Hou mystified by pairings, sits out round five
9/27/2017 – Since round four ended with the four leaders drawing their respective games, now seven share first with 3.5/4, from Magnus Carlsen to Alexander Lenderman. Hou Yifan, who faced her fourth consecutive female opponent, requested a half point bye in round 5, but contrary to rumour has no intention to withdraw for now. | Analysis by GM Elshan Moradiabadi; Photo: Alina L'Ami
Find the right combination! ChessBase 15 program + new Mega Database 2019 with 7.6 million games and more than 70,000 master analyses. Plus ChessBase Magazine (DVD + magazine) and CB Premium membership for 1 year!
Carlsen and six others tied for lead
The foremost games of the round were the two between the last players left with 100% scores: former FIDE World Champion Rustam Kasimdzhanov against Magnus Carlsen, and Alex Lenderman, who has often shared his deep analysis with ChessBase readers, against Pavel Eljanov.
Kasimdzhanov opened with 1.e4 without trepidation, perhaps wondering whether the World Champion would repeat his ‘Tiger Modern’ from round two. When Carlsen replied with 1…Nc6, essentially inviting all manner of unorthodox opening play, Kasimdzhanov backed down slightly, and a slightly offbeat Pirc ensued. If Magnus was hoping to just outplay his opponent, he also had to deal with his serious space disadvantage, which Rustam negotiated well, and they eventually drew after 72 moves.
It as nice to see both players leave the board in visble good spirits after the hard-fought game that had lasted no fewer than 72 moves. | Photo: John Saunders
Alexander Lenderman showed that his share of first was not a fluke, as he outplayed his higher rated opponent, Pavel Eljanov, for a good portion of the game, and even achieved what was likely a winning endgame advantage.
Lenderman ½-½ Eljanov
Alexander Lenderman has shown great chess so far | Photo: Alina L'Ami
Unfortunately for his supporters, this was shortly before the time control, and by the time move 40 was reached, his advantage had been reduced to a four vs three rook endgame that the Ukrainian held to a draw.
Fabiano Caruana, who made waves by beating Vladimir Kramnik in the very first round, will be kicking himself slightly as he missed a nice opportunity to get ahead in his game against Nils Grandelius.
Grandelius ½-½ Caruana
Though he strived hard to wrest the advantage again, he never quite got the same opportunity and they drew.
One advantage of a big open such as this compared to a tighter round robin is how many colleagues and friends the players get to run into | Photo: Alina L'Ami
One player who did not miss his chance was the young Indian grandmaster Vidit Santosh Gujrathi, who completely swamped his opponent Benjamin Bok.
Fabiano Caruana faced Nils Grandelius in round four | Photo: John Saunders
Benajmin Bok 0-1 Vidit Santosh Gujrathi (annotated by Elshan Moradiabadi)
WGM Enkhtuul Altan had to be a combination of shocked and delighted to get an opportunity to play against Vladimir Kramnik despite her modest start and playing on board 50. | Photo: Alina L'Ami
Hikaru Nakamura pressed hard throughout the entire game and endgame, but was unable to get his Indian opponent, GM S. P. Sethuraman to crack and they drew. | Photo: Maria Emelianova / Chess.com
While the tale of any event, short of a scandal or controversy, is inevitably comprised of the action by the top boards and players, such a large open with such a rich selection of strong players means that there will also be a wealth of fascinating struggles that have no direct effect on the podium, yet deserve to be seen and enjoyed. One such game was between young Argentinian GM Alan Pichot, former World under-16 champion, who overcame his 2700+ opponent, Zoltan Almasi from Hungary in an imbalanced tussle.
Alan Pichot and Zoltan Almasi showed that all games deserve to be checked out, not just the elite | Photo: Maria Emelianova / Chess.com
Alan Pichot 1-0 Zoltan Almasi (annotated by Elshan Moradiabadi)
Hou sits out fifth round
Speaking of controversies, it was with considerable surprise that we saw Hou Yifan paired against her fourth consecutive female player in the Isle of Man Open. This stands out for a variety of reasons, not least of which is Hou Yifan's very public protest in Gibraltar earlier this year, when she refused to play her last round after being paired against no fewer than seven female players. She already refused to take part in the latest Women World Champion cycle, effectively leaving the title open to any and all other colleagues, in order to focus on honing her skills against the best male players in the world. Her campaign and ambitions met with unprecedented success when she took clear first at the Biel GM tournament, but here once more, the pairings gods seem to have conspired against her.
Frustrated by the pairings, Hou has opted to skip round five, receiving instead a half point bye.
"It is difficult to imagine that the organisers are doing this on purpose," she said, reached for cooment via Skype. "There is no reason for that."
Results for round four (top 30)
Top pairings for round five
| Bo. | No. | Name | Pts. | Name | No. |
| 1 | 1 | Carlsen Magnus | 3½ | Granda Zuniga Julio E | 27 |
| 2 | 8 | Eljanov Pavel | 3½ | Kasimdzhanov Rustam | 18 |
| 3 | 12 | Vidit Santosh Gujrathi | 3½ | Lenderman Aleksandr | 46 |
| 4 | 3 | Caruana Fabiano | 3 | Xiong Jeffery | 30 |
| 5 | 4 | Anand Viswanathan | 3 | Grandelius Nils | 28 |
| 6 | 29 | Sargissian Gabriel | 3 | Nakamura Hikaru | 5 |
| 7 | 6 | Adams Michael | 3 | Shirov Alexei | 31 |
| 8 | 33 | Sethuraman S.P. | 3 | Gelfand Boris | 7 |
| 9 | 14 | Short Nigel D | 3 | L'ami Erwin | 34 |
| 10 | 35 | Sokolov Ivan | 3 | Rodshtein Maxim | 15 |
| 11 | 16 | Sutovsky Emil | 3 | Huschenbeth Niclas | 38 |
| 12 | 41 | Tari Aryan | 3 | Leko Peter | 17 |
| 13 | 47 | Pichot Alan | 3 | Rapport Richard | 19 |
| 14 | 20 | Movsesian Sergei | 3 | Deac Bogdan-Daniel | 49 |
| 15 | 21 | Adhiban B. | 3 | Harsha Bharathakoti | 97 |
| 16 | 23 | Jones Gawain C B | 3 | Swapnil S. Dhopade | 55 |
| 17 | 68 | Batsiashvili Nino | 3 | Riazantsev Alexander | 24 |
| 18 | 57 | Perelshteyn Eugene | 2½ | Vallejo Pons Francisco | 9 |
| 19 | 61 | Praggnanandhaa R | 2½ | Howell David W L | 13 |
| 20 | 62 | Brown Michael William | 2½ | Bok Benjamin | 32 |
Update 12:30: We reached Hou Yifan for comment who informed that she is not considering withdrawing at this time, but did request a ½ bye due to her frustration with her pairings thus far. Initially the bye was recorded as a zero point bye, but it was later updated to ½ point along with byes for Arkadij Naiditsch, Laurent Fressinet and FM Michael Babar.
Update 16:00: We received the following information on pairing procedures from the Chief Arbiter, IA Peter Purland:
I can confirm we are using Swiss Manager version 12.0.0.171 (21 June 2017) and Swiss Master version 5.6 (build 12). These programmes do not use the same pairing programme.
Our procedures are that I input results and make pairings with Swiss Manager, Deputy Chief Arbiter IA Arno Eliens does this in Swiss Master independently. Then we compare full pairings of both programmes and only if these are exactly the same do we publish the pairings.
Swiss Manager is based on the JaVaFo algorithm which is standard for Dutch Swiss pairings. See FIDE pairing rules, for details.
Links
- Official tournament site
- ChessBase Live Broadcast (with many games analyzed)
- Complete standings (Chess-Results)
Discussion and Feedback
Join the public discussion or submit your feedback to the editors
If this is not malicious, then the software itself is inadvertently set to pair like-genders.
@psychess as Macauley already pointed out, you are ignoring color. Also, as far as I know the fact that something is not very probable does not prove it is impossible. Someone, who studied statistics, like @EL2400 should know that lack of probability is not lack of possibility. The first round's pairings was described in a Chessbase article:
"In a public drawing of lots, the top players went on stage to draw a random name from a tombola, just like a raffle."
http://en.chessbase.com/post/isle-of-man-open-caruana-beats-kramnik
So, if there was a tombola with the names of participants for the first round, then how could possibly that be a part of a cheating scandal? How can someone add a tombola drawing's probability to the computer pairing's probability, implying that the organizers were cheating? Did they quickly change the tombola to another which contained female names only? Adding the probability of Hou choosing a female name from a tombola to a probability is only good to showing that Hou Yifan's pairing pattern was odd, but the probability of Hou playing a female player in the first round has absolutely no relevance in a debate about a supposed cheating. Due to the infinity of space, time and matter, unprobable events will occur. How big was the chance that you will born in the exact moment you were born at and not a second or more earlier or later? If you calculate that, you will see that the chance of that was way smaller than Hou Yifan's pairing, but you deem this to be normal and the other event, having a way higher probability abnormal. That is scientific bias. Of course, you do not wonder about the exact moment you were born at, since you had to born at a given moment and it happened to be the moment you were born at, but on the other hand, Hou Yifan had to play some players and they happened to be female. I agree with you that it is curious and interesting, but it is certainly not an adequate base of a cheating accusation. And the methodology was clearly faulty, ignoring color and the fact that we can exclude the possibility of a cheating in the first round. If someone can come up with a proof that there was a cheating at other rounds, then your conspiracy theory should be accepted as a fact. But up until then, your assumption is premature and I consider it to be unethical.
"In a public drawing of lots, the top players went on stage to draw a random name from a tombola, just like a raffle."
http://en.chessbase.com/post/isle-of-man-open-caruana-beats-kramnik
So, if there was a tombola with the names of participants for the first round, then how could possibly that be a part of a cheating scandal? How can someone add a tombola drawing's probability to the computer pairing's probability, implying that the organizers were cheating? Did they quickly change the tombola to another which contained female names only? Adding the probability of Hou choosing a female name from a tombola to a probability is only good to showing that Hou Yifan's pairing pattern was odd, but the probability of Hou playing a female player in the first round has absolutely no relevance in a debate about a supposed cheating. Due to the infinity of space, time and matter, unprobable events will occur. How big was the chance that you will born in the exact moment you were born at and not a second or more earlier or later? If you calculate that, you will see that the chance of that was way smaller than Hou Yifan's pairing, but you deem this to be normal and the other event, having a way higher probability abnormal. That is scientific bias. Of course, you do not wonder about the exact moment you were born at, since you had to born at a given moment and it happened to be the moment you were born at, but on the other hand, Hou Yifan had to play some players and they happened to be female. I agree with you that it is curious and interesting, but it is certainly not an adequate base of a cheating accusation. And the methodology was clearly faulty, ignoring color and the fact that we can exclude the possibility of a cheating in the first round. If someone can come up with a proof that there was a cheating at other rounds, then your conspiracy theory should be accepted as a fact. But up until then, your assumption is premature and I consider it to be unethical.
It seems to me that the most plausible way to get an estimate on the actual odds (not that they matter since one can directly verify the pairings' correctness) is to run a very large number of computer simulations, like how Jeff Sonas used to estimate chances of players winning chess tournaments.
I was thinking the same thing as Dr. Nunn said about directly calculating the probability being effectively impossible. Everybody and their dog seems to know just enough about probability to get into some trouble. Okay, 21/160 for a female in IoM round 1 is pretty easy. But the arguments I have seen from then on are not correct. To address a common sort of error... If every single way that Hou could get paired against a female in round 1 led to probability of x of getting another female, then 21/160 * x would indeed be the chances of getting 2/2 women. But this does not apply here, since some possible pairings are more likely to lead to a second female than others. (Whom she faces affects her chances of landing in each score group. Not to mention the fact that her chances of getting a woman in round 2 are highly dependent on the other players' round 1 pairings too.)
I was thinking the same thing as Dr. Nunn said about directly calculating the probability being effectively impossible. Everybody and their dog seems to know just enough about probability to get into some trouble. Okay, 21/160 for a female in IoM round 1 is pretty easy. But the arguments I have seen from then on are not correct. To address a common sort of error... If every single way that Hou could get paired against a female in round 1 led to probability of x of getting another female, then 21/160 * x would indeed be the chances of getting 2/2 women. But this does not apply here, since some possible pairings are more likely to lead to a second female than others. (Whom she faces affects her chances of landing in each score group. Not to mention the fact that her chances of getting a woman in round 2 are highly dependent on the other players' round 1 pairings too.)
@psychess - Aren't you ignoring color? She can't play just anyone in the score group.
@EL2400 - Care to provide details on your (evidently extensive) study of statistics? And what possible motive is there for a conspiracy?
psychess, is correct in his statistical calculations for the year 2017. But in my opinion, he must also multiply by the probability of last year 2016 (7 out of 9). And someone who knows and studied statistics like me knows that this is a cheating scandal. As I wrote it yesterday. After what I wrote, there would be no more cheatings against Hou Yifan. By the way, the poker organizations are world champions in statistics and lotteries. The reason and interest, which I think the organizers have to carry out these cheatings maybe I will explain later.
At the moment, I do not want to explain what I think is the reason.
At the moment, I do not want to explain what I think is the reason.
Thanks NimzoCapa. I just ran it myself and came to the same conclusion. In my first look I underestimated how much of a difference the colors would make.
And I just read the article on the topic on this site, and any doubts I had are now put to rest. What a remarkable coincidence.
And I just read the article on the topic on this site, and any doubts I had are now put to rest. What a remarkable coincidence.
There were 22 females out of 161 participants in the Master’s section. Obviously, Yifan Hou cannot play herself. Therefore, in the first round, where the pairings were random the probability that Hou Yifan would play a woman was 21/160 which is approximately equal to 0.13. It is just approximate because I’m rounding. Yifan drew the game so she could only play 44 people with a score of 1/2 (or if there was an unequal number of people with a score of 1/2 she could have played the lowest rated player who obtained a win or the highest rated played who lost). There were only two other women who drew that round so the probability she would play a woman in both rounds is 21/160 * 2/44 = an approximate probability of .006. In round three, there were four other women and 29 men who had a score equal to Yifan. However, she had already played Kostenkuik in round 1 so she couldn’t play her again. Yifan was paired with a woman and incidentally the other two women were also paired together. Therefore, the approximate probability for round 3 equals .006 * 3/33 = 00054. For round four, the calculation was .00054 * 4/34 so the final approximate probability = 0.000040 or about 1 in 25,000 are the chances that Yifan would play four women in a row.
Probabilities are calculated by estimating probability of the event occurring in each round multiplied by the probabilities in the subsequent rounds.
If she plays in Round 6 she is paired against a man. The probability of her playing a man in this round is 24/25 which is nearly equal to 1 so the probabilities will hardly change at all if she plays a man in round 6.
Probabilities are calculated by estimating probability of the event occurring in each round multiplied by the probabilities in the subsequent rounds.
If she plays in Round 6 she is paired against a man. The probability of her playing a man in this round is 24/25 which is nearly equal to 1 so the probabilities will hardly change at all if she plays a man in round 6.
We dug into this topic a bit more in "Investigating Hou's pairings" http://en.chessbase.com/post/hou-yifan-pairings-controversy-investigation
Whenever X is accusing Y of doing Z, the burden of proof is on X and it is unscientific to assume that X is right, as the null hypothesis should be assumed by default. In our case the accusation several people are describing here is based on the fact that it is extremely unlikely that a woman will be paired with other women only. However, accusations based solely on this are faulty, since if we take the pairings of any participant, we could easily find a pattern which is highly unprobable. Lack of probability is not equivalent with lack of possibility. So, if people are accusing the organisers, then those who do such accusations should prove their claim, since without proof they are playing with the honor of other people, which, in my opinion is highly unethical.
@dhochee Disclaimer: this is pretty tedious and potentially error-prone by hand, but here are some calculations for pairing Hou's round 2 of IoM. Also, I am not certified in pairing, but assuming that the pairing is meant to have people get the color they're due while having the players in as close to the "correct" rating order as possible, it will basically have to get the result as I describe it below.
There are an odd number of players with 1/1, so the top 0.5/1 (Gelfand) gets paired against the lowest 1/1 (Fischer). I count 44 remaining people with 0.5/1, so among this group they are paired 1 vs. 23, 2 vs. 24, etc. The "natural" pairings of the remaining 0.5/1 people (cut down to the part that interacts with Hou's pairing) are:
*Paehtz (2453) - Movsesian (2671)
Kiewra (2433) - Adhiban (2670)
Hou (2670) - Sundararajan (2426)
Raja (2423) - Jones (2668)
Riazantsev (2666) - *Arjun (2406)
I put an asterisk by people with the wrong color. We need to fix this in the least damaging way possible. The most "obvious" thing to do would be swap Paehtz and Arjun. This would leave:
Arjun (2406) - Movsesian (2671)
Kiewra (2433) - Adhiban (2670)
Hou (2670) - Sundararajan (2426)
Raja (2423) - Jones (2668)
Riazantsev (2666) - Paehtz (2453)
Now everybody has the right color but the ratings are rather out of order. We can improve this by rearranging the three 2400's with White and the two with Black to be in descending order:
Kiewra (2433) - Movsesian (2671)
Raja (2423) - Adhiban (2670)
Hou (2670) - Paehtz (2453)
Arjun (2406) - Jones (2668)
Riazantsev (2666) - Sundararajan (2426)
Can't do any better than this -- everybody has the color they are due and the rating order can't be improved without messing up the colors. We have reproduced Hou's round 2 pairings. That took forever, so I don't think I'll be trying it for future rounds, but it just goes to illustrate how funny-looking things can happen.
There are an odd number of players with 1/1, so the top 0.5/1 (Gelfand) gets paired against the lowest 1/1 (Fischer). I count 44 remaining people with 0.5/1, so among this group they are paired 1 vs. 23, 2 vs. 24, etc. The "natural" pairings of the remaining 0.5/1 people (cut down to the part that interacts with Hou's pairing) are:
*Paehtz (2453) - Movsesian (2671)
Kiewra (2433) - Adhiban (2670)
Hou (2670) - Sundararajan (2426)
Raja (2423) - Jones (2668)
Riazantsev (2666) - *Arjun (2406)
I put an asterisk by people with the wrong color. We need to fix this in the least damaging way possible. The most "obvious" thing to do would be swap Paehtz and Arjun. This would leave:
Arjun (2406) - Movsesian (2671)
Kiewra (2433) - Adhiban (2670)
Hou (2670) - Sundararajan (2426)
Raja (2423) - Jones (2668)
Riazantsev (2666) - Paehtz (2453)
Now everybody has the right color but the ratings are rather out of order. We can improve this by rearranging the three 2400's with White and the two with Black to be in descending order:
Kiewra (2433) - Movsesian (2671)
Raja (2423) - Adhiban (2670)
Hou (2670) - Paehtz (2453)
Arjun (2406) - Jones (2668)
Riazantsev (2666) - Sundararajan (2426)
Can't do any better than this -- everybody has the color they are due and the rating order can't be improved without messing up the colors. We have reproduced Hou's round 2 pairings. That took forever, so I don't think I'll be trying it for future rounds, but it just goes to illustrate how funny-looking things can happen.
I understand her ambitions, but if she starts to win against every female player with ease, she may start to complain. Making irrational moves or skipping rounds is irresponsible.
Good news for Hou Yifan:
For round 6 she is paired against a male GM from India.
She has White, and his rating is modest (about 190 Elo points
below her own). So, her chances are good that she will win,
giving her a rise out of the cloud of other female players.
For round 6 she is paired against a male GM from India.
She has White, and his rating is modest (about 190 Elo points
below her own). So, her chances are good that she will win,
giving her a rise out of the cloud of other female players.
dhochee: There were four other women with equal scores to Hou Yifan in round 3. Four of the women were paired together. There were 30 men with identical scores. Only Kostenuik played a man. The pairings for this round seem even stranger.
@dhochee:
Your lines about the pairings in round 2 are misleading.
You did not take in to account colors.
Among all players which had 0.5 points after round 1, two
groups were built: group A are those with White in round 1 and
group B those with Black in round 1. Both groups were
ordered according to Elo ratings. And then (almost) only
pairings between group A and group B were made. Having
this in mind, everything looks fine:
http://chess-results.com/tnr303618.aspx?lan=1&art=2&rd=2&fed=NOR&flag=30&wi=821
Your lines about the pairings in round 2 are misleading.
You did not take in to account colors.
