Visually handicapped Stein Bjørnsen (50) has received a two-year ban on domestic competition from the Central Board of the Norwegian Chess Federation (NSF). The verdict fell at a board meeting in Stavanger last weekend.
by Tarjei J. Svensen and Jonathan Tisdall
Bjørnsen has received the toughest penalty available to the federation.
On the basis of the facts presented, the Central Board of the Norwegian Chess Federation views that it is beyond reasonable doubt that there has been a violation of FIDE’s Laws of Chess, paragraphs 11.3a and 11.3b, whereby Bjørnsen, by using a Bluetooth earphone, has received information either from a chess analysis program or by external assistance, and that this has taken place at least during the National Championship in 2015 and the TV2 Chess International 2015/2016.
The man from Horten has also been forbidden from holding any official position in Norwegian chess, according to the federation verdict. Bjørnsen is currently deputy leader of Horten Chess Club.
All of his games played between July 2015 and January 2016 will be adjudged lost, NSF President Bjørn Salvesen said in a press release. The NSF also reported that the verdict had been reported to International Chess Federation FIDE.
Background
The 50 year old caught the attention of Norwegian chess circles last year. It started in the summer of 2015 when he won the Class 2 section (Norwegian Elo 1500-1750) as an unrated player with an impressive score of 8 out of 9.
He continued his progress by winning Hans Olav Lahlum’s Elo tournament in Fagernes last Christmas, with an undefeated 8/9 and a rating performance of 2267. That was an achievement, considering just a year earlier he scored 50 percent in the lowest section of the Norwegian Championship for the Blind, a performance at novice level.
His Fagernes performance in raised concern from opponents during the tournament, but the arbiters failed to address an important detail: Bjørnsen was playing with an electronic earpiece that proved to be a Bluetooth device. His explanation was that he wanted to record moves in order to be able to play them back later, but tech experts told Matt & Patt that the headset seen on photos during his games, is unable to connect to a recorder.
Without losing a single game, he continued his progress by winning Hans Olav Lahlum’s Elo tournament in Fagernes last christmas. His rating performance was 2267, winning with 8 out of 9. That was an achievement, considering he just a year earlier had finished a tournament in the lowest section of the Norwegian Championship for the blind scoring 50 percent, a performance that equals novice level.
His performance in Fagernes had already raised concern among his opponents during the tournament, but the arbiters failed to address an important detail: Bjørnsen was playing with an electronic earpiece that proved to be a bluetooth device. His explanation was that he wanted to record moves in order to be able to play them back later, but tech experts told Matt & Patt that the headset seen on photos during his games, are unable to connect to a recorder.
Analysis later showed that his games had an overwhelming correlation with an engine. According to analysis made by Ole Valaker, himself a master level player, journalist and chess teacher, Bjørnsen had a 96% match with powerful computer engine Rybka.
IM Atle Grønn, who also commented publicly about what he considered to be «a very computerish style», said that in the 298 moves and 18 games, Bjørnsen didn’t make a single tactical error.
GM Artur Kogan, who is actively involved in the battle against computer cheating, said to Matt & Patt:
«It’s completely clear that he has cheated here. Especially his game in round 8 leaves little doubt. I also want to add that it’s suspicious that a player at 2000 level plays like a grandmaster with perfect play in the critical phases, so this is a case that will be examined thoroughly.»
Speaking to Matt & Patt, Bjørnsen denied any cheating, claiming he had made the progress from beginner level to master by «playing through Magnus Carlsen and Viswanathan Anand’s games» and «studying chess for up to 10 hours a day» for months.
When the story broke, it caused headlines in Norwegian mainstream media. Bjørnsen appeared on national TV channels maintaining his innocence.
To NRK he said «My way of playing and learning chess, is different than others. I memorize all ways on the board and have to admit that my level of understanding is pretty average for a chess player. I’m good at remembering analysis made in advance.»
The Norwegian Chess Federation launched an investigation into the case, eventually suspending him until he had gone through a test. Bjørnsen initially said that he would be willing to take any test to prove himself, «naked if necessary», but eventually changed his mind stating he would prefer to be tested in a tournament.
