Post by hi224 on Apr 22, 2021 1:20:22 GMT
Hello, some of you might recognize my name I have been a writer here for some time, this particular write-up has been a strong passion project for me over the past few months. Aspects of this write-up include quantitative analysis which I have best tried to explain, this story lends itself to a visual format. So if you wish you can see my video on it here. For the purpose of this write-up paragraphs in Italics are 'exposition' from me. Feedback is very much appreciated.
[Long]
Introduction and Background:
In 1967 the meteoric rise of Bobby Fischer began, starting one of the most dominant runs in chess ever seen. Setting a record 20 game long winning streak against elite competition defeating grandmasters Mark Taimanov, Bent Larsen, Tigran Petrosian, Oscar Panno, Svetozar Gligoric, Wolfgang Uhlmann, Duncan Suttles and Jorge Alberto Rubinetti. All this culminating in a match between Fischer and Boris Spassky for the world chess Championship. With Bobby Fischer representing the USA and Boris Spassky the USSR the match would become highly publicized as the world was watching. Fischer played with an incredible and unseen level of skill often taking Spassky by surprise through intentionally playing a number of different unique openings. Eventually resulting in a dominant performance by Fischer; with a final score of 12 and ½ to 8 and ½ in favor of Fischer. leading to Fischer being in contention for the title of 'Greatest of All Time'.
Fischer became a household name, and spurred on a whole new generation of Chess players within the US and across the world. Featuring on talk shows such as, The Dick Cavette show and The Carson Tonight Show as well as appearing in major publications such as the cover of Sports Illustrated. However just as quick as he came, Fischer with the same speed disappeared from the public sphere; turning down future chess matches and hiding from his own popularity.
On the 17th of January 2008 Bobby Fischer would pass away in Iceland at the age of 64. Leaving the world wondering if he ever did reach his true potential, due to playing very few games after 1972.
After his death and in the years since then, a small rumor has persisted. This rumor centering around a particular website – The Internet Chess club.
In fact this very Sub had covered the mystery in the past, and saw a certain level of attention. Click HERE. If you wish to see a video covering the games, and what introduced me to the mystery click here.
The Games:
This section is best presented visually, and I have done my best in explaining what is happening if you don't play chess. I have included the chess algebraic notation for those who can visualize/read them. Its not 100% necessary to know the exact moves of the games, but it helps.
Nigel Short a Grand Master who was amongst the best of his time, and once challenging for the world championship would log into the Internet Chess club on the 4th of February 2001. He would encounter a Guest Player through an introduction from a third party in order play a series of Blitz Games. This player had the username “Guest 2563”, and would play Short (username Ural) for a total of 8 blitz Games under three minute time constraints. For those less aware of chess this constraint means both players have a total of 3 minutes to make all their moves, one wins by either Checkmating their opponent or the opponent running out of time.
Guest2563 would play with impeccable skill. Outplaying Short in the mid to late games displaying a high level of ability and resourcefulness, furthermore the Guest would play incredibly unique opening moves in each game. The dominance and creativity of this guest would become most visible in their sixth game, where they would play the oddest opening which involved walking the king and seemingly purposefully giving his opponent a stronger position. The Guest player would go on to win the game in 51 moves. Short would play this anonymous person 8 times, all of which were won by the Guest player in similar fashion.
Game 6 Algebraic Notation (Guest user is white, Short is Black):
1.f4 d5 2.Kf2 Nc6 3.Kf3 e5 4.c3 Nf6 5.d3 Be7 6.g3 0-0 7.Kg2 e4 8.d4 h5 9.e3 g6 10.h3 Kg7 11.c4 Rg8 12.c5 Rh8 13.Nc3 Be6 14.a3 Qd7 15.Bb5 Rag8 16.b4 Kf8 17.Ba4 g5 18.fxg5 Rxg5 19.b5 Nd8 20.Nge2 Rhg8 21.b6 c6 22.bxa7 Qc8 23.Bb3 Qa8 24.Rb1 Qxa7 25.Rf1 Qa8 26.Qe1 Qc8 27.Nf4 Rxg3+ 28.Qxg3 Rxg3+ 29.Kxg3 Bf5 30.Bd1 Ne6 31.Nxh5 Nxh5+ 32.Bxh5 Ng7 33.Bg4 Bxg4 34.hxg4 Qc7+ 35.Kg2 Ne6 36.Bd2 Ng5 37.Be1 Nf3 38.Rh1 Qd7 39.Kg3 Bd8 40.Ne2 Bc7+ 41.Nf4 Qe7 42.Bc3 Bxf4+ 43.exf4 Kg7 44.a4 Qc7 45.Rb6 f6 46.Rhb1 Kg6 47.Rxb7 Qd8 48.f5+ Kg5 49.Rg7+ Kh6 50.Rg6+ Kh7 51.Rh1+ Black Resigns.
The key takeaway here is that white (Guest User) played a very odd opening, and still won despite the opening putting white at a disadvantage.
Nigel Short was blown away by this players performance and so he pressed for confirmation as to who he was playing. In order to not ask directly if the Guest Player was Fischer, Short asked a question which only Fischer would know the answer to ‘Do you know Armando Acevedo ?’, the Player answered apparently immediately with ‘Siegen 1970’. Fischer had indeed played Acevedo in The Seigen Chess Olympiad in 1970 . Later Short would also ask the user who was the strongest blitz player they had ever competed against; to which the user replied ‘If I am who you think I am, I would answer Mikhail Tal’. Tal a former Champion had resoundingly defeated a 16 year old Fischer in Blitz Chess. Furthermore with the unique playing style that displayed a lack of respect for or knowledge of opening theory, which Fischer was very critical of (this is not to be understated) and the players incredibly strong mid to late game abilities as well as the Guest being incredibly resourceful led to Nigel Short saying he was 99% sure he had played Bobby Fischer. It was the return of the King.
This would be a huge deal for the Chess world, for if it were true it would mean that Fischer was still competitively viable in chess and an ex-champion who had never lost the title was now once again active. Short would further say that this user was a stronger Blitz player than the Champion Garry Kasparov, which would be a very large surprise as Fischer was old at the time being the age of 58. Like other sports, chess players abilities decrease as they get older and thus if Short’s claims were true Fischer would have had incredible longevity not seen before in chess.
Of course, this story was simply too great to ignore. Chess media have largely viewed these games as being played by Fischer leading to many articles being written and videos created – simply people just wanted to believe it to be true. Despite many dissenting voices the popular line being ended up believing it was Fischer, the myth has persisted for 20 years with media still creating content from it.
A Quantitative Gambit:
So to begin, others had taken a look at these games in 2001 and given some evidence of the user not being Fischer, however their results weren't given a lot of attention and due to technological advancements we can now analyze the games deeper. Link here.
Many aspects of the games were quickly drawn under question, in particular the main piece of ‘Evidence’ in favour of the Guest player being Fischer was immediately put under inspection. This piece of evidence being the question which Nigel asked enquiring about the Mexican Chess player Armando Acevedo, which apparently only Fischer could. However, there are a number of chess databases which give the user the ability to search through hundreds of thousands of games with a particular focus on incredible players such as Fischer. Two particular databases which were readily available at the time were Chess Base and Fritz. Someone could easily search the name Armando Avecedo in either database which would give the same result the guest player used being ‘Avecedo vs Fischer, Siegen Olympiad 1970’. Upon testing, this could be done within a very quick timeframe. Undermining the primary evidence which is often presented by those who believe it is Fischer. Furthermore the second question, being ‘Who was the strongest blitz player you have ever played ?’, could be answered in a similar fashion or by having a general knowledge of Fischer and the notable figures which he played.
Following this, it is important to note that Chess Engines had become increasingly strong – with Deep Blue defeating Garry Kasparov in 1997 and much progress being made by 2001 in this area. Multiple commercially available engines such as Rebel, Fritz, Chess Tiger and Gambit Tiger had been created. The strength of these engines in the quicker time constraints being Blitz were becoming increasingly dominant, with Rebel Engine in particular defeating Vishy Anand 5 to 3. Thus we can see a theory forming; what if these games were played not by a human but by a Chess Engine ?
See given my background, upon finding out about these games it immediately occurred to me that this would be a very easy hypothesis to prove/disprove (I was wrong about the easy part) but it is a mystery which can be directly disproven if certain tests are conducted and results produced.
This concept was previously tested against a second set of games played by the Guest user against the International Master Robert Fontaine otherwise known as Beber on the ICC. The testing which was conducted revealed that after a number of turns, the anonymous users moves would align perfectly with the engine Deep Fritz. When switching over to using a chess engine, the player typically would take an extended amount of time to make their following move – thus we can see that before the move Bishop takes C4, the anonymous player would take on average 3 seconds to move. However for this particular move the anonymous player took 12 seconds, possibly setting up an engine. Following this – every single move made by the anonymous player would have been made by the Chess Engine Deep Fritz in multiple variation mode; if the user selected the second variation. Another Chess engine Gambit tiger too would have given similar but not exact results.
Algebraic Notation White - Guest User and Black - IM Fontaine:
1.e4 c5 2.Ke2 Nc6 3.Ke3 g6 4.Nc3 Nd4 5.d3 Nf6 6.Kd2 d5 7.Ke1 Bg7 8.h3 0-0 9.a3 e5 10.Bg5 Be6 11.exd5 Bxd5 12.Nxd5 Qxd5 13.c3 Ne6 14.Be3 Rad8 15.Qa4 e4 16.dxe4 Nxe4 17.Rc1 a6 18.Be2 b5 19.Qxa6 c4 20.Rd1 Qf5 21.Nf3 Nxc3 22.bxc3 Bxc3+ 23.Nd2 Nc5 24.Bxc5 Bxd2+ 25.Rxd2 Rxd2 26.Kxd2 Rd8+ 27.Kc1 Qe5 28.Bxc4 Qc3+ 29.Kb1 bxc4 30.Rc1 Qb3+ 31.Ka1 Rd2 32.Qc8+ Kg7 33.Bf8+ Kf6 34.Qc6+ Kf5 35.Qc5+ Kf6 36.Qe7+ Kf5 37.Qxf7+ Kg5 38.Be7+ Kh6 39.Qf8+ Kh5 40.g4#
White (Guest) wins by checkmate.
However this testing was conducted on a set of games not played by Short, and it is imperative to test them. To begin, I had acquired a number of different chess Engines in the hopes some of them would produce similar results; however I found that the more recent engines being Stockfish 12 and Komodo 13 would produce on average vastly different moves than the ones played by the Guest User. This was to be expected, as both of these engines are incredibly strong and calculate to a deeper degree than the engines readily available in 2001. Thus it required me to find a number of 20 year old chess engines – and test them, a task not so easily achieved – as these engines by todays standards are highly inaccessible and very difficult to locate especially as many of them have had subsequent updates and re-releases.
The results of the new engines were around 50% similarity or <50%, really not great results for me, but I knew that the newer engines would cause issues. During my search for the old engines, I did stumble across a website which was dated but dedicated to the Chess Engine REBEL and its founders. I sent them an email, and to my surprise I got an enthusiastic response, not only did they have the engines they were willing to test them too. Please see acknowledgements at the bottom for further info.
Initial testing (process testset) conducted by those at Rebel revealed that the moves played by the Guest User in the Short games had a 58% congruency with the Fritz engine, meaning given the boards position the fritz engine would make the same following move 58% of the time. This was a relatively high result and an early indicator that there was something more behind these games.
