E5, Karpov pushed his pawn, and that was when Kasparov knew it was all over, the reigning king had blundered his rook. After an intense war of 24 battles, on 9th November 1985 Kasparov was crowned as the undisputed champion at the mere age of 22, a feat no other human had achieved. This was total dominance. Tinman, Ivanchuk, Short, Andersson, absolutely no one, could stand his might. Although the former World Champion, Karpov, challenged the new king on multiple occasions, all of these were close battles, which gave rise to one of the greatest rivalries the world has ever seen. Eventually, Garry Kasparov stood on the top, securing his position for the coming years. He was arguably the greatest chess player the world had ever seen. His aggression and offensive play, struck fear in his opponents. They said, “Kasparov doesn’t defeat his opponents, he destroys them”. It seemed as if he had superhuman abilities to make complex computations in his mind.
“There can be no chess program that can defeat me”, he boasted and in 1989, he defeated DeepThought, a chess computer. From 1986 to 1996, he had completed 10 years of total dominance, when he came across Deep Blue, a computer made by IBM, against which he escaped with a narrow victory margin; the relief was short-lived as a rematch was set-up. The world was watching Kasparov, as he was carrying the burden of the entire humanity on his shoulders.Everyone was rooting for him. The match,“Kasparov vs Deep Blue”, gained great media hype. Kasparov faced a 3.5 – 2.5 defeat against Deep Blue. He was so shocked by this result, he claimed that the machine was cheating. His proof being that in one instance, he had set up a deadly trap for the computer, a trick Kasparov had used on previous opponents. He was surprised when Deep Blue blundered and captured the poisoned pawn. This made the World Champion lose his tempo, eventually leading to a loss against Deep Blue. This blunder caused a series of accusations against Deep Blue as Kasparov never felt that a computer would fall for a trap like this for a computer lacks human emotion. Kasparov claimed Deep Blue’s moves were too “human-like”, perhaps even, that the machine was being fed moves by a human opponent.
At the end of the 20th century, the god of chess had been defeated at the hands of a man-made machine. Kasparov was in utter disbelief; little did he know that his defeat would go on to mark the dawn of a new period in chess history- “The Era of Chess Engines”. This was just a trailer for what was in store.
Chess is a battle of wits, believed to have originated from the game Chaturanga from India. 64 squares, 16 pieces, and 16 pawns. King, Queen, Rook, Bishop, and Knight; every piece has a unique style of moving. A bishop can only move diagonally, a rook, only linearly, a knight can move 1 and ½ squares on each turn, a queen can move linearly or diagonally in any direction, a king can move in any direction. The objective of the game of chess is to immobilize the opponent’s king with a threat to capture, that is, to create a “checkmate”. Due to a large number of possibilities of how a chess game can proceed, even supercomputers still haven’t figured out a definitive way to win the game. Chess Engines still rely on the opponent’s ability to make errors to win the game. In 1997, humans faced the first defeat against computers. For about the next 10 years, they traded blows, but by 2005-2006, the computers came on the top. The supremacy of chess engines happens mainly due to 2 factors; Moore’s Law i.e. the complexity of the hardware of the computer processors doubles almost every 2 years, and the improvements in the underlying software, increasing the ELO rating of Chess engines by 50 points almost every 2 years.
There are about 10120 ways in which a chess game can go, that’s actually more than the total number of electrons present in the Universe. Deep Blue used custom VLSI chips to execute the alpha-beta search algorithm in parallel. It was a brute-force search approach, and one of its developers even denied that it was artificial intelligence. All the games played against Kasparov were with standard time control which allowed the machine more time for computations. The first chess computer was attempted by Alan Turing around 1946 in which he assigned every piece a certain value, examined all white moves and all-black counter moves. After every move, a figure count was performed and the move which won the maximum material was selected. The problem with this method was that during the opening game almost all possible moves lead to similar results. This problem was solved by the Belle chess machine developed by none other than Ken Thompson, the developer of the Unix operating system, as he included an opening library and a hash map.
Deep Blue was a computer designed to play chess; programs that could run on regular workstations started to develop later. In 1998, a program, REBEL 10, defeated Vishwanathan Anand in blitz chess games (5 mins time control), although Anand, who was number two in the world back then, could defeat the program in classical chess games (standard time control). While millions of dollars were spent on production of Deep Blue hardware, REBEL software on a normal PC cost $2500. Later Israeli programmers came up with DeepJunior and claimed that it uses “opponent modeling”, that is it understands the weaknesses of opponents during game play and makes its moves accordingly. DeepJunior played nine grandmasters in a chess meet in 2000 and proved its mettle.
