The interesting and wide-open questions are then:
- Where exactly is the line between "playable" and "unplayable" in the opening? Or, how big is the gray area between "playable" and "unplayable", can it be reduced, and if so how and how far?
- What is the function from computer evaluation value to numeric probability? Or is this an apples-to-oranges comparison because of the inherent uncertainty in the value (dependent as it is on how much, how deeply, and which parts of the move tree the computer has searched, let alone the computer's underlying base (0-depth) evaluation function)?