Following early testing those at Rebel requested help from industry professionals on a public forum. The testing which was conducted on this forum was a centipawn loss analysis by the user “MWYOUNG”, this analysis is a way to quantify the difference between the move made by the chess player and the best move possible as noted by a chess engine. Thus the higher the centipawn result, the worse the players moves were when compared to the best possible engine moves. Thus in the case of recognizing the use of engines, the centipawn loss is to be close to zero on a scale of 0 to 1. Unfortunately for us this analysis produced the following results:
The guest user scoring an average of 0.51, while Nigel Short averaged a centipawn loss of 0.83. Using Former world Champion Garry Kasparov as an example, using a selection of 4 blitz games his average centipawn loss was 0.26. Thus indicating that the guest user did not use an engine, and in fact also indicating that Nigel Short played below his calibre.
These results were a little disheartening. However, these results got me thinking; and others too. The Centipawn loss analysis could not be applicable in this case – as it calculates games a whole including the opening moves. Which if you remember, were very odd and often gave the Guest User’s opponent a positional advantage. Thus if we were to presume the first 5-7 moves were made by a human, we can presume that this would highly negatively effect the centipawn loss analysis and thus it is not applicable. How about the congruency (process testset) analysis ?
Upon the reinspection of the congruency analysis it occurred to myself and those at Rebel, what if we removed the first 5-7 moves.
After removing the first 5 moves from the equation the congruency analysis would produce incredible results:
The Fritz engine would have a 68.6% similarity with the moves played by the guest user, furthermore Chess Tiger had a 71.4% similarity while Gambit Tiger had the highest level of similarities at 72.3%. This is huge indicator that something or some engine was behind many of the moves played the by the Guest User. Furthermore, certain games might not fall within the 5-7 initial human played moves and thus we can presume the true similarity would be much higher however continued testing would most likely result in diminishing returns and a perfect match would likely be impossible. Yet we will find that coupled with a few pieces of historical information largely ignored by those discussing these games; it was not fischer.
[Results]
Fritz 5.32: 68.6%
Chess Tiger: 71.4%
Gambit Tiger: 72.3%
Conclusion: A myth can travel half way around the world while the truth is still sitting in its shoes.
I’d like to begin by going direct to source, to Nigel Short and what he has said since the initial claim. Nigel in fact retracted these claims, citing new information and in his own words (in a tweet): he’d addressed this a long time ago; it’s just no one is interested in retractions. Which is true, in fact at multiple times Nigel has addressed this including and most recently in a stream with Grand Master Ben Finegold.
It is also to be noted other top chess players too questioned the claim (Show fischer caught out), and also were ignored.
Now lets see what Fischer had to say about this, Bobby Fischer in an interview too also denied ever playing online referring to it as:
‘A lot of BS’
Ok so this interview is linked at the bottom please read there.
Which makes a lot of sense for at the time, Fischer seemingly had strayed away from Chess and was mostly playing and working on his own variation of the game called FischerRandom – a type of chess designed to reduce the impact of opening theory.
And finally – I’d like to note as of writing (now two weeks ago) this Fredrik Friedel the editor of the website Chessbase mentioned he spent a number of weeks tracking down the player. Finding it was just a systems operator, playing out of Canada using a computer. Which is an impressive find, but also indicates that the myth lives on.
Really I hope you enjoy this and if you wish you can check out my video. There have been so many drafts of this write up it's insane, I hope I explained things well. A huge thank you to Ed from REBEL who did so much of the heavy lifting.
Acknowledgements:
Ed Schroder from REBEL rebel13.nl/
Everyone on the Prodeo forms If you wish to see the analysis, its best you go here.
[Links]
I cant link the tweet from Nigel, sorry reddit wont let me.
Chessbase
Chasebase look at comments Frederick Friedel Statement
The Guardian
Agadmator
Telegraph
Interview Radio interview, pre warning guys Fischer probably suffered from Paranoid Schizophrenia and said a lot of anti-Semitic things despite himself being Jewish. It's very difficult to listen to. Quote is at 1:30
Ben Finegold Stream Quote is at 23:30
If a major publication once again 'takes inspiration' from my work - at least reference me and those at REBEL and Prodeo.