Among all players which had 0.5 points after round 1, two
groups were built: group A are those with White in round 1 and
group B those with Black in round 1. Both groups were
ordered according to Elo ratings. And then (almost) only
pairings between group A and group B were made. Having
this in mind, everything looks fine:
http://chess-results.com/tnr303618.aspx?lan=1&art=2&rd=2&fed=NOR&flag=30&wi=821
@pompeaux Ah, okay. You're right, that tidbit about Swiss Manager and Swiss Master having the same outputs at IoM has no relevance to whether the Gibraltar pairings were a software glitch. For some reason I thought we were talking about the possibility of the same glitch happening in both tournaments, which is why I thought that was relevant. My apologies for the confusion.
Actually, after further consideration of GM Sambuev's pairing notes, and looking at pairings in a couple of rounds of IoM, it is not at all clear that there isn't manipulation. In fact, in almost every round (both in Gibraltar and IoM) there are pairings that do not result simply from rating. I understand that there are other factors that influence the pairings, but it is not sufficient to point out that the ratings of Hou's opponents were roughly in line with nearby boards.
Consider these seven pairings for round 2 of IoM, with three above and three below Hou's:
2675 - Rapport Richard
2456 - Kjartansson Gudmundur
2671 - Movsesian Sergei
2433 - Kiewra Keaton
2670 - Adhiban B.
2423 - Raja Harshit
2670 - Hou Yifan (F)
2453 - Paehtz Elisabeth (F)
2668- Jones Gawain C B
2406- Arjun Kalyan
2666 - Riazantsev Alexander
2426 - Sundararajan Kidambi
2662 - Akobian Varuzhan
2394 - Harsha Bharathakoti
A simplistic expectation would be that the ratings of the lower player would gradually decrease just like the ratings of the higher player, but that's not what we see. Of those 14 players, only 2 were women, and they were paired against each other, even though it didn't mesh with the ratings.
I'm looking forward to more evidence. I do agree, however, that if two programs match the output with no settings configured to give consideration to gender, it is still likely to be chance, but I'm starting to think Masquer might be correct.
Consider these seven pairings for round 2 of IoM, with three above and three below Hou's:
2675 - Rapport Richard
2456 - Kjartansson Gudmundur
2671 - Movsesian Sergei
2433 - Kiewra Keaton
2670 - Adhiban B.
2423 - Raja Harshit
2670 - Hou Yifan (F)
2453 - Paehtz Elisabeth (F)
2668- Jones Gawain C B
2406- Arjun Kalyan
2666 - Riazantsev Alexander
2426 - Sundararajan Kidambi
2662 - Akobian Varuzhan
2394 - Harsha Bharathakoti
A simplistic expectation would be that the ratings of the lower player would gradually decrease just like the ratings of the higher player, but that's not what we see. Of those 14 players, only 2 were women, and they were paired against each other, even though it didn't mesh with the ratings.
I'm looking forward to more evidence. I do agree, however, that if two programs match the output with no settings configured to give consideration to gender, it is still likely to be chance, but I'm starting to think Masquer might be correct.
I completely agree with moderncheckers! Well put, exactly my thoughts!
Thanks NimzoCapa. You answered my questions.
Masquer, while that option may be available in the software, that would obviously be reflected in the pairings for all players, not just for Hou. That clearly isn't the culprit here.
Actually, the thread with GM Sambuev's manual pairings revealed something very interesting that helps explain why this is happening. In the Dutch system for Swiss pairings, the top half of the field is paired against the bottom half. Of the 161 players in the Masters section at IoM, Hou is ranked 21, meaning that she will tend to play those with the same score who are closest to about rank 100, with a rating of ~2400, where most of the women masters are ranked. While it's still statistically improbable for her to be repeatedly paired against other women, this would increase the odds.
Masquer, while that option may be available in the software, that would obviously be reflected in the pairings for all players, not just for Hou. That clearly isn't the culprit here.
Actually, the thread with GM Sambuev's manual pairings revealed something very interesting that helps explain why this is happening. In the Dutch system for Swiss pairings, the top half of the field is paired against the bottom half. Of the 161 players in the Masters section at IoM, Hou is ranked 21, meaning that she will tend to play those with the same score who are closest to about rank 100, with a rating of ~2400, where most of the women masters are ranked. While it's still statistically improbable for her to be repeatedly paired against other women, this would increase the odds.
@NimzoCapa - They are using a different version of the software at Isle of Man, than they did at Gibraltar, with many bug fixes. That's why it doesn't make any sense to compare what they are doing now to what happened at Gibraltar. But thanks for not going into details!
@Masquer - I think it's strange that such programs require player categories of Male or Female for an Open. Would you expect a different result if the pairings were re-tested using the same programs but while registering Yifan as a Male?
@Masquer - I think it's strange that such programs require player categories of Male or Female for an Open. Would you expect a different result if the pairings were re-tested using the same programs but while registering Yifan as a Male?
As a tournament director, I am very familiar with Swiss pairing programs and I can tell you that such a program CAN be adjusted to ensure female vs female pairings whenever possible. All it takes is checking off a single box to trigger a preference for such pairings.
This feature could also be hidden in the publicly-available version of Swiss Manager. One would have to audit the actual software used by the organizers to find out exactly what's going on. Whoever thinks this is just a coincidence is just naive.
