The Evolution of Chess Engines
Today we're going to be diving into the history of chess engines! I assure you, it's going to be a very exciting story!
Chapter 1: The Beginnings
The story of autonomous chess stretches all the way to the Mechanical Turk of the 1770s. It was originally designed by Wolfgang von Klempelen was essentially a large wooden cabinet with a fake turbaned Turkish man behind it. He has something like a puppet arm to move the pieces. Inside was a complex system of gears and levers, whenever the opponent moved their piece a magnet would drop down giving the master inside of the box time to replicate his move on another chess board inside the box before the magnet lifted back up. As you probably realised this wasn't a true chess engine, however the hoax was only revealed after von Klempelen's death, which was in 1820. Even though this was a hoax this was one, if not the first to spark an interest in autonomous chess, which eventually developed into what we know today.
Chapter 2: The First True Theories
In the mid-20th century, Alan Turing, who was a famous figure in computer science laid a lot of the groundwork for chess algorithms, which was titled "Digital Computer Applied to Games." However this during his time was not applied to a computer, for several reasons, the main reason was that it was too complex for computers to calculate during that time period. However again in 1950 Claude Shannon another computer scientist published "Programming a Computer to Play Chess." Unlike Alan Turing's book, which introduced the concept of evaluation functions and search algorithms, these 2 books were the starting blocks of the modern chess engines.
Chapter 3: The First...Puzzle Solving Chess Engine
In 1951 Dietrich Prinz, a German computer scientist introduced the Ferranti Mark 1. One of the earliest chess "computers." However it was only capable of playing simpler mate solutions like mate in 2, but it's something.
However 5 years later, the first "true" chess computer came out, and it was called the MANIAC I, it was capable of playing a full game of chess albeit with simpler rules, but the MANIAC I marked a significant leap in chess computing and chess engines.
In the 1960s and 1970s computer scientists were racing through and making significant progress in chess computers, one of the most notable was the Mac Hack VI developed by Richard Greenblatt at MIT, specifically developed in 1967.
Now if we go to 1970-1980s then we'll find the rise of microcomputers, (aka smaller than giant shelves of computers.) Which happily enough brought chess computers to a broader audience, like most of us reading this right now. The most notable, or one of the most noticeable would be Chess 4.5, which was made by Larry Atkin and David Slate. Chess 4.5 won several chess computer championships at that time.
Chapter 4: The Start of the modern Chess engines
In the 1980s marked the era of the start of modern chess engines as we know of today. The 1980s introduced several new chess engines, the most popular of which was Fritz, which still exists today. Fritz was developed by ChessBase and had a rich history, and in fact it's still being updated, the most recent being Fritz 18 Neuronal.
A decade later from 1980s would mark the year of 1990, which also marked what people usually know as the decade when a chess engine beat the first chess GM, Garry Kasparov. And that chess engine was no other than Deep Blue! Although Deep Blue lost the match in 1996, it did win 1 game and one year later in 1997 it won the match against Garry Kasparov.
Black Resigned
Chapter 5: The Modern Times Begin!
The first notable chess engine released was our beloved Stockfish, which was in 2008. Developed by Tord Romstad, Marco Costalba, and Joona Kiiski, this engine was derived from Glaurung an open-sourced engine made by Tom Romstad in 2004. It quickly gained support because of its strong performance and open-source nature. And in August 2020, it adopted the Efficiently Updatable Neural Network (NNUE), which improved its evaluation accuracy by miles.
This was one of the first major games played by Stockfish, showcasing its power against Rybka
In 2010-2019, there were 3 notable releases 1 of which in 2012 was Komodo made by Don Dailey and Mark Lefler, it combined traditional with neural making it at the time, one of the top engines. In 2018 Chess.com acquired Komodo, which brought in a lot more supplies to update the engine, eventually evolving it into what we know as today as Dragon, which also used NNUE.
The second notable chess engine during the time span of 2010-2019 was AlphaZero, developed by DeepMind and released in 2017. Its goal? To master chess, shogi(Japanese chess), and Go. What made it so unique was that it trained solely on self-play.
The third notable one is Lc0 or Leela Chess Zero, developed in 2015, and in 2017 started getting competitive against Stockfish inspired by AlphaZero's self-training method. Lc0 was developed by Gary Linscott.
In 2023 a new kid arrived onto the block, Torch. A chess engine released by chess.com and developed by Andrew Grant (author of Ethereal), Finn Eggers & Kim Kåhre (authors of Koivisto), Jay Honnold (author of Berserk), and Michael Whiteley & Dietrich Kappe (current authors of Dragon). Quickly taking #2 and only losing to the rapid event, Torch proved itself to stay in the competition. In 2024 Torch 2.0 was released, which was primarily to improve performance, and to improve as an analysis tool.