[Long]
Introduction and Background:
In 1967 the meteoric rise of Bobby Fischer began, starting one of the most dominant runs in chess ever seen. Setting a record 20 game long winning streak against elite competition defeating grandmasters Mark Taimanov, Bent Larsen, Tigran Petrosian, Oscar Panno, Svetozar Gligoric, Wolfgang Uhlmann, Duncan Suttles and Jorge Alberto Rubinetti. All this culminating in a match between Fischer and Boris Spassky for the world chess Championship. With Bobby Fischer representing the USA and Boris Spassky the USSR the match would become highly publicized as the world was watching. Fischer played with an incredible and unseen level of skill often taking Spassky by surprise through intentionally playing a number of different unique openings. Eventually resulting in a dominant performance by Fischer; with a final score of 12 and ½ to 8 and ½ in favor of Fischer. leading to Fischer being in contention for the title of 'Greatest of All Time'.
Fischer became a household name, and spurred on a whole new generation of Chess players within the US and across the world. Featuring on talk shows such as, The Dick Cavette show and The Carson Tonight Show as well as appearing in major publications such as the cover of Sports Illustrated. However just as quick as he came, Fischer with the same speed disappeared from the public sphere; turning down future chess matches and hiding from his own popularity.
On the 17th of January 2008 Bobby Fischer would pass away in Iceland at the age of 64. Leaving the world wondering if he ever did reach his true potential, due to playing very few games after 1972.
After his death and in the years since then, a small rumor has persisted. This rumor centering around a particular website – The Internet Chess club.
In fact this very Sub had covered the mystery in the past, and saw a certain level of attention. Click HERE. If you wish to see a video covering the games, and what introduced me to the mystery click here.
The Games:
This section is best presented visually, and I have done my best in explaining what is happening if you don't play chess. I have included the chess algebraic notation for those who can visualize/read them. Its not 100% necessary to know the exact moves of the games, but it helps.
Nigel Short a Grand Master who was amongst the best of his time, and once challenging for the world championship would log into the Internet Chess club on the 4th of February 2001. He would encounter a Guest Player through an introduction from a third party in order play a series of Blitz Games. This player had the username “Guest 2563”, and would play Short (username Ural) for a total of 8 blitz Games under three minute time constraints. For those less aware of chess this constraint means both players have a total of 3 minutes to make all their moves, one wins by either Checkmating their opponent or the opponent running out of time.
Guest2563 would play with impeccable skill. Outplaying Short in the mid to late games displaying a high level of ability and resourcefulness, furthermore the Guest would play incredibly unique opening moves in each game. The dominance and creativity of this guest would become most visible in their sixth game, where they would play the oddest opening which involved walking the king and seemingly purposefully giving his opponent a stronger position. The Guest player would go on to win the game in 51 moves. Short would play this anonymous person 8 times, all of which were won by the Guest player in similar fashion.
Game 6 Algebraic Notation (Guest user is white, Short is Black):
1.f4 d5 2.Kf2 Nc6 3.Kf3 e5 4.c3 Nf6 5.d3 Be7 6.g3 0-0 7.Kg2 e4 8.d4 h5 9.e3 g6 10.h3 Kg7 11.c4 Rg8 12.c5 Rh8 13.Nc3 Be6 14.a3 Qd7 15.Bb5 Rag8 16.b4 Kf8 17.Ba4 g5 18.fxg5 Rxg5 19.b5 Nd8 20.Nge2 Rhg8 21.b6 c6 22.bxa7 Qc8 23.Bb3 Qa8 24.Rb1 Qxa7 25.Rf1 Qa8 26.Qe1 Qc8 27.Nf4 Rxg3+ 28.Qxg3 Rxg3+ 29.Kxg3 Bf5 30.Bd1 Ne6 31.Nxh5 Nxh5+ 32.Bxh5 Ng7 33.Bg4 Bxg4 34.hxg4 Qc7+ 35.Kg2 Ne6 36.Bd2 Ng5 37.Be1 Nf3 38.Rh1 Qd7 39.Kg3 Bd8 40.Ne2 Bc7+ 41.Nf4 Qe7 42.Bc3 Bxf4+ 43.exf4 Kg7 44.a4 Qc7 45.Rb6 f6 46.Rhb1 Kg6 47.Rxb7 Qd8 48.f5+ Kg5 49.Rg7+ Kh6 50.Rg6+ Kh7 51.Rh1+ Black Resigns.