To TV 2 in February, he said «I have no choice but to accept the test, but I demand that it take place when I’m ready for it – and that won’t be for a while. I haven’t touched a chess piece for one month, something that makes my level of play drop at once. I also won’t accept any chess theory in the test. What I’m good at, is moving pieces.»
Bjørnsen «I haven’t cheated consciously.»
The Norwegian Chess Federation finally made their decision in the case last weekend.
«I am shocked that I have not had a chance to respond before the verdict,» Bjørnsen told newspaper VG. «We want to respond to the report but we can’t manage it in four days.»
Bjørnsen has consistently denied cheating. When Matt & Patt spoke to him after the charges emerged, the 50-year-old said he understood that the accusations were coming.
«Yes, I understand, and that was why there so many checks (during the tournament) and (International Arbiter) Lahlum said to me that he expected (accusations) to follow.»
Bjørnsen also admitted that he was previously banned for doping in power-lifting.
«Yes, I was banned for a while, because I should have presented this earlier, so it was my fault in a way. But it didn’t become a case,» Bjørnsen cryptically told TV 2.
After the NSF verdict Bjørnsen was asked by VG why so many chess players were suspicious of him.
«Because I have made massive progress and because I use equipment – while I improve. All blind players have used similar equipment for years, but they tend to come last,» he told the newspaper.
He also had an explanation for TV 2 after the federation report.
«I still assert that I have not cheated!» Bjørnsen said. «I understand that it looks strange to have used Bluetooth earphones, but that is a long way from getting information in my ear. I haven’t gotten any. I haven’t cheated consciously.»
The reasons behind the Bjørnsen ban
The NSF has published three reasons behind the Bjørnsen ban: Photos that show with high probability that Bluetooth earphones have been used, games that demonstrate the use of a computer and the rejected offer to undergo testing.
The clearest evidence is the 18 games that were analyzed by Professor Kenneth Regan, the man who has developed special anti-cheating software for FIDE.
Regan operates with three levels associated with correspondence with computer moves: if the coefficient is under 2.75 the play is human and there is no cheating; from 2.75 to 4.75 there is a high probability of cheating; over 4.75 is patently cheating.
In the report, which the NSF cites, Regan concluded: «Full test of both tournaments is arguably over 5.00.»
«My work is «data-driven» – it tallies myriad games by players of all levels, originally down to 1600 but now to 1000,» Regan explains to Matt & Patt. «Each level has a characteristic rate of agreement with various top engines and rate of error as judged by them. The rates give a strong linear fit to Elo ratings, and importantly, the variances behave as in so-called «normal distribution.»
How does Bjørnsen’s score compare to other players accused of cheating?
«That I cannot say except a generic (and true) remark that scores are all over the map, and that over 2/3 of allegations/whispers that come to my attention produce scores that «exonerate» and explain,» Regan said.
FIDE in its report writes that it would have been useful to also have tournaments between the two Norwegian events analyzed. Games have not been available from a tournament in Germany, the Braille Baltic Cup 2015, so the analysis was limited to the 18 games from the two Norwegian events.
In the NSF report Bjørnsen’s extreme rating improvement in just 18 months is also discussed. Comparing his rather weakly played games in the autumn of 2014 to the master class production in the TV 2 International in Fagernes, a highly unusual development is charted. Bjørnsen has now played 50 FIDE-rated games without loss. According to FIDE calculations, Bjørnsen leapt up to 1750 at the Norwegian Championship in 2015 and up to 2075 at Fagernes.
The NSF commented on the earphones:
«Bjørnsen claims that the earphones used during Fagernes disappeared after the last round, and there is no use in examining these. The original Jaybird box was sent to the NSF. A relatively large amount of photo evidence from the event leaves no doubt that Jaybird Bluetooth earphones were used at the Norwegian Championship, the Horten Championship and the TV2 Chess International.»
Bjørnsen’s lawyer: «Frightening process»
Stein Bjørnsen’s legal counsel Øivind Rief Aarmo is upset that they have not had sufficient time to consider the verdict.