This feature could also be hidden in the publicly-available version of Swiss Manager. One would have to audit the actual software used by the organizers to find out exactly what's going on. Whoever thinks this is just a coincidence is just naive.
You were concerned about the possibility of the Gibraltar program being bugged. Another program matching its output shows there is no such bug.
Look at the colors in the pairings to explain rating deviations in the pairings. Small swaps in rating, accroding to specific rules, are done to equalize colors. I do not wish to go over details that people explained at length when it happened, but you can find further explanations there and in the chessbase comments for the original article.
Look at the colors in the pairings to explain rating deviations in the pairings. Small swaps in rating, accroding to specific rules, are done to equalize colors. I do not wish to go over details that people explained at length when it happened, but you can find further explanations there and in the chessbase comments for the original article.
@NimzoCapa, What does the selection of two programs for "this tournament" have to do with Gibraltar pairing integrity? Also, in your citation, GM Sambuev stated that Hou was paired according to ratings in Gibraltar, but Piorun Kacper 2651 had a rating identical to Ms.Hou. Plus, they had the same score through 3 rounds, yet Hou was paired with women each round, whereas Piorun was paired with men.
The organizers in this tournament are using two different pairing programs that give identical pairings -- see the 16:00 update. (No good reason they'd lie about this since such a claim is easily verifiable/falsifiable.) There would have to be an identical bug in both programs, which seems hard to believe. If just Swiss Manager had some glitch where it handles female players wrongly, it would not match the output of the other program here.
Also, GM Sambuev (among others) has checked the Gibraltar pairings by hand:
http://forum.chesstalk.com/showthread.php?15226-Gibraltar-2017&p=110537#post110537
Also, GM Sambuev (among others) has checked the Gibraltar pairings by hand:
http://forum.chesstalk.com/showthread.php?15226-Gibraltar-2017&p=110537#post110537
@NimzoCapa - Actually, that doesn't verify the correctness of Hou's pairings in Gibraltar because it doesn't control against a glitch in the software. TheJoker would have to run the same test on some other FIDE-sanctioned Swiss management software in order to verify the correctness of the pairings by Swiss Manager.
@dhochee Multiple people reported that they verified the correctness of Hou's pairings against women at Gibraltar, some by hand and some by running the programs. For instance, here is user TheJackpot's post on the ChessBase comments a few months ago:
"Took the SwissManager tournament file from chess-results.com, created a TRF / FIDE rating report file, imported it, verified the pairings.
Round 1: differences, which is to be expected: people show up late, ratings get corrected, mistakes fixed, etc.
Round 2, 3, 4: equal to the pairing in Gibraltar
Round 5: a few differences in the group of people with 1.5 and 1 out of 4, nowhere near Hou. My educated guess: results of previous rounds were corrected after round 5 was paired
Round 6, 7, 8: equal to pairing in Gibraltar
Round 9: in the lower echelons 2 pairings were adjusted (the black players exchanged), due to (probably) israeli not playing irani
Round 10: equal to pairing in Gibraltar."
"Took the SwissManager tournament file from chess-results.com, created a TRF / FIDE rating report file, imported it, verified the pairings.
Round 1: differences, which is to be expected: people show up late, ratings get corrected, mistakes fixed, etc.
Round 2, 3, 4: equal to the pairing in Gibraltar
Round 5: a few differences in the group of people with 1.5 and 1 out of 4, nowhere near Hou. My educated guess: results of previous rounds were corrected after round 5 was paired
Round 6, 7, 8: equal to pairing in Gibraltar
Round 9: in the lower echelons 2 pairings were adjusted (the black players exchanged), due to (probably) israeli not playing irani
Round 10: equal to pairing in Gibraltar."
A couple comments on why conspiracy theories are utterly dismissible here. (Not to be construed as anti-Hou, one of my favorite players. She is understandably frustrated with bad luck, not asking fans to make wild accusations on her behalf.)
1. If anything, an unscrupulous tournament organizer wishing to manipulate pairings would want to give Hou as high-rated of opponents as possible. Spectators want to see Hou face the best players, which is why organizers invite her to way more invitational tournaments than anybody else around her rating. A game between Hou and another woman has zero intrigue to most spectators. She either wins or the game is dismissed as a meltdown, not exciting either way. On the other hand, Hou vs. 2800 generates immeasurably more publicity whenever it happens than any other sub-2750 player vs. 2800. (I guess maybe if the organizers wanted a surefire scandal for publicity, they could pair Hou vs. women intentionally. But scandal discourages strong players from registering for future editions, so this also seems against their best interest.)
2. An organizer manipulating pairings in a big tournament would be the easiest thing to catch ever. Anybody who has this software can see for themselves with 100% certainty whether the pairings have been manipulated. (Swiss Master can be downloaded for free, and tournament directors have also been using the non-free Swiss Manager for ages.)
3. Those concerned with the issue of the programs being closed source... are you seriously suggesting the programmers are in on the anti-Hou conspiracy? For instance, a quick Google search shows that the author of the Swiss Manager program wrote the first version of the code well before Hou was born, and has no discernible affiliation with Isle of Man or FIDE. And the author of Swiss Master would also have to put in anti-Hou code too, for the program outputs to match. In fact, they'd have to be working together, to make sure their anti-Hou code works identically. Quite crazy.