The key takeaway here is that white (Guest User) played a very odd opening, and still won despite the opening putting white at a disadvantage.
Nigel Short was blown away by this players performance and so he pressed for confirmation as to who he was playing. In order to not ask directly if the Guest Player was Fischer, Short asked a question which only Fischer would know the answer to ‘Do you know Armando Acevedo ?’, the Player answered apparently immediately with ‘Siegen 1970’. Fischer had indeed played Acevedo in The Seigen Chess Olympiad in 1970 . Later Short would also ask the user who was the strongest blitz player they had ever competed against; to which the user replied ‘If I am who you think I am, I would answer Mikhail Tal’. Tal a former Champion had resoundingly defeated a 16 year old Fischer in Blitz Chess. Furthermore with the unique playing style that displayed a lack of respect for or knowledge of opening theory, which Fischer was very critical of (this is not to be understated) and the players incredibly strong mid to late game abilities as well as the Guest being incredibly resourceful led to Nigel Short saying he was 99% sure he had played Bobby Fischer. It was the return of the King.
This would be a huge deal for the Chess world, for if it were true it would mean that Fischer was still competitively viable in chess and an ex-champion who had never lost the title was now once again active. Short would further say that this user was a stronger Blitz player than the Champion Garry Kasparov, which would be a very large surprise as Fischer was old at the time being the age of 58. Like other sports, chess players abilities decrease as they get older and thus if Short’s claims were true Fischer would have had incredible longevity not seen before in chess.
Of course, this story was simply too great to ignore. Chess media have largely viewed these games as being played by Fischer leading to many articles being written and videos created – simply people just wanted to believe it to be true. Despite many dissenting voices the popular line being ended up believing it was Fischer, the myth has persisted for 20 years with media still creating content from it.
A Quantitative Gambit:
So to begin, others had taken a look at these games in 2001 and given some evidence of the user not being Fischer, however their results weren't given a lot of attention and due to technological advancements we can now analyze the games deeper. Link here.
Many aspects of the games were quickly drawn under question, in particular the main piece of ‘Evidence’ in favour of the Guest player being Fischer was immediately put under inspection. This piece of evidence being the question which Nigel asked enquiring about the Mexican Chess player Armando Acevedo, which apparently only Fischer could. However, there are a number of chess databases which give the user the ability to search through hundreds of thousands of games with a particular focus on incredible players such as Fischer. Two particular databases which were readily available at the time were Chess Base and Fritz. Someone could easily search the name Armando Avecedo in either database which would give the same result the guest player used being ‘Avecedo vs Fischer, Siegen Olympiad 1970’. Upon testing, this could be done within a very quick timeframe. Undermining the primary evidence which is often presented by those who believe it is Fischer. Furthermore the second question, being ‘Who was the strongest blitz player you have ever played ?’, could be answered in a similar fashion or by having a general knowledge of Fischer and the notable figures which he played.
Following this, it is important to note that Chess Engines had become increasingly strong – with Deep Blue defeating Garry Kasparov in 1997 and much progress being made by 2001 in this area. Multiple commercially available engines such as Rebel, Fritz, Chess Tiger and Gambit Tiger had been created. The strength of these engines in the quicker time constraints being Blitz were becoming increasingly dominant, with Rebel Engine in particular defeating Vishy Anand 5 to 3. Thus we can see a theory forming; what if these games were played not by a human but by a Chess Engine ?
