It used to be that the two-squares rule had a maximum of 5 consecutive back-and-forths of the same piece between the same two squares. Now it is 3. As is spelled out in Vincent de Boer's thesis on Stratego programming from 2007, the possibility of placing 3 interposing moves at any point, will greatly inflate the required search depth to solve tactics.
I am in the process of programming a new Stratego AI and have found the same problem to be still a major obstacle on modern computers.So I wondered if a shortcut was possible that still gives the same game-theoretic results as for the official ISF rules, but which is a lot cheaper to compute.
Modified two-squares rule: a piece cannot move more than 1 consecutive time between the same two squares, except to capture a piece that was placed on the just vacated square, or if the opponent revealed a new piece of information (moved an unknown piece, or revealed a piece's rank).
Note that this is not a proposal to modify the ISF rules, just an experiment to greatly reduce a computer program's search depth to solve tactics.
Question to the experts: is this rule modification consistent? In other words, are there any position where this rule (with at most 1 consecutive move, except for captures or to act on new information) gives a different outcome than the ISF rules (with at most 3 consecutive moves)?
Edited by TemplateRex, 25 June 2018 - 12:25 PM.