Chess, Artificial Intelligence and Cloud Computing: The Challenge
Origin of the game
The origin of the game of chess is linked to a figure of an Indian dignitary who is presumed to have lived in India between 500 and 600 AD. The game then arrived in Europe around the year 1000 thanks to the Arabs. This inventor seems to have been called Sassa or Sissa; In some legends he is the son of an Indian king, in others a minister, and in still others a Brahmin.
The funniest legend sees Sassa as King Kaid’s minister, who bored after bringing peace to the kingdom, asks his minister for help. Minister Sassa talks to him about this game and teaches him about it. The king is satisfied and offers him a reward. Sassa, modestly, says that he wants a silver Diram for the first square, two for the second, four for the third, and so on (the board has 64 squares).
That is, for each square it must have a Diram number = 2n-1, which for the 64th square alone means 9.223372037×10¹⁸ Diram. The total of Sassa’s request to his king is an expression of this type:
There are other versions of the same legend, but with a king of Persia and instead of Dirams, grains of rice or wheat.
The only element common to the various legends is the name of the inventor Sassa or Sissa.
Man-Machine Challenge
Chess has always been considered a very intelligent game, which requires great intellectual abilities and which develops the same skills and concentration.
It helps children to develop their intellectual abilities (moving them away from modern tools that depress them) and the elderly to stave off brain aging.
Chess and computers have always gone hand in hand since the beginning of computer science, precisely to justify the complexity of the game and man’s desire to develop algorithms capable of playing chess and able to compete with international grandmasters.
The first algorithms capable of playing chess were born in the late 1940s and early 1950s (Alan Turing was one of the scientists who signed these papers).
The first software saw the light of day in 1967 (Mac Hack VI), then came programs for home computers (apple, spectrum, commodore).
The question begins to spread:
“Will computers ever beat the man on the chessboard one day?”
In 1969 International Master David Levy bet over a thousand pounds with artificial intelligence pioneer John McCarthy who, ten years later, would not be defeated by a computer.
John McCarthy, for those who don’t know, is also considered the father of the Cloud Computing model.
We have to wait until 1996-1997, when IBM’s DeepBlue supercomputer starts to win a few games against the world champion Kasparov.
Kasparov played against DeepBlue 6 times in those two years:
1. e4 c6 {Kasparov plays the Caro-Kan for game six of the match} 2. D4 D5 3. NC3 DXE4 4. Nxe4 Nd7 5. NG5 NGF6 6. Bd3 e6 7. N1f3 h6 8. Nxe6! {This is the move that shocked the world at the time of the match. Kasparov simply collapses after this sacrifice.} 8… Qe7 9. O-O fxe6 (9… Qxe6??? 10. Re1) 10. Bg6+ Kd8 11. Bf4 {Black has a very difficult time developing and finding a plan.} 11… b5 12. A4 BB7 13. Ke1 Nd5 14. Bg3 Kc8 15. axb5 {The computer open lines to the enemy king. White has a decisive advantage.} 15… cxb5 16. Qd3 Bc6 17. Bf5 {All of the tactics are flowing for Deep Blue. All pieces are participating in the hunt for the king.} 17… EXF5 18. Rxe7 Bxe7 19. c4 {Kasparov resigns! What a game and sacrifice by Deep Blue.} 1-0
DeepBlue makes a move (a horse sacrifice – Nxe6!) that shocked the world, Kasparov will quit after 19 moves. Such a match inspired the famous film: “ Game Over, Kasparov and the machine“
Nowadays, after a couple of decades, the competition between computers and chess grandmasters has died down.
If a grandmaster can examine dozens of alternatives for each move by imagining the development of the game forward by 6-8-10 moves, a current supercomputer goes beyond thousands to millions of moves.
Chess computers can count on a huge database of games, openings and endgames and an enormous computing capacity.
Current World Chess Champion Magnus Carlsen boasts ELO rating highest ever 2882. While it is estimated that the ELO rating of current supercomputers reaches 3500-4000.
Computer Vs Computer
Therefore, nowadays, just as there are chess tournaments between men, international chess competitions for super computers only have been born.
Alongside the raw computing power, scientists challenge each other by optimizing algorithms and recently exploiting neural networks:
World Computer Chess Championship ▶
https://icga.org/
Top Chess Engine Championship ▶ https://tcec.chessdom.com
By now, computer chess games are so optimized that a draw is the most common outcome.
The computer has now become a training and teaching tool for all chess players, both beginners and grandmasters.
The chessboard, from being a battlefield between great and sublime human minds (in the past also the scene of the Cold War), becomes a clash of bits between computing power, algorithms and artificial intelligence.
Alpha-Zero , DeepMind’s artificial intelligence algorithm, now owned by Google, learns autonomously using neural networks.
To learn how to play chess, Alpha-Zero played against himself by basic rules. In just a few hours, it can play 44 million games.
That is, like a human player, with over a century of experience and able to remember every move he has ever played, without missing a beat due to seniority.
For some time now, programmers have been using heuristic evaluations, a lateral thinking that guarantees better results than pure mathematical strength. In essence, the search is guided by allowing the computer to recognize the most promising strands, reducing the complexity of the algorithm both in a temporal sense (the duration of the search) and in a spatial sense (the memory occupied by the process).
That is, the computer “thinks outside the box.”
Last April, the open-source machine learning engine LC0 won the
Computer Chess Championship
and made history.
This August, a chess engine (Chiron) developed by Italian Ubaldo Andrea Farina, came second at the 2019 World Computer Chess Championship held in Macau, China.
It is the first time that an Italian program has finished in the top three in the world championship.
The undefeated champion Komodo could count on a 128 Intel Xeon cores, while the Italian engine on 48 Intel Xeon cores.
Both algorithms came out of the competition undefeated; Chiron won two matches and drew eight, winner Komodo won 5 and drew 5, winning the competition.
Chiron uses a cloud service
This is Amazon AWS EC2, specifically the c5.24xlarge with 96 vCPUs, using only the real physical cores, so 48.
The CPUs were two Intel Xeon Platinum 8175M.
On this hardware, Chiron had an average speed of 60-70 million positions per second.
According to Farina, using a cloud server was necessary because he couldn’t find a sponsor.
In the end, the total cost of the competition was $239.
Sources:
- https://www.ultimouomo.com/storia-della-sfida-tra-uomo-e-computer-a-scacchi/
- https://en.chessbase.com/post/kasparov-on-the-future-of-artificial-intelligence
- https://science.sciencemag.org/content/362/6419/1087
- https://www.dday.it/redazione/31961/il-motore-italiano-chiron-e-arrivato-secondo-ai-campionati-del-mondo-dei-programmi-di-scacchi-abbiamo-chiesto-al-suo-sviluppatore-come-ci-e-riuscito
- https://www.chess.com/news/view/lc0-wins-computer-chess-championship-makes-history
- https://www.chess.com/article/view/deep-blue-kasparov-chess
Do you want to play chess with us?
Write to us and book your chess challenge against the CEO of VMEngine.
If you win, you can win up to $500 in trial credits for AWS services.
