Chinook Seminar - Review

It was an entertaining seminar by Dr Jonathan Schaeffer this afternoon, although it did have its fair share of hyperbole. The majority of the talk was about the journey from the start of Chinook (his checkers playing program) until it defeated Dr Marion Tinsley for the World Man-Machine Checkers Championship.
There were a couple of things in the talk that stood out to me as a chess player. Firstly, Chinook was started the day after his chess program "crashed and burned" in the 1989 World Computer Chess Championship. I guess like a lot of chess players, losing can create a "fight or flight" response in computer chess developers. Secondly I disagreed with his reasoning behind why a lot of effort was put into computer chess and not computer checkers (He cited publicity surrounding the 1963 Samuels Checker Program as the cause, I just think computer chess was more of a challenge). And thirdly, one of his slides concerning the 1994 Man v Machine Checkers Match had a picture with GM Raymond Keene in the background, but this went unremarked by the presenter.
What was really interesting was his motivation for "solving" checkers. The final match between Chinook and Tinsley ended with the first 6 games drawn, before Tinsley resigned for health reasons. The health reasons turned out to be pancreatic cancer and Tinsley died before another match could be organised. But just as in chess, players often become stronger after they die, and a number of checker players claimed that Chinook wouldn't have beaten Tinsley in his prime. Short of building a time machine the only way to disprove that claim was get Chinook to a point where it could provably not lose. Which he then did.
At the end of the talk I asked a question about whether it was a "strong" solution or a "weak" solution. It turned out the solution is a "weak" solution. In a "strong" solution, it is possible to assess every position as a win,loss or draw. What Chinook did was start with a specific opening move and assess every position leading from that. They did this successfully, finding that if Chinook played a specific first move then by best play the opponent could only draw. They then checked all the other openings but only established that the positions leading from the other openings were either draws or losses, with best play. The reasoning was simply that as they already had a winning/drawing first move, Chinook wasn't going to play anything else as its first move. But if Chinook played second and the opponent chose a different first move, they only needed to establish that opponent couldn't win with that move. They didn't go as far as to establish the best play value for the other first moves (which may have indeed been a loss for the first player), as this was unnecessary for the overall proof.