Pente Masters 2020: Solving the seed problem
Posted:
Jul 14, 2020, 9:46 PM
Richardiii is the lowest seed in the tournament. With the current format, this means that if he makes a deep run, he will be placed in the same section as the highest seed every time. This is highly unfair to pente_gon.
Zoeyk is also slightly misplaced as the 10th seed. This might not seem terribly problematic, but it has the potential to compound the problem of Richard's seeding. For example, in Round 1 it is entirely plausible that 17-20 players advance, including all of the top 10 seeds and richard, which would lead to the following absurd scenario for Round 2 Section 1:
1 pente_gon 10 zoeyk ... 25 richardiii
It is obvious that we need to take action to prevent this from occurring. I have tried to think of measures that would be (a) as fair as possible to all competitors and (b) minimally disruptive to the current tournament format. I have come up with two options:
Option 1: After Round 1, if Richardiii has advanced, reseed him according to his post-Round 1 rating and everyone else's pre-tournament rating. You could repeat this process after every round or only do it once.
Option 2: After each round, if two players who were in the same section in that round are again placed in the same section, then switch out the lower seed with the closest seed that did not face any other player in that section in the previous round. Repeat as necessary.
Option 1 appears to be less complicated and most likely easier to code, but is also less equitable in that it singles out Richard. You can justify this, however, by making it the new standard for provisional players in masters tournaments.
Option 2 may be more difficult to implement, but it has the advantage that it is completely equitable and more fully solves the problem of repeated matchups. Personally, I vote for Option 2 if it is plausible to execute.
It's no use going back to yesterday, because I was a different person then.
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 16, 2020, 2:05 AM
I don't have (much) time to devote to this.
Option 2 runs into a problem when, for example, all players in a section tie and the next round will have 3 sections or less. I do love the pigeon hole principle, but it's working against us now.
Option 1 seems hard to justify, I could have only let established ratings participate in the first place as well. Currently, all provisionally rated players (regardless of their rating) are seeded below all established ratings.
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 16, 2020, 3:02 AM
While I understand that you have little time to devote to this, I also think you'd agree that it would be a shame if the most stacked pente tournament in 5 years had its competitive integrity seriously compromised by two top players facing each other in the same section up to 5 times in a row.
There has to be a minimally invasive, minimally time-consuming solution to prevent this from happening. Both of my proposed solutions have flaws, but I think we need to choose something, and those are the two best ideas I was able to come up with. I'm open to other ideas.
I don't think Option 1 is hard to justify given this very special circumstance that, his pente.org TB rating notwithstanding, richardiii is one of if not THE most established pente player in the country, having been ranked in the top 10 for at least two full decades if not more. I think if an exception can ever be made, it can and must be made for him.
Option 2 does run into more problems, but later on in the tournament I think there's more of an understanding if you get placed in the same section as another person twice in a row. Let's at least try to avoid that for the first 3 rounds.
I don't know much about coding at all, but I'm sure I speak for myself and zoeyk (who is an admin here and may be willing to help?) in offering my services to help get this done.
It's no use going back to yesterday, because I was a different person then.
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 16, 2020, 6:12 AM
Eh, I think Pgon and Rich should just plan to come to some sort of gentlemen's agreement to save their best ammo against each other until they face each other in the final section. That's what I'd do in one pair of their shoes if a coded solution weren't in the cards. Why waste their best ammo in early rounds when both are likely to advance regardless?
Message was edited by: watsu at Jul 16, 2020, 6:14 AM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 16, 2020, 7:27 AM
That solution works in the abstract, but let's focus on actualities and suppose that my worst-case scenario (which has a fair chance of happening) comes to pass: zoey ends up in their section in Round 2. Now, if pgon and rich abide by this "gentlemen's agreement" and hold off on their strongest lines, they will most likely split their set. Suppose also for the sake of argument that zoey beats richard and splits with pente_gon. The fourth player most likely loses to all three. This results in zoey and pente_gon advancing, and richard forever regretting pulling his punches against pente_gon, because if he had won the set he would have advanced as well.
Agree that there is a serious problem here?
It's no use going back to yesterday, because I was a different person then.
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 16, 2020, 7:56 AM
Well, as a hypothetical scenario I agree with you that having all three of them in the same section in round two could lead to a less than desirable outcome for one of them, assuming all three don't have gentlemen's agreements. Or, they could all three advance out of that section even without any gentlemen's agreements in effect between any of them. I still think all of them might fare better planning for the possibility of agreement contingency than not. No top 5 rated seed is going to be happy to have both Zk and Rich in their section before the third round, so splitting them up if possible would be of probable benefit to the strength of the tourney, but if I were in Vegas I'd bet on a final section of those three splitting games all around. I'd be happier if they weren't all in the same section before the final, despite what that means for my odds of advancing past round two. But, whether and how the tournament seeds for the next rounds should be tweaked to avoid a premature clash which ends someone's tourney chances... I'm not going to offer any ideas or vote on my preferences among options.
