Widely unknown pioneering chess "paper machine" by Gunter Sсhliebs
Posted: Thu Jun 06, 2019 11:52 pm
Turochamp, Machiavelli, SOMA, SOMAC, Shannon's programs — we know a lot about them, even there are several reconstruction attempts especially by Chessbase team (Frederic Friedel, Martin Feist). However with a German attempts it is known a little. Zuse's first works on Plankalkül provides us with some move-generation routines, but I can't find any sources for his ideas on move selection. Guys from PyTuroChamp team (https://mdoege.github.io/PyTuroChamp/engine.html) made their reconstruction of Zuse's engine (https://github.com/mdoege/PyTuroChamp/b ... er/plan.py) and they even digged piece values from somewhere: 1 for a pawn, 2 for a knight or a bishop, 3 for a rook and 4 for a queen. I hope they don't invented these strange numbers themselves but I'm still unable to find any clue in Zuse's online archive.
But the paper machine of Günter Sсhliebs from DDR was never referenced in Western sources. In Soviet books and articles it was referenced multiple times (especially in the works of legendary Anatoly Kitov — [1], [2], [3]) among with Shannon's work. The original source for Schliebs' work is [4]. I never seen [4] (but started some attempts to obtain the scan of this article), but Kitov's works provides several details, at least evaluation function structure and some feature scores and also the point that Schliebs' "program" was fixed-depth searcher.
Evaluation features of Schliebs' paper machine:
queen — 9 points
rook — 5
knight, bishop — 3
pawn — 1
backward pawn penalty — 0.5
isolated pawn penalty — 0.4
doubled pawn penalty — 0.3
mobility (for each field for "most strong pieces" (?))
One work of Kitov occasionally assign 0.5 score for the pawn instead of 1.0 (mistake?)
1. Китов А.И. (1956). Электронные цифровые машины // http://www.kitov-anatoly.ru/naucnye-tru ... vye-masiny
2. Китов А.И., Криницкий Н.А. (1959). Электронные цифровые машины и программирование // http://www.computer-museum.ru/books/ecm_i_prog.pdf
3. Китов А.И., Криницкий Н.А. (1958). Электронные вычислительные машины и программирование // http://elib.ict.nsc.ru/jspui/bitstream/ ... ov1958.pdf
4. Sсhliebs G. (1953). Über die Gründzuge eines Programms für eine Schachspielende Rechenmaschine / Funk und Ton, 1953, vol. 7, pp. 257—265.
But the paper machine of Günter Sсhliebs from DDR was never referenced in Western sources. In Soviet books and articles it was referenced multiple times (especially in the works of legendary Anatoly Kitov — [1], [2], [3]) among with Shannon's work. The original source for Schliebs' work is [4]. I never seen [4] (but started some attempts to obtain the scan of this article), but Kitov's works provides several details, at least evaluation function structure and some feature scores and also the point that Schliebs' "program" was fixed-depth searcher.
Evaluation features of Schliebs' paper machine:
queen — 9 points
rook — 5
knight, bishop — 3
pawn — 1
backward pawn penalty — 0.5
isolated pawn penalty — 0.4
doubled pawn penalty — 0.3
mobility (for each field for "most strong pieces" (?))
One work of Kitov occasionally assign 0.5 score for the pawn instead of 1.0 (mistake?)
1. Китов А.И. (1956). Электронные цифровые машины // http://www.kitov-anatoly.ru/naucnye-tru ... vye-masiny
2. Китов А.И., Криницкий Н.А. (1959). Электронные цифровые машины и программирование // http://www.computer-museum.ru/books/ecm_i_prog.pdf
3. Китов А.И., Криницкий Н.А. (1958). Электронные вычислительные машины и программирование // http://elib.ict.nsc.ru/jspui/bitstream/ ... ov1958.pdf
4. Sсhliebs G. (1953). Über die Gründzuge eines Programms für eine Schachspielende Rechenmaschine / Funk und Ton, 1953, vol. 7, pp. 257—265.