«We were sent this last Monday along with the associated material. We were then told that a decision was due the following Saturday, five days later. We didn’t like this since the material was primarily complete in January, except for one report from February 23.»
«We have requested feedback for a long time, since February, and had none until five days before he decision was to fall. As a lawyer I find this a completely frightening process.»
«I cannot understand how one can proceed this way. If there is to be a real opportunity for appeal, one has to be given a chance to examine the evidence. Also to have a chance to gather alternative documentation, in the case of complicated calculations and statistics and probability, as there were here, so that we could verify it,» Aarmo said.
Aarmo added that his client was terribly disappointed with the verdict.
«He completely disagrees with this. When we saw the schedule we understood that there was no real chance to protest before the decision was processed. He had hoped that the federation would see that and delay the proceedings and give us a chance to give input. That was his first reaction,» Aarmo said.
Bjørsen has already stated to VG and state broadcaster NRK that he will appeal the verdict.
NSF president: «Must accept that the proceedings took time.»
NSF President Bjørn Salvesen admitted to a time consuming process, and had this to say to Aarmo’s comments:
«They wanted more material than we had prepared. They would not accept this, but they have to accept that the NSF has taken the time it has. We have been extremely careful not to go wrong, this is virgin territory since there are no FIDE rulings to refer to. We are treading new ground so it was important for us to use the time needed and we have gathered everything in a report, which was ready a week in advance,» Salvesen said.
«For me, Stein Bjørnsen has been a member of the NSF who has been innocent until the verdict last Saturday. This is also how we have treated him. We have had telephone conversations, he received the FIDE report as soon as it was finished. He got the FIDE report immediately.»
«The lawyer came into this later. He got a chance to respond. We have received a lot of information. The report was ready a week before the Central Board meeting on Saturday and was sent to Bjørnsen and his lawyer. The elements in the verdict are the FIDE report, the use of illegal equipment and that he has not cooperated in taking a test,» Salvesen added.
«The claim that he got the information just a week before is taking things to extremes, because the main factors have been know the entire time – the report itself, the final edits and adjustments were sent a week before.»
Could this have been done differently, or are you satisfied with how the NSF has handled the matter?
«Considering that this is new ground it has gone quite well. If we get another such case it will go faster since we now have a precedent. But this is a very special case with many delicate aspects. We took the time we needed,» Salvesen said.
The NSF president explained that there was a test ready for Bjørnsen and he originally agreed to it, but he reversed his position.
«What happened was that Stein Bjørnsen confirmed that he would take part in a test after a decision by the Central Board. We selected qualified people and we contacted Bjørnsen and arranged a meeting between him and these people, and then we received a letter from his lawyer that a test was out of the question.»
«We saw no reason to proceed with this after such a direct refusal. We might have expected some demands regarding the testing.»
There were no negotiations about how to carry out the testing?
«No, we didn’t get that far. The lawyer simply rejected any such procedure. We found this regrettable, because that halted our process for him.»
Salvesen objected to Bjørnsen’s argument that he did not receive information about how the test would be administered. It was precisely this he was invited to discuss, the president said.
«When we contacted Bjørnsen after establishing a test committee, it was so he could make suggestions about how to take into account his special situation as visually handicapped, and also to accommodate a time period where the test could be done. The argument that he didn’t want to take the test because he didn’t know what he was getting into collapses when he rejected the offer to take part in forming the test,» Salvesen said.
Urkedal: «Wishful thinking that he hasn’t cheated.»
The Norwegian chess scene has been eagerly awaiting the NSF decision on the much-discussed Bjørnsen case. Earlier this year the 50-year-old was quarantined after widespread accusations of cheating. Bjørnsen preferred to answer critics by playing rather than being tested.
The affair has attracted national – and international – media attention. One public voice among the accusers was Norway’s latest grandmaster, 22-year-old Frode Urkedal.
«I would say that it is wishful thinking to believe that he has not used computer assistance. It seems that the media are very obsessed with what equipment he has used, but that is not primarily why he is being accused. It is the games,» Urkedal told Matt & Patt earlier.