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 16, 2020, 8:28 AM
Another not impossible scenario is that Rich, Pgon and Zk could all split sets among them and only two of them advance because one of them failed to win both games against a fourth player. Sort of like what happened in live play in 2003 with Gary Barnes, Dmitri King and Ilurath all splitting sets, but Ilurath won one less game against another player than the other two did. Also a less than ideal outcome to have happen in a section.I ETA: I'd say if Rich is willing to play enough games to get an established rating between now and the next round, he should be seeded for the next round based on his established rating. In general terms, I think this reseeding based on ratings at the end of each round makes sense, and would handle cases where a player had a provisional rating initially, but was playing enough sets to have an established rating by the end of a round. Didn't I say I wasn't going to vote in my last post? Oh, well... that's my vote for a code alteration. Reseed tourneys in each round based on current ratings at the time the next round starts, perhaps making it based on a rating grab scheduled to happen at an unknown time somewhat prior to the start of the next round so that there's less potential for attempts at rating gaming...
Message was edited by: watsu at Jul 16, 2020, 8:53 AM
Message was edited by: watsu at Jul 16, 2020, 9:00 AM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 17, 2020, 4:29 AM
I'm very reluctant towards making changes, especially since the tournament has already started. And concepts like competitive integrity don't resonate much with me since it sounds very fuzzy, undefinable, unquantifiable.
The only change that comes close to a bit of reason is sorting the players on every round not according to pre-tournament seed, but using their rating. The big hiccup there is that right now, all provisional ratings are sorted below all established ratings, and for reason: provisional ratings jump wildly and are unreliable. So this change would be tailored to R3's case, and sabotage future sister tournaments where participants are rated below 1800.
You mention fairness, but this was the deal everyone took on when they signed up for the tournament, I have huge doubts about the fairness towards other players when changing the system after the start. I'd love to see a thunderous final with all the strongest players, but not at the expense of other participants. And change or no change, a spectacular final looks like it's in the books already, with or without R3.
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 17, 2020, 4:01 PM
it seems to me that to win the tournament, you have to beat the last man standing. richardiii beats pentegon, or zoeyk beats madhippo. still leaves r3 and the hippo. it will be a good game, just without karlw and watsu (who has said he probably wont finish)
Posts:
2,233
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 17, 2020, 10:58 PM
just temporarily change me and richards rating to a rating you think fits our skill levels, then after the tourney switch the ratings back. i doubt me or richard care about our turn based ratings. of course ask rich for his opinion maybe. this would fix the problem yes?
Or if this is unfair to other participants never mind. just shooting ideas.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 17, 2020, 11:31 PM
Okay rainwolf, you claim that the term "competitive integrity" which I used is fuzzy and undefinable. Let me try to make it concrete.
a) In tournament formats other than single elimination, there is almost always a rule that prevents repeat matchups whenever feasible, usually by performing a minimally disturbing rearrangement of the matchups for that round. The reasons are many: if one player has the upper hand in the matchup (which is almost always the case), it is cruel to repeatedly punish the disadvantaged player rather than give them a fresh chance. Furthermore, the point of formats such as round robin is to give players the chance to face as many opponents as possible.
b) Seeding in tournaments is specifically designed to separate top players into their own divisions so that they don't face each other until late in the tournament.
If I understand correctly, this is a seeded, round-robin tournament in which
(a) the division of the sections has no safeguards to prevent repeat matchups
(b) the seeds are inaccurate and are actually working contrary to their intended purpose (you could argue that it was richard and zoey's fault for not having an accurate rating, but I would call that being petty).
Why have seeds at all if they aren't going to work as intended? Why have a round robin tournament if the same players are going to play each other each round?
This is what I mean by competitive integrity. I believe at the very minimum, a few lines of code to prevent repeat matchups would go very far to making this a better tournament.
It's no use going back to yesterday, because I was a different person then.
Re: Pente Masters 2020: Solving the seed problem
Posted:
Jul 18, 2020, 2:41 AM
I believe a more accurate seeding based on comparative TB historical playing strength would have been as follows: 1. spavacz 2. be_water 3. zoeyk 4. richardiii 5. karlw 6. madhippo 7. pente_gon 8. watsu If this seems like an arbitrary list, well it's actually based on the highest established historical TB ratings for each player from Brainking. If we'd had happyjo and nosovs join the tourney, the seeds would've been even more inaccurate. Since Rich made no attempt to increase his provisional rating before entering the tourney, I have to conclude that he either knew that provisional players would be seeded below established players, didn't care where he was seeded or both. Since a comparable field is unlikely to assemble in the future (not just my opinion) how this tournament's seedings are handled is unlikely to be needed as a precedent going forward. I do agree with Karl that avoiding continued matchups between Rich and Pente_gon would be preferable in this case, but like I said Rich doesn't seem to be making any playing efforts towards an established TB rating here, which would help place him in relation to other Pente.org ratings for seeding here or commenting in these threads.
Message was edited by: watsu at Jul 18, 2020, 3:03 AM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