See given my background, upon finding out about these games it immediately occurred to me that this would be a very easy hypothesis to prove/disprove (I was wrong about the easy part) but it is a mystery which can be directly disproven if certain tests are conducted and results produced.
This concept was previously tested against a second set of games played by the Guest user against the International Master Robert Fontaine otherwise known as Beber on the ICC. The testing which was conducted revealed that after a number of turns, the anonymous users moves would align perfectly with the engine Deep Fritz. When switching over to using a chess engine, the player typically would take an extended amount of time to make their following move – thus we can see that before the move Bishop takes C4, the anonymous player would take on average 3 seconds to move. However for this particular move the anonymous player took 12 seconds, possibly setting up an engine. Following this – every single move made by the anonymous player would have been made by the Chess Engine Deep Fritz in multiple variation mode; if the user selected the second variation. Another Chess engine Gambit tiger too would have given similar but not exact results.
Algebraic Notation White - Guest User and Black - IM Fontaine:
1.e4 c5 2.Ke2 Nc6 3.Ke3 g6 4.Nc3 Nd4 5.d3 Nf6 6.Kd2 d5 7.Ke1 Bg7 8.h3 0-0 9.a3 e5 10.Bg5 Be6 11.exd5 Bxd5 12.Nxd5 Qxd5 13.c3 Ne6 14.Be3 Rad8 15.Qa4 e4 16.dxe4 Nxe4 17.Rc1 a6 18.Be2 b5 19.Qxa6 c4 20.Rd1 Qf5 21.Nf3 Nxc3 22.bxc3 Bxc3+ 23.Nd2 Nc5 24.Bxc5 Bxd2+ 25.Rxd2 Rxd2 26.Kxd2 Rd8+ 27.Kc1 Qe5 28.Bxc4 Qc3+ 29.Kb1 bxc4 30.Rc1 Qb3+ 31.Ka1 Rd2 32.Qc8+ Kg7 33.Bf8+ Kf6 34.Qc6+ Kf5 35.Qc5+ Kf6 36.Qe7+ Kf5 37.Qxf7+ Kg5 38.Be7+ Kh6 39.Qf8+ Kh5 40.g4#
White (Guest) wins by checkmate.
However this testing was conducted on a set of games not played by Short, and it is imperative to test them. To begin, I had acquired a number of different chess Engines in the hopes some of them would produce similar results; however I found that the more recent engines being Stockfish 12 and Komodo 13 would produce on average vastly different moves than the ones played by the Guest User. This was to be expected, as both of these engines are incredibly strong and calculate to a deeper degree than the engines readily available in 2001. Thus it required me to find a number of 20 year old chess engines – and test them, a task not so easily achieved – as these engines by todays standards are highly inaccessible and very difficult to locate especially as many of them have had subsequent updates and re-releases.
The results of the new engines were around 50% similarity or <50%, really not great results for me, but I knew that the newer engines would cause issues. During my search for the old engines, I did stumble across a website which was dated but dedicated to the Chess Engine REBEL and its founders. I sent them an email, and to my surprise I got an enthusiastic response, not only did they have the engines they were willing to test them too. Please see acknowledgements at the bottom for further info.
Initial testing (process testset) conducted by those at Rebel revealed that the moves played by the Guest User in the Short games had a 58% congruency with the Fritz engine, meaning given the boards position the fritz engine would make the same following move 58% of the time. This was a relatively high result and an early indicator that there was something more behind these games.