1. If anything, an unscrupulous tournament organizer wishing to manipulate pairings would want to give Hou as high-rated of opponents as possible. Spectators want to see Hou face the best players, which is why organizers invite her to way more invitational tournaments than anybody else around her rating. A game between Hou and another woman has zero intrigue to most spectators. She either wins or the game is dismissed as a meltdown, not exciting either way. On the other hand, Hou vs. 2800 generates immeasurably more publicity whenever it happens than any other sub-2750 player vs. 2800. (I guess maybe if the organizers wanted a surefire scandal for publicity, they could pair Hou vs. women intentionally. But scandal discourages strong players from registering for future editions, so this also seems against their best interest.)
2. An organizer manipulating pairings in a big tournament would be the easiest thing to catch ever. Anybody who has this software can see for themselves with 100% certainty whether the pairings have been manipulated. (Swiss Master can be downloaded for free, and tournament directors have also been using the non-free Swiss Manager for ages.)
3. Those concerned with the issue of the programs being closed source... are you seriously suggesting the programmers are in on the anti-Hou conspiracy? For instance, a quick Google search shows that the author of the Swiss Manager program wrote the first version of the code well before Hou was born, and has no discernible affiliation with Isle of Man or FIDE. And the author of Swiss Master would also have to put in anti-Hou code too, for the program outputs to match. In fact, they'd have to be working together, to make sure their anti-Hou code works identically. Quite crazy.
There are many comments around the line: "it is easy to verify what the parings would be using the software paring program used by the organizers".
But I did not see any such verification as of yet. Maybe I did not look well enough.
But I did not see any such verification as of yet. Maybe I did not look well enough.
As a statistician, a very quick mathematical calculation suggests that the probability of Hou Yifan's actual oppponents arising at Gibraltar/IOM being female was around 1 in 3 billion, give or take few - sufficiently infinitesimal to warrant an investigation! But as a qualified arbiter, I sympathise with tournament organisers and would wholeheartedly defend their reliance on what are very good pairing programs.
It seems probable therefore that there is indeed a problem, perhaps previously undetected, in the pairing program software and/or possibly with the way female records are identified when entered. It would be very prudent of the organisers to continue with manual pairings for the remaining rounds, then to conduct software and user compliance tests after the event.
It seems probable therefore that there is indeed a problem, perhaps previously undetected, in the pairing program software and/or possibly with the way female records are identified when entered. It would be very prudent of the organisers to continue with manual pairings for the remaining rounds, then to conduct software and user compliance tests after the event.
As others have noted, the rules for Swiss pairing are clearly defined, to the point that two different software programs will produce the same pairings. It should be relatively easy to verify whether or not Hou is the victim of manipulation or crazy chance.
I had the same thought when this happened in Gibraltar, and I'm curious if an unbiased investigation was ever performed. It seems highly unlikely, although possible, that she is simply getting paired against women by chance.
I had the same thought when this happened in Gibraltar, and I'm curious if an unbiased investigation was ever performed. It seems highly unlikely, although possible, that she is simply getting paired against women by chance.
I'm bracing myself for another Fool's Mate from Yifan...
@Keith Homeyard: The motto says exactly that we should be united and thus it should not make a difference to us who the person against whom we're playing is.
@Aighearach: There is no logical connection between the protasis and the apodosis in your first sentence of the last paragraph. And to play the game professionally is to concentrate on playing every game as well as you can instead of using up your energy on questioning the pairings. If it is a manipulation, in which I doubt, then it is a most harmless one. I mean, in which way exactly are these pairings any worse for her then any generic pairings in which she plays players with the same number of points as her?
@Aighearach: There is no logical connection between the protasis and the apodosis in your first sentence of the last paragraph. And to play the game professionally is to concentrate on playing every game as well as you can instead of using up your energy on questioning the pairings. If it is a manipulation, in which I doubt, then it is a most harmless one. I mean, in which way exactly are these pairings any worse for her then any generic pairings in which she plays players with the same number of points as her?
@moderncheckers: It is absolutely normal in an open tournament to get a half point bye, and if there is some controversy over the pairings it is often used as a solution.
Other notable players who received half point byes are listed above in the article. In local 5 round opens it is common for people to take 1 or 2 half point byes to make their schedule work. Sometimes if you pre-register for the tournament and don't check in on time before the first round, you get a half point bye as long as you show up before pairings are completed for round 2.
If you don't understand the problems with her having her pairings manipulated, perhaps you're not really prepared to feel vicariously offended either? This is about having an opportunity to play the game professionally, it isn't about people's feefees.
Other notable players who received half point byes are listed above in the article. In local 5 round opens it is common for people to take 1 or 2 half point byes to make their schedule work. Sometimes if you pre-register for the tournament and don't check in on time before the first round, you get a half point bye as long as you show up before pairings are completed for round 2.
If you don't understand the problems with her having her pairings manipulated, perhaps you're not really prepared to feel vicariously offended either? This is about having an opportunity to play the game professionally, it isn't about people's feefees.
I totally disagree with modern checkers, it's not a matter of colour ,sex or anything else to me - it's most important whether there is any manipulation rather than simple human adjustment to the pairings! Remember the FIDE motto Gens una sums , is this a good example?