The next human-computer chess match was played in 2002 when reigning world champion, Vladimir Kramnik, squared up against a new program, DeepFritz. Unlike previous encounters, this time playing conditions were favorable for the world champion. When Kasparov played DeepBlue, the program was allowed to be modified after each game in a match, now the program had to be frozen sometime before the match. In addition to this, Kramnik was given a copy of the DeepFritz code a month before with which he could practice. All the games were played in classical format and after every 56 moves, he could adjourn the game, go back and try combinations with the code and come back prepared the next day. Even with so many advantages on his side, Kramnik could only manage a 4-4 draw against DeepFritz. Later that year Fritz-7 ushered a new era into world chess as it had the ability to play chess online on the Playchess server with thousands of chess players.
In a span of one year, Fritz evolved to X3DFritz with a virtual board, 3D glasses, and a speech-recognition system. From a brute-force algorithm-based computer to a 3D chess program having an ability to evaluate three million positions per second with a speech-recognition system, AI in chess had come a long way in less than a decade. Kasparov, the god of chess, played out a drawn match against this machine and was rewarded with $175,000 and a golden trophy for being able to hold himself against this super algorithm. The following two years saw Man vs Machine World Team Championships, the who’s who of Chess competed in the human team and the best chess computers geared up in the machine team. The computer team beat the men on both occasions making a statement that the gap between human ability and computer intelligence is widening.
As the computational skills of modern-day computers got better day-by-day, their gameplay got more robust to such an extent that today’s world champion, Magnus Carlsen, cannot score a single victory even if he plays 100 games against the world’s best chess program! Every chess engine that we can load today on our phone would be an equal opponent for a grandmaster.
The world computer chess championship has been conducted since 1974 but it was only after 2000 that people realized that there is no contest between a human and a machine. As of today, the Komodo chess engine has won it the most times with 4 consecutive victories from 2016 to 2019. Yet it isn’t the boss!
AlphaZero is one of the most recent developments in chess engines. It is developed by Google’s DeepMind Technologies. It consists of mainly two parts, a neural network and an algorithm called Monte-Carlo Tree Search. The Zero in AlphaZero indicates that it was based on Zero Human Knowledge, zero opening libraries, and databases. AlphaZero was trained by playing games with itself. In a mere time interval of 9 hours and about 44 million games, AlphaZero became the greatest chess player in the world. In about 4 hours of playing with itself, AlphaZero was able to defeat Stockfish.js (ELO 3428), the supposed champion of Chess Engines. AlphaZero can calculate about 60,000 positions per second, Stockfish calculates about 60 million positions per second. This has brought about a revolution in recent chess engines. The fact that an engine can outperform every other engine in the world, with no prior human knowledge like opening strategies, no midgame or endgame theory, is simply astounding. Whatever knowledge the program has is through self-learning.
In the past few years, chess engines have caused a major impact. Chess games, when played by GrandMasters were too difficult to understand, even more, difficult to predict who was winning. Chess engines have largely impacted that, as now, we not only know what is happening in the game, but the engines also explain which moves are better and why, making chess much more interesting. Recently, Chess streamers around the world have gained a lot of popularity. GM Hikaru Nakamura has gained almost half a million followers and about 10k daily viewers on Twitch. Streaming has opened a very new stage to earn revenue for Chess players. Today we have downloadable apps on mobile with which we can practice chess, the best example is Play Magnus app which is based on the journey of World Champion Magnus Carlsen in chess and has different levels starting from Magnus age 5 to Magnus age 29, so it has difficulty levels ranging from a novice Magnus to the world champion whom you and I cannot defeat.
From a time when bragging grandmasters thought that there can never be a computer smart enough to beat them, to an era in which it is impossible to find a human who could beat a chess engine, the journey of AI in chess has truly been fascinating! The progress AI has made in chess is unparalleled, no other sport in the world has provided such a platform for programmers to flourish. Today developing and working on chess engines has become a full-time job for many engineers. Thus, chess has given a huge scope for technological progress. When it comes to being a battle of wits, this game truly trumps all.
-Anurag and Rohan