That's a laaaaaat of information with 3 questions.
Play against Computer - how does it work?
Sort:
I would like to know the approximate rating strengths of the ten levels. I can reliably beat it on 5, but at 6 I have to concentrate and think every move through, and the computer still wins half of the time. I've noticed some flaws in its play at 6; it will unwisely move its king, forfeiting the right to castle, and sometimes when I play PxP, it will not recapture immediately, allowing me to take more pawns. "A pawn ahead is worth a little trouble." --Wilhelm Steinitz
It's not easy to programs computers to play like humans at different rating levels. So it's probably the same engine for each, but at lower levels it just intentionally blunders early in the game.
Hello
I have a question: If you click 'Play' --> 'Computer' you can play against the computer.
There, you can choose between 10 difficulty levels.
My question: How do they realize these 10 difficulty levels?
Background: I play the computer at Level 3 or 4. I ALWAYS find myself winning in the opening, the computer blunders large amounts of material, most of the time 1-3 pawns and one piece.
However, the fewer pieces (and pawns) are on the board, the more accurate the computer plays. No more blunders, or even mistakes sometimes in the endgame. It's quite hard to win an endgame for me against him.
So I ran the analysis mode, and sure enough, 90% of the blunders / mistakes happen in the opening / middlegame.
Here's my theory: The 'Difficulty levels' aren't difficulty levels at all, but time-restrictions for the engine.
Setting it on a lower difficulty actually means you play the same, strong engine, but with greatly limited calculation time.
As long as there are many pieces, there are many calculations for the engine, and because there isn't enough time to find the best moves, it 'Blunders'.
As soon as most of the pieces are off the board, the possibilities become way less, and therefor it plays more accurate.
Can anyone verify that my theory is true?
And if it is, why doesn't the calculation-time get reduced even further dynamically, once the pieces come off the board, so the Computer-opponent plays a consistent difficulty for a whole match?