It seems to me that chess engines are closely analogous to such filters. At the end of a search to depth (d) in any given position, the average branching factor of the number of evaluated nodes (n) is effectively the engine's 'Q':
Code: Select all
Q = n ^ (1/d)
There seem to be two main advantages to such a scheme. First, engines could be compared on entirely different hardware. Second, given a wide enough mix of test positions - strategic, tactical, opening, end-game - it would be possible to identify clearly where any engine's individual strengths and weaknesses lie.
So, does the panel think there is any merit in the idea?