Cyrus Chess

     One of my first computers in the early 1980s was a Radio Shack TRS-80 Color Computer 2, generally known as the CoCo2; it plugged into a TV and the programs ran off cassette tapes. You could buy a magazine dedicated to the computer and spend evenings and Saturday mornings programming it in BASIC to do all kinds of neat things. 
     Naturally, when the chess program Cyrus came out, I had to have a copy. Cyrus was written by a fellow named Richard Lang and it was his first chess program. Lang started programming in January 1981 and made his tournament debut at the 2nd European Microcomputer Chess Championship at the PCW Show in 1981 in London. Cyrus was the clear winner with 5 out of 5 in a field of 12 microcomputers. 

CoCo 2

     Lang was immediately offered two contracts by David Levy and Kevin O’Connell, one for Cyrus, and one to work as programmer for Intelligent Software. Lang accepted and Cyrus IS-Chess for the Sinclair ZX Spectrum was his first commercial entry, followed by programs for various dedicated chess computers merchandised by Intelligent Software, as well a further improved version of Cyrus, Cyrus II. In an 2003 interview, Richard Lang stated that there is still much of Cyrus in current versions of Chess Genius.   
     Back in the day, Cyrus was one of the world's first chess games designed for strong players, but it was probably one of the least known due to its limited distribution. 
     Cyrus had varying level of difficulty based on how long it was allowed to think and the time it took over a move on each level could differ significantly from the average. Times were 1, 10, 20, and 30 seconds, 1, 2, 6 and 12 minutes and infinite. 
     When Cyrus came out for the CoCo 2 it was claimed that it had been greatly improved since it had won the second EMCC. The manual claimed that it offered advanced players a challenging game under tournament conditions and in the fast mode it was an ideal opponent for beginners. Some of the features were: it could play against itself, it could show you its thinking, it allowed for setting up positions for problem solving or analysis and you could print out games. 
     In the following game Cyrus succeeded in defeating Cray Blitz, a program written by Robert Hyatt, Harry L. Nelson, and Albert Gower to run on the Cray supercomputer. It was derived from "Blitz" a program that Hyatt started to work on as an undergraduate. "Blitz" played its first move in the fall of 1968, and was developed continuously from that time until roughly 1980 when Cray Research chose to sponsor the program. Cray Blitz participated in computer chess events from 1980 through 1994 when the last North American Computer Chess Championship was held in Cape May, New Jersey. Cray Blitz won several ACM computer chess events, and two consecutive World Computer Chess Championships, the first in 1983 in New York City, and the second in 1986 in Cologne, Germany. The program Crafty is the successor to Cray Blitz. 
     Analyzing this game between Cyrus and Cray Blitz with Stockfish (and Komodo 8) leads to a curious phenomenon I have seen before. At moves 30, 33 and 40 they show an evaluation of 0.00, but after you actually make their recommended move for black, the evaluation jumps back to the correct one of a 2-3 Pawn advantage.
     For example, make 30.Nb5 on the board and the engines show 30...Rb7 with no further moves and the evaluation immediately dropped to 0.00. This happened with Stockfish, Komodo 8, Crafty 23.01, Fire 4, Fritz 5.32, Giraffe, Rybka 2.3.2a, Roce, SmarThink and SugaR PrO. Only Fritz 12 and Houdini 1.5a, which also recommend 30.Nb5, show white with a substantial advantage after 30...Rb7; they also, as is normal, show additional moves. I don't know why this happens, but I have run into it a few times when analyzing games on LSS where engine use is allowed. Thoughts? Can anybody duplicate this situation?