Related to the Fisher-Benko my ignominous Fritz 11 (on his relative ignominous hardware and owner and so on...) see 17. h4 after 3 minutes of rumination.
In a five minutes game it's a fail but in the time condition of a tournament match i think it's not a bad performance.
I can suppose that my Fritz 11 in a tournament match should stand in front of a genius like him and it's seems ridicoulous also to comment
Where is my fault in this argument ?
What is the score?
h4 should return a winning score, so +100cps and more.
oh no..no winning score..equal position but best move..forgot this sorry :-)
Lyodomil, of course engines can be beaten in a handicap game (although this is very challenging for an unassisted human). When I said a PGN, I meant of a game of standard chess from the starting position.
Stockfish 7 likes Kh1 from about 50,000,000 nodes: are you sure Stockfish 8 never likes it?
What position are you talking about?
White move 13 in the queens knight versus c-pawn odds game you posted recently.
I have already posted a number of standard game pgns, just browse this thread, as well as 'The Secret of Chess' and 'Stockfish is blind'.
You have posted no PGN of a standard chess game in this thread. Please do.
He cannot. Such game(s) would fail verification. Especially now that we know that you have a powerful machine and you know how to use the engine.
oh no..no winning score..equal position but best move..forgot this sorry :-)
This is a very deep position and requires long analysis, but I guess white should be winning.
What is your main line?
Lyodomil, of course engines can be beaten in a handicap game (although this is very challenging for an unassisted human). When I said a PGN, I meant of a game of standard chess from the starting position.
Stockfish 7 likes Kh1 from about 50,000,000 nodes: are you sure Stockfish 8 never likes it?
What position are you talking about?
White move 13 in the queens knight versus c-pawn odds game you posted recently.
I have already posted a number of standard game pgns, just browse this thread, as well as 'The Secret of Chess' and 'Stockfish is blind'.
You have posted no PGN of a standard chess game in this thread. Please do.
So, if not Kh1, what?
Everything loses.
This was a very STC game, so you should not analyse it with much time.
You better read the whole thread, but here once more a game vs SF 6, characterised by extreme closedness:
I like this game, because SF did not get a chance, and the human won without taking minimal risks.
I think this is the way to play against top engines.
He cannot. Such game(s) would fail verification. Especially now that we know that you have a powerful machine and you know how to use the engine.
Why don't you self-verify yourself on absurd statements?
As said, all I am waiting for is SF 12 or Komodo 15 to crush them.
Challenges fill me with adrenaline.
Unfortunately, as said, they will delay new releases.
Talkchess forum knows: as soon as they release a new version, I start beating it, as it is new and interesting, so they are delaying a new release as much as possible.
An analysis with PGN-Spy of the above game (1.f4 Nf6...) gives:
UNDECIDED POSITIONS
Positions: 27
T1: 8/25; 32.00% (std error 9.33)
T2: 11/22; 50.00% (std error 10.66)
T3: 13/21; 61.90% (std error 10.60)
LOSING POSITIONS
Positions: 0
Most probably you used Stockfish 7 or 8 to win that game.
Sorry, but I cannot second that statement. The main thing is the sample is way too small. The second is that the statistics are not very extreme (most good players would have games with similar stats, and it might be quite normal for a professional player as an average. To emphasise, averages and individual games are not the same evidence!)
To be honest, I like this game: it's a really nice win. I am not entirely clear how black got suckered, but it did. I too like this sort of closed game with a sacrifice or two (mostly in the king's indian (occasionally reversed) for me).
Besides Stockfish 7 & 8, there are also other engines that can win against Stockfish 6, like Houdini or Komodo. I suggest you check this game using them. I stop doing this, because my computer has to weak processor.
I would not expect big differences: engines tend to agree a lot because of their similar strengths. If anything a mere 62% T3 is quite strong evidence of little or no computer assistance. Think about it: 38% of the time on those moves he picked a move the engine did not rate in the top 3.
An analysis with PGN-Spy of the above game (1.f4 Nf6...) gives:
UNDECIDED POSITIONS
Positions: 27
T1: 8/25; 32.00% (std error 9.33)
T2: 11/22; 50.00% (std error 10.66)
T3: 13/21; 61.90% (std error 10.60)
LOSING POSITIONS
Positions: 0
Most probably you used Stockfish 7 or 8 to win that game.
Can you explain what this is all about.
What kind of Spy is that?
How can I play myself and use SF at the same time, I just don't understand.
Besides Stockfish 7 & 8, there are also other engines that can win against Stockfish 6, like Houdini or Komodo. I suggest you check this game using them. I stop doing this, because my computer has to weak processor.
First and major clue:
SF, Houdini and Komodo NEVER play closed positions, they NEVER play The Stonewall, unless from a book, with either black or white, they NEVER play the King's Indian, with either black or white.
I don't think you should be lloking any further than that.
The obvious hypothesis is I am just stronger than SF.
The simple truth is I like playing engines, and I like beating the top engines, as simple as that.
So, I will play against SF until I win a game, and then 5 and then 10.
As simple as that.
You can not beat the perfect player, but an engine that has so many imperfections, certainly you can.
We know Stockfish is imperfect from its losses to other engines. AlphaZero provided some more evidence for that.
[The "PGN Spy" software is to see if a person's moves have been assisted by computer, but the stats posted do not strongly support that conclusion.]
As I said, my computer is weak. When a stronger computer is used, the percentages are higher, I say this by experience on the cheating forum.
Besides, it depends on what engine was used for play/analysis.
I agree that someone that knows intimately the programming of SF, like LT, can by playing many games, find a path in which SF loses. But this doesn't makes LT the greatest player in the world (or the best GM). Let's see him defeating Komodo 10 or the last version of Houdini.
He has hinted it took a lot of trying to get successes, but his play makes sense, and you can see how Stockfish has horizon problems when there are certain types of sacrificial moves. This could well be due to an evaluation function that is too materialistic blinding it to the value of positional sacrifices.
Lyodomil, of course engines can be beaten in a handicap game (although this is very challenging for an unassisted human). When I said a PGN, I meant of a game of standard chess from the starting position.
Stockfish 7 likes Kh1 from about 50,000,000 nodes: are you sure Stockfish 8 never likes it?
What position are you talking about?
I have already posted a number of standard game pgns, just browse this thread, as well as 'The Secret of Chess' and 'Stockfish is blind'.