Following early testing those at Rebel requested help from industry professionals on a public forum. The testing which was conducted on this forum was a centipawn loss analysis by the user “MWYOUNG”, this analysis is a way to quantify the difference between the move made by the chess player and the best move possible as noted by a chess engine. Thus the higher the centipawn result, the worse the players moves were when compared to the best possible engine moves. Thus in the case of recognizing the use of engines, the centipawn loss is to be close to zero on a scale of 0 to 1. Unfortunately for us this analysis produced the following results:
The guest user scoring an average of 0.51, while Nigel Short averaged a centipawn loss of 0.83. Using Former world Champion Garry Kasparov as an example, using a selection of 4 blitz games his average centipawn loss was 0.26. Thus indicating that the guest user did not use an engine, and in fact also indicating that Nigel Short played below his calibre.
These results were a little disheartening. However, these results got me thinking; and others too. The Centipawn loss analysis could not be applicable in this case – as it calculates games a whole including the opening moves. Which if you remember, were very odd and often gave the Guest User’s opponent a positional advantage. Thus if we were to presume the first 5-7 moves were made by a human, we can presume that this would highly negatively effect the centipawn loss analysis and thus it is not applicable. How about the congruency (process testset) analysis ?
Upon the reinspection of the congruency analysis it occurred to myself and those at Rebel, what if we removed the first 5-7 moves.
After removing the first 5 moves from the equation the congruency analysis would produce incredible results:
The Fritz engine would have a 68.6% similarity with the moves played by the guest user, furthermore Chess Tiger had a 71.4% similarity while Gambit Tiger had the highest level of similarities at 72.3%. This is huge indicator that something or some engine was behind many of the moves played the by the Guest User. Furthermore, certain games might not fall within the 5-7 initial human played moves and thus we can presume the true similarity would be much higher however continued testing would most likely result in diminishing returns and a perfect match would likely be impossible. Yet we will find that coupled with a few pieces of historical information largely ignored by those discussing these games; it was not fischer.
[Results]
Fritz 5.32: 68.6%
Chess Tiger: 71.4%
Gambit Tiger: 72.3%
Conclusion: A myth can travel half way around the world while the truth is still sitting in its shoes.
I’d like to begin by going direct to source, to Nigel Short and what he has said since the initial claim. Nigel in fact retracted these claims, citing new information and in his own words (in a tweet): he’d addressed this a long time ago; it’s just no one is interested in retractions. Which is true, in fact at multiple times Nigel has addressed this including and most recently in a stream with Grand Master Ben Finegold.
It is also to be noted other top chess players too questioned the claim (Show fischer caught out), and also were ignored.
Now lets see what Fischer had to say about this, Bobby Fischer in an interview too also denied ever playing online referring to it as:
‘A lot of BS’
Ok so this interview is linked at the bottom please read there.
Which makes a lot of sense for at the time, Fischer seemingly had strayed away from Chess and was mostly playing and working on his own variation of the game called FischerRandom – a type of chess designed to reduce the impact of opening theory.
And finally – I’d like to note as of writing (now two weeks ago) this Fredrik Friedel the editor of the website Chessbase mentioned he spent a number of weeks tracking down the player. Finding it was just a systems operator, playing out of Canada using a computer. Which is an impressive find, but also indicates that the myth lives on.
Really I hope you enjoy this and if you wish you can check out my video. There have been so many drafts of this write up it's insane, I hope I explained things well. A huge thank you to Ed from REBEL who did so much of the heavy lifting.
Acknowledgements:
Ed Schroder from REBEL rebel13.nl/
Everyone on the Prodeo forms If you wish to see the analysis, its best you go here.
[Links]
I cant link the tweet from Nigel, sorry reddit wont let me.
Chessbase
Chasebase look at comments Frederick Friedel Statement
The Guardian
Agadmator
Telegraph
Interview Radio interview, pre warning guys Fischer probably suffered from Paranoid Schizophrenia and said a lot of anti-Semitic things despite himself being Jewish. It's very difficult to listen to. Quote is at 1:30
Ben Finegold Stream Quote is at 23:30
If a major publication once again 'takes inspiration' from my work - at least reference me and those at REBEL and Prodeo.