-2

I have been following the controversy surrounding the pairings of Hou Yifan at the IOM chess tournament 2017. She was paired with 4 consecutive female players, this is following her last open swiss tournament outing in Gibraltar where she was paired with 7 female players from 10 pairings, including round 1 - 4 consecutive female-female pairing as in IOM 2017.

The question is can we not design a simple non-deterministic swiss pairing system that does not penalise the "outliers" of the female population of players? It is clear that rules make it more likely for Hou Yifan to be paired with women as there is an elo gap in mean the population of male elite GMs and female elite GMs and the lowest elo on the highest score and highest elo on lowest score pairing pattern used increases the likelihood of these gender-specific pairings. Any ideas?

Surely we can create opt out for gender pairings which occur consecutively?

11
  • 5
    I don't think you fully understand how the Swiss system works. If Hou Yifan was matched with another woman in the first round, then they were in different halves of the elo pool. She gets the exact same treatment as all the other GMs from her half (I'd call that equality).
    – Annatar
    Sep 27, 2017 at 7:07
  • In fact, the upper half is expected to win their games against the lower half, and are then matched against each other because they have the same score. Assuming that the frequency of female GMs is higher in the lower than the upper half (your argument about mean of population), she has a lower chance to play against another woman in the second round already.
    – Annatar
    Sep 27, 2017 at 7:09
  • 1
    @Annatar: as it happens, at Isle of Man, the 1st round was paired completely at random. The rest is Swiss as usual. Sep 27, 2017 at 7:15
  • 8
    And of course the question remains of why she would care about the gender of her opponents at all..
    – Annatar
    Sep 27, 2017 at 7:22
  • 2
    The real question is whether these pairings are "bad luck" or whether they have been manipulated to somehow "penalize" Hou Yifan. ChessBase has a report on their website now discussing the Gibraltar pairings, concluding that both SwissMaster and SwissManager would always have given these pairings. So it seems that indeed the stars just haven't aligned for Hou Yifan, and this was just a coincidence. (The first round pairings of IoM are another issue though.)
    – TMM
    Sep 28, 2017 at 13:20

3 Answers 3

8

The current algorithms are deterministic. I have no idea why you would want a non-deterministic algorithm. The advantage of the deterministic ones we currently use is that they can be easily done by hand (and were done by hand until quite recently), so if a TD changes them manually that can be detected.

Gender of the players is not an input to the algorithm at all. It does not play a role whatsoever.

Hou Yifan had a very unusual draw in this year's Gibraltar and the first four rounds of Isle of Man; but that's just a coincidence. Something like that has never been noticed with any other women or other tournaments or other years.

3
  • That's a tautology... what's the advantage? Oct 2, 2017 at 13:47
  • 1
    Yes, I know what the word means. A deterministic system always has one clear next step and will always lead to the same output given the same input, a nondeterministic system doesn't. Hou Yifan didn't ask for a non-deterministic one, and so far it's always been seen as an essential feature that the algorithms used must be deterministic, so it can be easily checked that the pairings haven't been tampered with. Hou Yifan suspected tampering, so she would be very much in favour of that. You are the first one I ever hear who wants a nondeterministic algorithm, and I'm asking for a reason why. Oct 2, 2017 at 19:17
  • 2
    Tampering is a big problem because organisers try to give their local boy who is trying for a GM norm easier pairings, or just their friends. If the algorithm were non-deterministic, they could always claim it was just randomness. Deterministic is fairer, the same rules for everyone, all the time. Oct 2, 2017 at 19:32
3

Surely we can create opt out for gender pairings which occur consecutively?

Yes, we could. We can make whatever rules we like, after all. But I don't think we should.

Such a rule would have to come at the expense of some other pairing rule. We currently have rules that players cannot play each other twice, that players with equal scores should play each other, that colors should be equalized, etc. Which of the existing rules are less important than equalizing which gender a player will play?

Such a rule would also imply that playing females is somehow undesirable, or gives an inherent advantage or disadvantage compared to playing males with the same rating. I don't think these are views FIDE wants to encourage.