In 2012, it wasn't Hou Yifan who was screwed, it was Judit. She was paired with Hou Yifan, Humpy and Mariya Muzychuk. Although for 2012 it was quite exciting for the #1 women to be paired with the #2 and #3 women of the time.
Can anyone just skip a round, saying
"I protest against pairing me against unusually many [XXX]",
as [XXX] insert any of the following:
women/men/white people/black people/Muslims/Buddhists/Christians/foreigners/citizens of country XYZ/fat people/thin people/bald people/black-haired people/old people/[label of any other group, especially a label of a minority group in the given environment because then even a short series of games against them will seem statistically significant],
and even get a half-point refund?
How is such protesting not discriminatory, no matter if the observation was technically correct or not and if the person protesting was themselves a member of said group or not?
And exactly what difference does it make against a person of what sex or against a member of what social group one is playing? What is so appalling in playing against a member of a "wrong" group? Can some of the "FIDE-is-cheating"-theorists explain it to me?
I am inclined to feel vicariously offended on behalf of other woman players in the tournament because it seems to me as if by her protest Hou had meant "I am better than all of you, and I don't want to play against you any more".
And by the way, no matter against whom you're paired, if you win all the games, you win the tournament, and thereby best show your strength.
"I protest against pairing me against unusually many [XXX]",
as [XXX] insert any of the following:
women/men/white people/black people/Muslims/Buddhists/Christians/foreigners/citizens of country XYZ/fat people/thin people/bald people/black-haired people/old people/[label of any other group, especially a label of a minority group in the given environment because then even a short series of games against them will seem statistically significant],
and even get a half-point refund?
How is such protesting not discriminatory, no matter if the observation was technically correct or not and if the person protesting was themselves a member of said group or not?
And exactly what difference does it make against a person of what sex or against a member of what social group one is playing? What is so appalling in playing against a member of a "wrong" group? Can some of the "FIDE-is-cheating"-theorists explain it to me?
I am inclined to feel vicariously offended on behalf of other woman players in the tournament because it seems to me as if by her protest Hou had meant "I am better than all of you, and I don't want to play against you any more".
And by the way, no matter against whom you're paired, if you win all the games, you win the tournament, and thereby best show your strength.
In the report of the previous round, it was mentioned that "Hou Yifan, incorrectly of course" suspected something to be wrong with the pairings. The expression used was so absolute that I assumed that the author had received an exact copy of the executable of Swiss Manager used in the tournament, had it disassembled by a specialist, who analysed it and opined that there is was no gender bias in the pairing algorithm. Only, the link to that analysis was missing. Can we have it now?
If what I wrote is not clear, the idea is that the person running the software could simply forbid the pairings around Hou's standing that *don't* pair her with a woman.
And of course, if they wanted to be really brazen, the software does allow manual pairing:
(again, from the Swiss Manager handbook)
"Pairing Players in Swiss System Tournaments
If it is required by organizational reasons to pair players manually, is this possible with {Pairing/Set
player...} or {Pairing/Set new player...}. The difference between the menu items is that with
{Pairing/Set player...} you can set the players for the next round and with {Pairing/Set new player...}
you can set/change the players from the first to the current round."
And of course, if they wanted to be really brazen, the software does allow manual pairing:
(again, from the Swiss Manager handbook)
"Pairing Players in Swiss System Tournaments
If it is required by organizational reasons to pair players manually, is this possible with {Pairing/Set
player...} or {Pairing/Set new player...}. The difference between the menu items is that with
{Pairing/Set player...} you can set the players for the next round and with {Pairing/Set new player...}
you can set/change the players from the first to the current round."
@Macauley JaVaFo appears to be closed source (it's written in Java, but we don't get to see the Java), as is Swiss Manager, which IoM claims to use.
The Swiss Manager handbook, here http://swiss-manager.at/unload/SwissManagerHelp_ENG.pdf , states that there is an option to set "Forbidden Pairings". Therefore, the tournament could have easily manipulated the pairings such that Hou faces women, provided there is a woman remotely close in standings.
That said, if someone has the full version and simply runs the pairings through it, yielding the same pairings that Hou faced in the tournament, bias could be disproven.
The Swiss Manager handbook, here http://swiss-manager.at/unload/SwissManagerHelp_ENG.pdf , states that there is an option to set "Forbidden Pairings". Therefore, the tournament could have easily manipulated the pairings such that Hou faces women, provided there is a woman remotely close in standings.
That said, if someone has the full version and simply runs the pairings through it, yielding the same pairings that Hou faced in the tournament, bias could be disproven.
Strange that it is FIDE itself that is cheating, and not the players. This is what happens when you set up a system that ensures corruption.
The only solution is to create a new international chess organization where the votes are based on the cumulative ratings of the players in your country, instead of giving each country one vote, even countries that don't care about chess. Obviously many of those countries will vote based on politics or kickbacks.
The only solution is to create a new international chess organization where the votes are based on the cumulative ratings of the players in your country, instead of giving each country one vote, even countries that don't care about chess. Obviously many of those countries will vote based on politics or kickbacks.
@billvan61 The pairing procedure is transparent. Every one can check the pairings:
pairings.fide.com/images/stories/downloads/newsectionc04web.pdf.
pairings.fide.com/images/stories/downloads/newsectionc04web.pdf.
