When all else is equal, I certainly can choose between the two, because I have nothing else to go on. But two different moves, searched to two different depths, is a completely different animal. This problem has been studied at length and the conclusion has been "NFG" each and every time. You can find lots of papers. Schaeffer even encountered this in a different way, by running two different searches on the same moves, one a normal search, one a tactics-only search (material-only eval). The idea being that the tactical search could go deeper and refute the "normal move" by finding a deep tactic. But then what? You can't let the tactical search choose the move, it has no positional knowledge. So you are back to square zero and all you can do is force yourself to go a ply deeper to see if the positional search can find something, or else re-start the iteration excluding the failed move to see if another can do better after it gets verified.jwes wrote:By the same logic, you can't choose between depth 20, eval=+1.3, and depth 20, eval +2.5. If there is not a very high correlation between scores at ply n and ply n+1, standard computer searches could not work at all. I parsed some log files and came up with a correlation coefficient of 0.995.bob wrote: this is useless.
How can you choose between depth 21, eval=+1.3, and depth 19, eval +2.5?? You can't. This has been tried in the past, by Newborn, by Schaeffer, and by others. There is no way to compute any sort of equivalence function so that you can decide which of the above is better. The depth 19 move might be even higher at depth 21. Or it might be way lower. The only way to discover this is to search both moves to the same depth. Anything else is beyond hopeless and is a coin flip. You can't even choose between depth 21, +1.5, and depth 19, -2.0, because by depth 21 the -2.0 score might be +5.0...
BTW there is a _high_ correlation between iterations. 85% of the time the best move at N-1 is still best at N...