Re:Magnus Carlsen Beats Fabiano Caruana to Win World Chess C...
by Someone on Wednesday November 28, 2018 @05:07PM
(#18556137)
”It's currently +3-18=14. That's 3 wins for Caruana's position, 18 wins for Carlsen's, and 14 draws. That's about as decisive a position as you can get in high-level chess.”
Today’s games stand at +3-0=0 or, taking the last game as the draw it would have been if Caruana didn’t have to win, at +2-0=1.
I haven’t done the math, but I’m sure taking a “best of four” sample from the distribution that was taken from gives you better odds than taking a n=1 sample from the distribution that gave you +3-18=14 (about 9% chance of losing, about 50% of winning’, followed by (if that draw is a tied game) that n=4 sample.
And that’s ignoring the fact that that +3-18=14 is from machines that are way different from humans. When given a chess position that’s like a minefield, with a single winning sequence of 40 moves, and all other moves a loss, a computer will happily declare the position a certain win, but chances are no human will find that winning sequence.
Today’s games stand at +3-0=0 or, taking the last game as the draw it would have been if Caruana didn’t have to win, at +2-0=1.
I haven’t done the math, but I’m sure taking a “best of four” sample from the distribution that was taken from gives you better odds than taking a n=1 sample from the distribution that gave you +3-18=14 (about 9% chance of losing, about 50% of winning’, followed by (if that draw is a tied game) that n=4 sample.
And that’s ignoring the fact that that +3-18=14 is from machines that are way different from humans. When given a chess position that’s like a minefield, with a single winning sequence of 40 moves, and all other moves a loss, a computer will happily declare the position a certain win, but chances are no human will find that winning sequence.