They show number of blunders (score dropping 1.5 or more) categorized by player strength and current position evaluation according to either Komodo or Stockfish. The conclusion is that people blunder much more in worse positions, but they blunder most in dead even positions.
I'm not following all of the theory involved, but this makes me wonder if the "firewall" is just an effect of the scale on which they're measuring positions.
Chess engines evaluate a large range of positions as having value 0.00. In fact, an omniscient chess engine would evaluate every position as either having value 0.00 or as being "mate in N" (where N could be unreasonably large).
So maybe the blunders from positive positions are smaller because, as evaluated by a sufficiently strong chess engine, they mostly drop you from +(small number) to 0.
Comments
How do you read the charts? I can't make head or tail.
They show number of blunders (score dropping 1.5 or more) categorized by player strength and current position evaluation according to either Komodo or Stockfish. The conclusion is that people blunder much more in worse positions, but they blunder most in dead even positions.
At least I think that's what it says:-)
I'm not following all of the theory involved, but this makes me wonder if the "firewall" is just an effect of the scale on which they're measuring positions.
Chess engines evaluate a large range of positions as having value 0.00. In fact, an omniscient chess engine would evaluate every position as either having value 0.00 or as being "mate in N" (where N could be unreasonably large).
So maybe the blunders from positive positions are smaller because, as evaluated by a sufficiently strong chess engine, they mostly drop you from +(small number) to 0.