. If they lose, then they start at the bottom and must work their way up. If multiple people start on the same day and end up at the bottom, the outcome of that day's games shall determine the order. The person losing against the weakest player will be at the bottom. The person winning against the stronger player will be higher.

Ken: Again, with the ELO system, this does not seem to be necessary. Everyone starts at the same rating.

  1. To keep things simple, a player may not play more than three ladder challenges per meeting.
  2. The ladder will be updated after each meeting that has results to report.

Issues and Questions

Ken: So the ladder is not meant to represent an accurate picture of who is better than who, per se, but rather intended to be a sort of competition concerned with who beats who most recently. More like a tourney, than a rating exercise. So if I beat a high ranked player using a handicap, I should be allowed above them on the ladder. But, how are handicaps and ranks being determined?

ed: If we have AGA-rated members, we could use that. Otherwise, I would propose that we set the rule that we accept either AGA, KGS, or DGS as the rating standard, though I am not thrilled about using DGS. In DGS, I could be consulting with people and books, running variations on each move, and all sorts of stuff. That would "artificially" inflate one's rank. On the other hand, games played in person or live on a go server, are less likely to have "cheating."

Ken: Per the last issue, do we need to start maintaining our own records of player ranks?

ed: That would be useful. Many clubs do this. There's software available that can facilitate it. I could write a web application that can put it in a database and make it public. However, I would propose someone volunteer to be "record keeper" or statistician or whatever. I would work with them on the web app.

Ken: Should each player choose a method for which they would like their rank established?

ed: I think there ought to be a standard, like AGA first. If you have a rank with AGA, KGS, IGS, and LMNOP, then I say AGA takes priority, followed by KGS. I say that because IGS ranks start with 22 kyu, and KGS's and AGA's ranking systems are similar enough to make easy comparisons. So if you don't have an AGA rank, then your KGS rank. If not KGS, DGS. If not DGS, then take your chances by challenging whomever you like. But with this last option, you only get one shot. You lose... you start at the bottom and work your way up.

Ken: How shall it be updated?

ed: I propose that the Record Keeper collect the results of the week's games and enter them all at once into the web application. That way, Monday morning, everybody could see the most recent ladder on the website.

Ken: Do we avoid the whole rank issue and only allow even games?

ed: That would make things a heck of a lot easier. I think that makes things a lot more competitive and interesting.

Ken: Do we let players just decide what their rank is and use the honor system and work out handicap issues with the other player? If there is a sandbagger, how do we handle that?

ed: I think the social consequences of sandbagging ("Hey, a$$h0le!") will enforce the honor system very well. I have very strong opinions about sandbagging, so I would be in favor of banning someone from the ladder for a long time for this heinous offence against nature, the Bushido code, and everything else.

Ken: Same thread really, but what if we keep track of handicaps given, and modify them per the outcome of ladder games only. For instance, the first time I (15 kyu) meet Alex (5 kyu), I am given 9 stones. I win by 14 points. The next time we play, I am only given 8 stones. If I lose, it stays at 9 stones. (No handis above 9). So handicaps are set the first time relative to rank, and after that relative to the last time the two players had a game and the results of that game. Does this make any sense?

ed: The issue with that is at the lower ranks, individual games are not necessarily representative of someone's true strength. In other words, I can make some brilliant 5 kyu opening moves, make some idiotic 30 kyu attacks in the middle game, and hold my own with a 15 kyu endgame. This kind of complexity only makes even games look better.

Ken: Given the above. If a person's rank on the ladder gets very far out of whack with their actual rank, perhaps there should be a way to adjust ranks. Maybe upward only. It still seems we need to look into how and when we change ranks on the Ladder.

ed: I might be misunderstanding you, but all this really matters in the initial assembly of the ladder. Once the ladder is set, let the outcome of the games decide. Things will settle into their natural balance based on the invisible hand of fate. And skill. And luck.

Ken: This does not seem to hard to do: [ext] http://gobase.org/studying/articles/elo/

ed: This would be a snap to implement. All it would take is for everybody's initial rank to be set, and then let the games begin!

