Comment by gowld

Are you saying that, to encode a valid position, ignoring the cost to encode both the starting position and the move definitions, you can encode the move sequence using less information than encoding a single position?