what can programs see that 2700 GM cannot

[Vinvin]

About humans, a top-50 player make usually one blunder (around 1 pawn) each game.

[rabbits23]

Robert : I think any master who would take the rook sacrifice on offer by white would have to be in time trouble or a very ordinary player. The very
fact that a major piece was being offered in such a simple fashion would
alert most players to be be very, very cautious.Yes it might have been a major blunder on the part of white nevertheless the position is one which
separates the clever from the not so clever.
Allan

[Uri Blass]

check out 23.Qf3, a 2600 move. another blunder from chessbase.
now show me a one move program blunder, I have shown you two but I have a thousand more

By your definition in every game that a program lose there is a blunder assuming the opening position is a draw(if you use the perfect player to analyze).
One move change the real score from draw to mate against the losing program.

If you use programs of the future they will show you the blunder and the evaluation may be changed from draw to mate in 30 against the program in one move.

I think that definition of a blunder by the number of pawns you lose based on the engines of today on the hardware of today is clearly wrong.

For me a blunder is a move when it is easy to understand why it is wrong.
A move that lose even if it lose the queen for nothing based on engines is not a blunder if humans cannot understand fast why it lose the queen.

Uri

[Uri Blass]

I usually check GM tournament games posted at Chessbase.
I have been doing this for years.
I try to understand, using program analysis, why the GM lost the game.
I have found that almost every loss is because the GM does not understand a position and blunders.
Then I run the misunderstood position in the program and see if the program understands.
I have found that almost every time a 2700 GM does not understand the position, the program does.
There are thousands of these examples available.
What I find interesting is that in 2015 the Super GM still continually misunderstands positions, very often.

here is an example:
some guy on the internet called this a Kasparov immortal game when actually his opponent made a 4 point blunder, 24...cxd4.
I would not call this an immortal game.
I would call this a mickey mouse game, Kasparovs opponent didn't have a clue so Kasparov gets praise, stupid logic.
/quote]

Do you understand why it is a blunder?
In other words can you explain to other people why cxd4 is wrong(engines say is not an explanation and you need to give specific lines that you calculate) and how much time you need for it.

If you need some minutes to convince other people that cxd4 is wrong then I do not consider it as a blunder.

Blunder is something that you can explain to a person who does not know the position why it is wrong in a few seconds from my point of view.

Uri

[jdart]

This is nothing new. Even Gms get tired or distracted and occasionally suffer from "chess blindness." They can sometimes be incredibly accurate and precise, but they don't do it 100% of the time. And this is why computers are now very hard even for GMs to win against. The computer's level of play tends to be much more consistent and it will not overlook an opponent's simple blunder. This is not to say they play flawlessly; they don't. But to make up for this they almost never miss things that can be found with shallow to medium depth search.

--Jon

[Uri Blass]

Robert : I think any master who would take the rook sacrifice on offer by white would have to be in time trouble or a very ordinary player. The very
fact that a major piece was being offered in such a simple fashion would
alert most players to be be very, very cautious.Yes it might have been a major blunder on the part of white nevertheless the position is one which
separates the clever from the not so clever.
Allan

I do not agree with you.
The mistake that topalov did by capturing the rook is a mistake that many players can do without knowing the position earlier.

Can you show me a convincing evidence that capturing the rook is losing(not by engine score after a long time but by specific lines)?

There are many possible lines and not only the game line.

Edit:I doubt if a human can be sure mathematically that capturing the rook is losing.

A strong human player may suspect it without being sure about it and avoid the mistake but it is not something that I call a blunder.

If you see some attack of the opponent that you cannot calculate to the end and do not evaluate it correctly then it is not a blunder by my definition.

[APassionForCriminalJustic]

This is nothing new. Even Gms get tired or distracted and occasionally suffer from "chess blindness." They can sometimes be incredibly accurate and precise, but they don't do it 100% of the time. And this is why computers are now very hard even for GMs to win against. The computer's level of play tends to be much more consistent and it will not overlook an opponent's simple blunder. This is not to say they play flawlessly; they don't. But to make up for this they almost never miss things that can be found with shallow to medium depth search.

--Jon

It is not just the avoidance of say obvious blunders (or consistency) that make computers play chess at astronomical levels. It is how deep they can see things. They are positionally now very strong, and they can win games that look to master players to be draws (I have countless games that can show this - one little tiny inferior move and the game could be lost literally). It is unbelievable what can engines can see; the ideas and variations that they can come up with is absolutely disgusting - and it really does put any human player to shame. Of course I am speaking about Stockfish, Komodo, and Houdini mostly. But of course countless engines can do this relatively easily.

You are not going to beat an entity that can see positions, even those extremely complex, with such massive depth, and precision. Heck, even drawing today's best chess engines at classic LTC time control(s) is very difficult now. You will see no master player even waste his or her time anymore with Man versus Machine events. It is both pointless and FUTILE.

Many people always seem to focus on avoidance of so-called easy blunders as the bulk of an engine's strength. There is simply more to it then that, beginning with what they see in depth and ideas, and winning games that really would never be seen by super GMs. 2700+ rated players are great players, but they are not even close to the level of today's best chess engines. In fact, I think that I read somewhere actually quite recently that an engine like Komodo 8 would be approximately 3550 rated in FIDE tournament play according to experts. It would be a slaughter fest that no mere mortal could effectively contend with.

[Stan Arts]

Ofcourse. As with any skill where machines or technology overtakes human capability.
Speed, strength, precision etc. That does not mean particular skill in humans is any less impressive or enjoyable from a sport perspective.

Part of why I enjoy human chess is the pressure, psychology and subsequent mistakes. Just how impressive it is what superGM's do see instead of what/when they don't.

You seem to enjoy perfect/scientific chess. To each their own but why focus on human chess? Computers have gotten better than humans and that's useful for checking ideas but standing on the sideline criticising doesn't make sense to me.
Same as 100 meter dash world record standing on the sideline going, whoa I know of a rocket sled that's atleast 200x faster! n00bs!

Computerchess is interesting to me from a programming perspective, from a chess perspective not very much. Criticising humans seems to me poor taste in this regard, but to each their own.

[Dann Corbit]

check out 23.Qf3, a 2600 move. another blunder from chessbase.
now show me a one move program blunder, I have shown you two but I have a thousand more

By your definition in every game that a program lose there is a blunder assuming the opening position is a draw(if you use the perfect player to analyze).
One move change the real score from draw to mate against the losing program.

If you use programs of the future they will show you the blunder and the evaluation may be changed from draw to mate in 30 against the program in one move.

I think that definition of a blunder by the number of pawns you lose based on the engines of today on the hardware of today is clearly wrong.

For me a blunder is a move when it is easy to understand why it is wrong.
A move that lose even if it lose the queen for nothing based on engines is not a blunder if humans cannot understand fast why it lose the queen.

Uri

I like your definition, but I would add to it.
A move is a blunder if one computer sees instantly that it loses but the other computer does not see it.

[JuLieN]

Here, engines see a mate in one. Kramnik doesn't.