Jon Ostriker recently earned the title of Correspondence Grand Master to become the 10th US player to receive the title. Ostiker was interviewed in this month’s Chess Life magazine and freely admits to being a centaur. That’s a term, often derogatory, for a player than uses an engine, but as I have pointed out in previous posts, these days it includes many players at the top of most Internet sites and all official titled players playing for the ICCF. Officially a centaur is described as a player who uses both chess engine input and his own strategic knowledge. It’s that last part, uses his own strategic knowledge, that distinguishes a centaur from a pure engine user.
In the article Ostriker explained the necessity of using engines at his level of play and also explained some of the dangers in letting engine be the final judge of what move should be played…at least at his level of play.
Ostriker emphasizes a point I have made many times when he said, “If a mid-level player had a fast chess computer and …always followed the recommendation of his chess engine, he would probably finish last in an ICCF candidates tournament.”
The reason, as he pointed out, is that engines don’t always correctly evaluate a position and they do poorly, for example, at recognizing a fortress, where, for example, one side may be down a piece, but the other side is unable to force a win using his material advantage. In order to be successful at his level of play a centaur must use his judgment to determine how much he can rely on the engine in each position.
He said that very often he has to ignore the engine’s top or second recommendation because the engine’s ply horizon is not deep enough to correctly evaluate key lines. As a result they will often miss the strategic importance of certain themes: positional sacs of the exchange, fortresses, Zugzwang and very deep sacrificial attacks. I’ve mentioned in a previous post that I have noticed a tendency of engine evaluations to be in disagreement with GM evaluations in positions with material imbalances. I also opined that I will rely on the GM’s judgment over the engines in every case.
As Ostiker points out, engines are just another tool at the international level. You have to include a lot of other tools (mentioned in previous posts) in your arsenal. He mentions one key point to knowing when to start looking deeper into the position and questioning its evaluation is when you see evaluations fluctuating. Also, as I previously mentioned, he says CC at that level requires skills most of us don’t have: research, patience and excellent positional judgment.
In the interview Ostiker also commented that to him the first CC world champion, C.S. Purdy, remains a chess hero and he recommends the book How Purdy Won. Purdy wrote that in many complex chess positions strategy is almost irrelevant. What matters is finding the right tactical stroke.
An even better source for Purdy material is probably Bob Long's site: The Chess Museum