Bjørnsen has four weeks to appeal the verdict to the NSF’s Rules Committee (RU), which will then pass their own judgment. This process occurred the previous time a cheating case was tried.
«He has four weeks to appeal from the time of publication. This is a good legal safety valve in our system,» Salvesen said, and points out that the case will then be considered by completely different people.
The RU is led by International Arbiter Hans Olav Lahlum, with IM Eirik Gullaksen and Arne Tjølsen as members. Since Lahlum was tournament director of the event Bjørnsen won in Fagernes he has disqualified himself, and a deputy – either Øyvind Bentsen or Trond Romsdal, will take his place.
Is it fair to say, that rather than «patent cheating», all the FIDE Report (Regan) did is show that Bjornsen performed «patently above» the expectation from his presumed Elo? If he is really improving, is this a surprise? Even a 1700 player, can play like 1900 or more if he gets into the right type of positions (openings), has had time (freedom from work) to study, etc. Add in possible additional factors (which at least from the lawyer’s words, Bjornsen was not allowed to present to NSF, if any did in fact exist), and the error bars start dropping from «patent» to «likely» to…
Don’t get me wrong, I think where there’s smoke there’s fire, but to base the decision on the «clearest evidence» being just that he played good moves in a statistical sense, well, I have to agree with Aarmo, that it is rather frightening. The other elements (Bluetooth and failure to accept the test), are MUCH more clear to me. Also, *specific* move sequences (as given in the example), rather than any comparison against a (supposed) baseline by Regan has more impact in my book.
Or to put in another way, a «96% match to Rybka» (if true) would be quite substantial evidence, but if I understand Regan’s FIDE method properly, all it is saying is that Bjornsen played at something like a 2300 level rather than a 1700 level on average, which just isn’t such a clinching argument.
According to the verdict (http://www.sjakk.no/filarkiv/2016/04/Pressemelding-Dom-i-Klagesak.pdf) the assumed rating in the analysis was 1750 during the Norwegian Championship, and 2050 during Fagernes. The score (referring to the number of standard deviations away the player was away from the expected agreement with a chess engine) using these assumptions was over 5. The probability of managing to score five standard deviations better than normal is less than 1 in 3 million.
This does NOT mean that the probability of Bjørnsen having scored this well by honest means is less than 1 in 3 million, and equating the probability of a rare feat with the probability of being innocent is known as the «prosecutor’s fallacy». Since there are very many chess players in the world, there is a fair chance that a very unlikely feat will be accomplished by someone. Nonetheless, it is not surprising that the accomplishments have raised suspicion.
I agree with you about the evidence regarding the Bluetooth earphone. The pictures are direct evidence of having used prohibited equipment. In a sense, Bjørnsen is correct in saying that «I understand that it looks strange to have used Bluetooth earphone, but that is a long way from getting information in my ear». Unfortunately, the claim that he was using this incompatible earphone to listen to the recorder is an even longer way off. In any regards, even if he was just listening to music on the earphone it would still be prohibited.
Regan himself said once that a 1 in 1000 chance with outside indications, is better evidence than a 1 in million chance found by data-mining. Of course, he has reasons to be proud of and promote his methods, but I don’t think applying them to the 1750-2050 range is the same as something like Ivanov matching Houdini/Rybka to some high percentage. Since we are talking about results still well within in the realm of humans (only 2300 Elo), and especially with the right to respond, Bjornsen should get a chance to give an explanation of how he had an «extraordinary development» (usedvanlig sjakkutvikling) in strength, though it seems that’s essentially what the NSF did, and his refusal (with lawyerly excuses) makes me doubt he can successfully mount this to persuade the panel. The Regan/FIDE data (in this case specifically) should only be a preliminary indication in my opinion, not persuading on its own. Though if I read the report correctly (thx for link), they did not put this evidential element as quite so central as the blog article did.
1900’s don’t play tactically perfect chess.
And 2300s don’t play match Rybka at 96 rate (or most gms for that matter.)