basically A* which im sure you know does path finding in practically every game for the last 5-6 years possible longer. its a good algorithm but unless highly optimised it could really bottleneck your game.
so this method basically implements a version of a* which is easy to program but is highly optimised (algorithmly), instead of programming specific shortcut, which claims to improve by 40% percent or so.
i have yet to implement myself but when i get a chance ill let ye know. i do know the guys from the game companies who were there were pretty impressed, scribbling like notes like made and asking questions, which you knew were specifically aimed towards their next games or own purpose. one guy from codemasters i think, told me he was going into work tomorrow and changing their implementation to see if he could get more out of his ai.