Ken: As always you are the voice of wisdom, wit, and reason. The Bushido code and Sharia law it will be! Wait - no. I think even games makes life simplest - and some people really hate handicap games. Also, I think it can be allowable for a high ranked player to offer a lower ranked player a handi if they want to. No mandatory handicapping. That said, why not do a simple ELO rating thing and let that determine where you are on the Ladder? It may mean that if you win, you may not necessarily move up though. That might be a bummer...

Ken: Is there a way to avoid establishing the initial seeding? Just say, everyone starts at 0 and the best rise up? Maybe that means a staggered start. As in a real ladder, not everyonne climbs it at once. You go in an order. Maybe the first saturday, a couple of people start the ladder. Their game sets the first two rungs. Then a third and fourth person can challenge each of them and enter the ladder. But maybe everyone starts with a 1500 ELO rating or something and we just play it out.

ed: OK, let's say this. Our standard for determining rank is ELO. Leave it to me to come up with the technology for maintaining this data, so assume the "how" is taken care of. Everybody starts at 1500. Everybody plays even games. If a handicap is given, then WHITE's rating is reduced by some constant number of points per stone (how much?) But handicaps are discouraged if only because they muck up the math. Over time, as more games are played, the player's skill will get this ELO rating closer to the truth, i.e. his/her true strength against everybody in the club.

Ken: I'm cool with all of that. But maybe we keep the math and rules simple. It is White's perogative to offer a handi if they want to. If they do, they do it at their own peril of decreasing their own odds, but are otherwise not penalized.

Ed: So, I guess we come up with some point-equivalent for a handicap stone. 50-100 ELO points per stone? I dunno. That will be easy enough to accomodate in my web app.

Ken: Do we even need that? I think it should be entirely up to the White player if they want to give that stone (out of the goodness of the heart), but not really have that accounted for in the ELO system. Keep it all simple. Maybe we just ban handicap games altogether, kind of like the Championship on LG. All even games. Theoretically, you should be able to move up the ladder by knocking off people a few rungs above you, not shooting for the top every time. This time when you win - there is no "Well with a 9 stone handicap" caveat. If you win, you win. Straight up.

Ken: The other thing I was thinking was giving a bonus for a skillful resignation. The rating system only cares whether or not you win, not by how much or if you resigned. Perhaps if you resign before move 100 or 150, you get a small percentage of your rating back. If you were supposed to drop by 20 points, maybe you get 20% of the drop back or something. Maybe it is too much trouble. Likewise if you really crush someone, maybe there should be a bigger rating reward. Probably unnecesary but just a thought.

Ed: While this is all technically possible, I think the games should stand on their own. This is the way it is in all other go tournaments, ladders, and whatnot. At least from my research. A win is a win; a loss is a loss. If you win by .5 or 200, it doesn't matter.

Ken: Yeah I am cool with that. Nevermind!

Phil: So... are we using Elo to determine position on the ladder, or a system of rungs where if you beat someone you move up to their position regardless of Elo ranking? I'm a trifle confused.

Ed: It's intentionally confusing. We want to foil terrorists who might seek to undermine our great Republic by getting in on our Ladder. ELO only would determine position. At first, there would probably be ties. But over time, it would be less and less likely that two people would have precisely the same ELO score, especially if we store their score to two or more decimal places.

  1. To remain on the ladder, a player must pay a fee of ??? per month/year (to be determined.) Players who do not pay are dropped from the ladder, though their rating does not change. Once they pay again, they are readmitted to the ladder with the ELO score they had before.

Ken: Let's just make it free. Let's build participation before funding.

Ed: Done. The fee idea was to a) identify who was active and drop inactive players and b) provide money for possible prizes or incentives. But I agree the fee idea is something we can re-visit later.


This is a copy of the living page "LGCLadder/Discussion" at Sensei's Library.
(OC) 2007 the Authors, published under the OpenContent License V1.0.
[Welcome to Sensei's Library!]
StartingPoints
ReferenceSection
About