Hats off to the Dragon for holding out for so long. Leela seems very balanced, turning a spatial advantage into a material one and showing coolness under fire.
One way you can imagine this is if the table base losses are just evaluated as a loss. So the computer thinks, okay I am choosing between a loss in 1 move or a loss in 2 moves. But yeah definitely funny.
I like to play the modern, but also the Austrian against the modern, so it's always cool to see how others treat the game from the opening. These videos are phenomenal, thank you
There is a database of all legal 7 piece positions (2 Kings and 5 other pieces) where the end (Win, Loss, Tie) is known with perfect play. So in the game if the Queen took the rook there would be a position with 7 pieces and the database says the position is winning for black with perfect play which the computers can do. So white doesn't take the rook to put itself in that position.
@@ryanswanson126 White to win, not black. (Black's turn to move anyway, if Komodo saw a table-base win for themself, wouldn't it be the logical choice? These bots aren't prideful, to my knowledge.)
@@nicbentulan my understanding is yes, you look the position up in a table and it gives you the answer. The computers are trained or programmed with those tables in mind so they use that to inform their play.
@@nicbentulanExactly right, all chess positions with seven pieces or less have now been solved. In a few years perhaps they will have completed eight piece tablebases, and so on. I guess the entire game of chess won't be solved until 32-piece tablebases have been produced! 😄
2:45 - I analysed this with Stockfish 10, and I noticed some interesting things here. First off, initially Stockfish thinks f5 is 0.0. Then at Depth 22 it thinks f5 is +0.6 and e5 is +0.5. Then lastly at Depth 27, it boosts the f5 evaluation to +0.9
Very interesting game, lots of position strength grinding from white, black looks desperate just a few moves in. Could someone elaborate on the endgame tablebase thing? Is it triggered whenever the 8th last piece is taken, to check if one side or the other is doomed at that point? Why does it make sense for computer games, which don't lack time to play uninteresting endgame moves? Quite curious on that
The engines have taught me a lot about the power of waiting. The old me would have maybe snatched that pawn on D6, but they are confident in never decreasing pressure. BC1! to me is the best move in this game. Obvious advantage gained by the exchange, but they want to get the most juice from every squeeze, instead of busting their nut early.
This position reminds me a lot the Gran Prix attack, i've studied and played counterless hours the pawn sacrifice agaist the fianchetto/dragon it's really tough for black to come out from that attack
Very interesting opening I always thought Nfd7 was the only non-losing move but it turns out that Bd7 is better and may be drawing against e5 (but probably still lost?). I’ve played this opening a lot against the Old Benoni.
Komodo is clearly inferior to the more streamlined Leela. But what impressed me were the alternative lines (usually for Komodo) you put on. Did you calculate those yourself, Jerry?
I mean this game is from the Top Chess Engine Championship (TCEC), it's pretty much constant games of engines against eachother, you can go to the site and see the winners and games from past seasons
Not for about a decade at this point. The team behind it never meant to revolutionize chess, they did by accident. And they just moved on to a new project, their Go engine. After thtat I have no idea what they did, but point is chess is done for deep mind.
Leela = Wesley So Stockfish = Magnus Carlsen - because of 9LX game in 2022Jan Deep Blue = Bobby Fischer Garry Kasparov= Garry Kasparov - Deep Blue is American Komodo = Hans Niemann - Larry Kaufman told me 'Magnus doesn't generally play such great openings, he strives to get the game out of book as early as possible usually. I think the issue here is that his greatest strength is the endgame, but FRC games are much more likely to be decided in the middlegame as the players are on their own so early. That's probably why he doesn't shine as brightly in FRC as in Classical chess.' (??) = Hikaru Nakamura ? I don't know any other modern engines besides Stockfish because Stockfish Leela because beat Stockfish Komodo because Larry Kaufman
