Comb through the very best games of the very best players of all time and you will come to a different conclusion, if you really understand chess.
I only check 2600 to 2800 rated players.
prefer tournaments played in last ten years.
blunders have not decreased with modern study methods.
I only check very best games of very best players.
Terry name two of the very best players that you refer to and I will post the blunders.
You might be surprised by what the computers call blunders when in fact they are the winning moves in some cases!
I would suggest going through Kasparov's best games back in the eighties!
I am not sure what this proves, because you will find blunders in almost every engine game where one engine loses to the other.
Dan Corbit
Oh really?
I check most long time control games by programs and I don't find any.
How many games were in the TCEC tournament?
Show me 1 game that had a 4 point blunder single move.
ok a 3 point
ok a 2 point
show me some of these program blunders by the top programs using strong hardware, Dan Corbit. I want to see them.
kgburcham
So far, I have 140K of positions with a full piece blunder by one side or the other. I guess it will be 3-4 times that many when the script finishes tomorrow morning (all high level opponents from CCRL at 40/40).
The 2 pawn blunder file is twice as large as that.
The 1 pawn blunder file is twice as large as the two pawn blunder file.
I am using Thomas McBurney's utility "Game Analyzer" to produce the list.
I am not sure what this proves, because you will find blunders in almost every engine game where one engine loses to the other.
Dan Corbit
Oh really?
I check most long time control games by programs and I don't find any.
How many games were in the TCEC tournament?
Show me 1 game that had a 4 point blunder single move.
ok a 3 point
ok a 2 point
show me some of these program blunders by the top programs using strong hardware, Dan Corbit. I want to see them.
kgburcham
Finally someone with a brain. All of this stupid crap about engines making pitiful blunders is like saying fish can talk. The point here Dan is that engines are MASSIVELY superior to our best human players. We shouldn't care about the naturally small inferior moves that engines HAVE to make in order to lose. Engines are at a level of incredible skill and ability - my friend who is a FM can speak volumes as to the strength of engines like Stockfish. In fact, Stockfish on my 20 core system plays at a level that he has NEVER seen before. And he has a lot of experience with engines both old and new.
So with a top program, and with very good hardware you essentially have the chess world right in your finger tips. You can train, analyze, and see ideas, variations that GMs might never even think of. That is stunning.
Engines do not make pitiful blunders (or make them very rarely). I do not know of anyone who has ever said that. They will, on occasion, make a blunder the size of a queen. You or I might not ever see that blunder but their opponent engine will (it may be 15 plies away or more where the issue becomes forced).
Super GMs make pitiful blunders on very rare occasions (I believe that there is even one instance of a GM getting nailed by a "Fool's Mate"). If you or I play 100 games against a super GM, I doubt if we will ever detect any blunders.
Super GMs will sometimes understand a position better than an engine does.
Chess engines are better tactically than the best super GMs. I don't think anyone sensible will argue that today.
Is it possible to redo this game with the original engines to proof it's not a book mistake ?
It is almost certainly a book mistake.
But those are the moves played in the game.
Chess programs make mistakes due to the following:
1. Bad book lines
2. Hash collision {super-rare}
3. Programming error
4. Out-thought by the opponent
5. etc.
Similarly, humans make chess mistakes for various reasons:
1. Distracted
2. Ill
3. Not taking the game seriously
4. Out-thought by the opponent
5. etc.
I am not sure what this proves, because you will find blunders in almost every engine game where one engine loses to the other.
Dan Corbit
Oh really?
I check most long time control games by programs and I don't find any.
How many games were in the TCEC tournament?
Show me 1 game that had a 4 point blunder single move.
ok a 3 point
ok a 2 point
show me some of these program blunders by the top programs using strong hardware, Dan Corbit. I want to see them.
kgburcham
A more philosophical view :
It depends how deep one search : a "2 points" mistake becomes a "3 points" mistake when you search deeper. And all "mistakes" are "mate mistakes" if one search deep enough.
Is it possible to redo this game with the original engines to proof it's not a book mistake ?
It is almost certainly a book mistake.
But those are the moves played in the game.
Chess programs make mistakes due to the following:
1. Bad book lines
My point is : in the case of a bad book line, it's not the engine who make the mistake.
The computer chess system made the mistake. When the game in question was played, opening books were really needed to make a program play good chess. Less so today.
If a chess program has no book and makes a bad move that is another kind of mistake.
It used to be easy to fool them with things like Stonewall, but it is a lot harder today.
Even at that, if you look at Sedat's contests you can download the files and see book lines that cause engines to fail.