And any rule that makes things "non-deterministic" makes it harder for FIDE to verify that the pairings were not manipulated. (If we did have such a rule, why would it not be deterministic?)

It is clear that rules make it more likely for Hou Yifan to be paired with women as there is an elo gap in mean the population of male elite GMs and female elite GMs and the lowest elo on the highest score and highest elo on lowest score pairing pattern used increases the likelihood of these gender-specific pairings.

First of all, people with "highest" and "lowest" scores are very unlikely to play each other. Perhaps you meant "higher" and "lower".

Second, there are pairing rules about minimizing the number of people who get the same upfloat/downfloat as the previous round, or two rounds ago, so while this may be a factor for some players, it won't happen every round, or even every other round.

Third, in both Gibraltar and Isle of Man, she did not float (in either direction) in any of the rounds, so this was pretty much irrelevant. It was sheer luck that she was paired with women.

2
  • The opt outs for people in the same team, federation and family. An opt out for players of the same gender is not as silly as it sounds since most women play in a gender specific tournaments and have not got the same or similar population spread as the open tournaments. That is the reason for this opt out. Think approximation to the normal distribution of two populations with different means. Oct 2, 2017 at 13:58
  • Did those two tournaments accept opt outs for people in the same team/federation/family? I'm guessing not, as that would have affected things in the report on Gibraltar pairings that TMM referenced in his comment.
    – D M
    Oct 2, 2017 at 15:43
0

The question is can we not design a simple non-deterministic swiss pairing system that does not penalise the "outliers" of the female population of players?

What you are saying is sexist and very offensive. You are saying that when Hou Yifan is paired against other female players she is "penalised".

It is correct and fair that the pairing systems are required to be gender neutral.

Your suggestion that -

Surely we can create opt out for gender pairings which occur consecutively?

offends against one of the basic principles laid down in the General handling rules for Swiss Tournaments in the FIDE Handbook -

A. Pairing Systems

...

5 It is not allowed to alter the correct pairings in favour of any player.

6
  • 2
    I would be grateful if you did not resort to personal comments here. I am free to ask a question here without being accused of being sexist and being offensive. The question is about gender. If Hou Yifan didn't have a problem with her pairings I would not have posted this question. Is Hou Yifan sexist for wondering if her pairings of 4 consecutive females in two consecutive opens is truly objectively by non-gender biased rules or subjective or manipulated in some way. Is she sexist? I think not so please do not accuse me of being sexist.If questions about gender offend you then don't read them Oct 2, 2017 at 14:04
  • 1
    "Is Hou Yifan sexist for wondering if her pairings of 4 consecutive females in two consecutive opens is truly objectively by non-gender biased rules or subjective or manipulated in some way." - It was perfectly reasonable for her to wonder, as such pairings are unlikely. But resigning in protest when things were verifiably not manipulated was not reasonable.
    – D M
    Oct 2, 2017 at 15:24
  • Totally agree @DM. I wonder why a lot of people think my question implies the FIDE laws are at fault or the organisers or software are to blame. FIDE Laws work and the organisers and pairing software are blameless. Hou Yifan is to blame for assuming a non-bias statistical trial like a swiss tournament pairing system is biased. I am asking intellectually what can be done to take it beyond all doubt. It is a question and I'm only asking it. I have a right to ask any question, right? Oct 2, 2017 at 18:55
  • @Michael FWIW my comment was aimed at Brian Towers. (My previous comment saying this answer is offensive was removed, presumably because Brian thought my comment hurt his feelings. Apparently Brian is the only one allowed to offend others here.)
    – TMM
    Oct 3, 2017 at 20:20
  • @DM You are assuming Hou Yifan also knew this. At the time she likely did not know these pairings are verifiable, and that verifying this would show the pairings were not manipulated. (If she did, I am sure she would not have made a big deal out of it.)
    – TMM
    Oct 3, 2017 at 20:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.