A Game That Outlasted the Day (4)

by ChessBase
4/14/2026 – It should be clearly emphasized that the longest chess game of all time can only arise by chance. Any prior agreement between the opponents before the game—such as, "Let’s play the longest game in history today!" – or any such understanding reached during the game, automatically turns them into cheaters, with all the ensuing consequences. Estonian chess expert and trainer, Valery Golubenko, tells us about the struggle to regulate very long theoretical endings.

Your personal chess trainer. Your toughest opponent. Your strongest ally.
FRITZ 20 is more than just a chess engine – it is a training revolution for ambitious players and professionals. Whether you are taking your first steps into the world of serious chess training, or already playing at tournament level, FRITZ 20 will help you train more efficiently, intelligently and individually than ever before. 

What to do about very long endings

By Valery Golubenko

In his book, in the chapter “Queen and Pawn vs. Queen”, Mark Dvoretsky focuses mainly on endings with a knight’s pawn and on the following game:

In that position, Botvinnik immediately made a mistake by giving check on f4, after which the position became a draw. Yet he still won the game — thanks to Paul Keres!

The point is that eight years earlier Botvinnik had played a similar ending against Grigory Ravinsky, which Keres analyzed in great detail and came to the erroneous conclusion that the safest squares for the defending king were a4 and a5. As a result, the Bulgarian grandmaster went to a5 and lost! My own analysis today shows that the truly safe squares are h8, h7, and h6 — that is, in Minev’s case, a3, a2, and a1!

In the initial position, according to the Nalimov Tablebases, White mates in 60 moves. What does this mean in practice? Checkmate with two queens against one is achieved in about 12–13 moves — which means that the stronger side must make more than 45 only moves before getting the second queen! For a human, this is impossible. After Minev’s mistake, the mate distance “collapsed” immediately to 35 moves! And it was precisely this sharp drop that allowed the world champion to convert the game to a win.

FIDE officially modified the 50-move rule to allow 100 moves in specific, hard-to-win endgames in 1984:

  • Rook and bishop vs. rook
  • Two knights vs. a pawn (specifically when it was safe to do so)
  • Queen and pawn on the 7th vs. queen.

Then, in 1989, immediately after the game Ivan Nikolić–Goran Arsović, FIDE reduced the allowed limit for these positions from 100 moves to 75 moves. In my opinion, the drawn endgame R+B vs. R should never have been included in this list at all — the drawing technique there is quite simple. The Nikolić–Arsović game was played under this setting, was adjourned many times, and ended in a draw on move 269.

Ironically, when the development of chess engines revealed even more positions in which a win requires more than 50 moves, FIDE (in 2001) reinstated the 50-move rule for all positions without exception! What is the logic here?!

Finally, in 2014, FIDE introduced the rule of an automatic draw after 75 moves, and this rule was finally applied in the game Gorkov–Golubenko.

Only now has it suddenly dawned on me: the game Ivan Nikolić-Goran Arsović should have ended after Black’s 161st move! The point is that after 111…d5 the following position arose:

and the pawn on d5 was captured only on move 167! But in 1989 the endgame R+B vs. R+P was not among the exceptions (this is not R+B vs. R at all!) and the 50-move rule applied to it! Goran Arsović clearly intended to claim a draw (as any normal chess player down a piece would!), but why did he do so under the 100-move rule (168+101=269) after many days of play, and not under the 50-move rule 108 moves earlier?? Perhaps the tournament table from the Serbian chess magazine Mat, issue no. 1–2/1989, will provide an answer to this question?

Translation from Serbian of the last paragraph under the tournament table:

“It should be noted that at this tournament Goran Arsović and Ivan Nikolić played the longest game in the world. The game will most likely enter the Guinness Book of Records. The game lasted 269 moves and was played over 20 hours and 15 minutes.”

We are talking about Goran, not his brother Zoran Arsović, who won the tournament. However, it seems that this is already a different story…

Previous articles on the subject

About the author

Valery Golubenko is a FIDE Trainer, Estonian Rapid Chess Champion (1993–1994), and a multiple-time Estonian team champion in classical chess, rapid and blitz. He holds a PhD in Applied Mathematics and is the author of his own definition of the imaginary unit in higher mathematics.

Valery is the father of two daughters, Valentia and Alexandra, (who lost their mother in 2012). Valentina won the Girls Under-18 World Championship in 2008 and became the first Woman Grandmaster from Estonia. Her younger sister Alexandra competed in the European Youth Championships in 2015 and 2019 in Estonia, in rapid, blitz, and chess problem solving, coming close to winning the problem-solving championship. By education, Alexandra is a game designer and artist.

This photo was taken at the Zagreb Zoo in 2015. Alexandra (left) was returning from the European Youth Championship in Poreč (where Valery served as head of the Swedish delegation), while Valentina had flown in for the Croatian Team Championship.


Discussion and Feedback Submit your feedback to the editors