Retro Computer Chess part 1: The paper tiger
In 1770, Wolfgang von Kempelen presented the world to The Turk, a chess playing automaton unlike anything the world had ever seen. It was all a hoax, however; The mechanical chess player was operated by a human player concealed inside the cabinet. But the very idea of a machine beating a human player at chess obviously tickled the imagination of people centuries ago. A lot has changed since then.
In these blog posts, I wish to take a look back at some of the earlier days of computer chess. Using emulators and other resources, I will be testing myself against a number of older chess programs and engines. A lot of these will be from the 70s and 80s. But I will start even further back.
In 1913, more than a century ago, the german mathematician Ernst Zermelo published the paper "Über eine Anwendung der Mengenlehre auf die Theorie des Schachspiels" ("On an Application of Set Theory to the Theory of the Game of Chess"), describing a mathematical approach to playing chess. This approach, called the Minimax algorithm, is still essential to most chess engines constructed to this day.
Fast forward to the late 1940s, and we find a brilliant young man named Alan Turing:
Turing is most famous for creating a general purpose electric computer that was used during WW2 for cracking codes created by the german Enigma machines. This effort was hailed by Churchill as the single biggest contribution to the allied victory against the germans, and has been estimated to have shortened the duration of WW2 with somewhere between 2 to 4 years. So, what do you do when you have single handedly shortened the biggest war in human history? How about a nice game of chess? So, young Alan designed a chess playing program, sometimes referred to as Turochamp or Turbochamp. Unfortunately, he did not have the means to actually construct it, so the only recorded game he did with it was performed with him calculating all the moves by hand using pen and paper, in a game against Turings colleague Alick Glennie.
Turing actually did try to implement his program on the Ferranti Mark I computer at Manchester University, but he never did manage to complete it. However, more than 50 years later, some programmers at Chessbase picked up Turings old code, brushed it up, and implemented them as a Chessbase compatible engine and put it up for download on their website. They made a couple of alterations, including handling of situations like stalemate and lone kings, which had bugs or were not handled in the original code, but apart from that the engine should perform like Turings original paper engine.
Unfortunately, Chessbase has pulled the setup file for the engine, so actually finding a copy today can be problematic. But if you can get your hand on it, you can find yourself playing games like this:
The Turing engine has even been seen playing against Gary Kasparov himself:
So, what can be said about the Turing chess program? At 2 plies search depth, it will easily fall into tactical traps. And the evaluation function has it's quirks, such as early queening and making room for the rooks by moving the rook pawns. But good ol' Alan certainly did a lot of things right. And with the Chessbase engine running on a modern computer, capable of performing deeper searches than Alan himself could hope to do using paper and pencil, it should also serve as an engine that can be both interesting and challenging to a beginner player.