The red move of 2 to 7 seems very strange to me… I think you would have to be very very good to see this as a good move. or maybe it is not a good move i know in a game i would have never made that move…
Well, 2-7 is a not a good move. But the engine claims it’s the best of a bunch of lousy options. I agree that it’s not a very human try. I think a human would probably try 17-21 where the problem seems to be that red has no access to a double corner because of the white piece on 32. That combined with white’s easier access to additional kings and control of the center of the board spells disaster for red.
The puzzle should probably try that path but I don’t have a good way of adjusting it yet. If you open the ‘defenses’ button next to 2-7 it will show you some lines occurring from other moves.
I didn’t have time to look at this problem in any detail but this comes up very often. The computer will choose a move that it evaluates to lose more slowly. This often means pitching a piece just to delay the inevitable. It is a very non-human approach, and when I publish analysis in my columns, I will often substitute a move that someone might actually play over the board, as there is little educational value in an artificial move that may be “better” by some objective criterion but really makes no human sense, and more importantly, no difference in the end result.
But we have to be careful, too. I’ve seen instances where the computer comes up with some amazing deep line of play starting with a nearly unthinkable move.
Indeed. For this reason, when the position in the puzzle occurred in the source game the problem generator uses the defense that was tried, even if it’s not the best move. But if the attacking player missed the opportunity or chose one of the alternate lines, so the position didn’t occur, then I have to use the computer move.