EVEN COMPUTER ANALYSIS CAN NOT ALWAYS FIND THE BEST MOVE


  • 4 years ago · Quote · #21

    Uddipto

    sorry i have made a mistake in the diagram. I will cure this when i will be on my pc.

  • 4 years ago · Quote · #22

    Irontiger

    Moses2792796 wrote:

    @cookiemonster, this is simply wrong.  There are many positions where Houdini can't find the best moves that are relatively obvious to humans

    If "many" = a thousand of possible positions (or ten thousands or...), maybe.

    If "many" = a significant fraction of the positions that occur in real games, not compositions, no.

    There is no strategical concern that materialize past a 50-move horizon. And today's computer strength tend to this...

  • 4 years ago · Quote · #23

    vannorden2

    17. Bxf8 is a blunder (... Qxf7) so it is impossible that it has been suggested by any computer.

    Computers are mostly right if you give them some time. I use Hiarcs (only available for Mac) which comes up with the most active/ aggressive continuations.Impossible to beat for me, unless you limit it's time to 5 seconds and allow yourself to think minutes.

  • 4 years ago · Quote · #24

    C-nack

    vannorden2 wrote:

    17. Bxf8 is a blunder (... Qxf7) so it is impossible that it has been suggested by any computer.

    Computers are mostly right if you give them some time. I use Hiarcs (only available for Mac) which comes up with the most active/ aggressive continuations.Impossible to beat for me, unless you limit it's time to 5 seconds and allow yourself to think minutes.

    If you beat an engine that has 5 seconds for move while you have minutes, then this is a pretty shitty engine.

  • 4 years ago · Quote · #25

    TetsuoShima

    i hope today my mind wont fail me.

    When you play knight to g5, you d pawn will be unprotected. If he takes it he also guards the his pawn on e6 so that your tactic doesnt work. Also you bishop is hangin and if he takes it with the knight, he gets a free move because it would be check.

    You cant move the bishop because the queen is attacked and would drop.

    If you move the queen you other bishop probably falls.

    My guess is if you try to exchange pieces then like queen takes queen and bishop takes bishop to save the piece the position probably even gets worse, because you will suffocate....  he probably later will either take the c pawn and have a free pawn or you will have to defend awkwardly so you position will be completly losing.

    I dont know if you really wanted help or if it was symbolic for something else, but i hope i could help you and that my analysis was correct.

     

    why bishop takes f8 is good i dont know. Seems like losing...

    if the other position was wrong as well, then i at least had something to do

  • 4 years ago · Quote · #26

    e4_guy

    16.Qxe6+ is strongest move. Gains material + saves the bishop.

  • 4 years ago · Quote · #27

    TetsuoShima

    e4_guy wrote:

    16.Qxe6+ is strongest move. Gains material + saves the bishop.

    you are right

  • 4 years ago · Quote · #28

    Uddipto

    corrected the mistake

  • 4 years ago · Quote · #29

    Uddipto

    vannorden2 wrote:

    17. Bxf8 is a blunder (... Qxf7) so it is impossible that it has been suggested by any computer.

    Computers are mostly right if you give them some time. I use Hiarcs (only available for Mac) which comes up with the most active/ aggressive continuations.Impossible to beat for me, unless you limit it's time to 5 seconds and allow yourself to think minutes.

    thanks , i did not notice the mistake

  • 4 years ago · Quote · #30

    Irontiger

    After the change, it looks better.

    I side with Houdini for this one. A pair of minor pieces are usually better than rook + pawn (because that's the trade your line makes - when Houdini"s line ends up a pawn for White, yours is rook + 2 pawns vs. knight and bishop), and it's even more true here where few pawns have vanished from the board (which favors the minors over the rooks).

  • 4 years ago · Quote · #31

    C-nack

    Ng5 leads to a lost endgame.

    Qxc7 is not the best move.

    Qxe6+ is the best move.

  • 4 years ago · Quote · #32

    Uddipto

    C-nack wrote:

    Ng5 leads to a lost endgame.

    Qxc7 is not the best move.

    Qxe6+ is the best move.

    how a lost endgame?

  • 4 years ago · Quote · #33

    C-nack

    Uddipto wrote:
    C-nack wrote:

    Ng5 leads to a lost endgame.

    Qxc7 is not the best move.

    Qxe6+ is the best move.

    how a lost endgame?

    Just check it with your houdini. Black has a passed d pawn, white rook is restricted, white king is in the corner. Doesn't look very promising. (unless your engine finds a better solution, I'm not on my computer so I can't prove it with a concrete engine variation).

  • 4 years ago · Quote · #34

    Uddipto

    ya . it is so . thank u c-nack

  • 4 years ago · Quote · #35

    Irontiger

    cookiemonster161140 wrote:

     We may not quite be there yet but when computers are able to effectively abandon unproductive lines and focus more on that which will likely bear fruit in a chess position then we're doomed.

    Depending on the definition you're giving to that, we are already there. (Or never will be.)

  • 4 years ago · Quote · #36

    Uddipto

    Irontiger wrote:
    cookiemonster161140 wrote:

     We may not quite be there yet but when computers are able to effectively abandon unproductive lines and focus more on that which will likely bear fruit in a chess position then we're doomed.

    Depending on the definition you're giving to that, we are already there. (Or never will be.)

     

    a point to noteCool

  • 4 years ago · Quote · #37

    Uddipto

    Moses2792796 wrote:

    @cookiemonster, this is simply wrong.  There are many positions where Houdini can't find the best moves that are relatively obvious to humans because they are part of common strategical themes, just start setting up various middlegame positions and fiddle around inputing moves to see that it's not that uncommon for Houdini to overlook good moves.  If you spend much time analysing with Houdini you'll find these positions frequently.  Brute force is simply too inefficient to find long term plans, hence why good centaurs will be stonger than engines for the forseeable future.

    To put it another way, the reason why humans, who are significantly weaker than engines in overall play, can still signifcantly assist engines because they derive their strength from completely different abilities.  How many nodes do you think even a super GM calculates over the course of a game?  I don't know but I'd guess it wouldn't be much, definitely not more than a few-thousand.  How then is it possible that they are still not significantly weaker than strong engines, despite the fact that they calculate billions of nodes for each move?

    agree with you.

  • 4 years ago · Quote · #38

    xxvalakixx

    The computer is right here. You should have played Qxc7+ then Bxf8.
    Then you don't lose any material. Instead, you gave two pieces for a rook.

  • 4 years ago · Quote · #39

    Uddipto

    yes yes . i got it


Back to Top

Post your reply: