Doubt On the Use of Computers

I've started to become really distrustful of engines. To the point where I'd rather have another human check my analysis than the iron beasts. There is a few reasons for this but one of them an increasing cognitive dissonance with the way computers think. I used to be really in sync with them, now I almost feel like I can tell when they are wrong and right, and its never when I think I am wrong or right.

Stockfish is nearly always right in non-endgame situations, but as with any computer, what is right for a computer is often not right for humans.

I used to try and use Stockfish lines exclusively, and in general, it's not a bad idea.

However, a most common computer-type move that really should be considered an 'error' by humans is whent he CPUs go for unnecessary tactical complications. It's not unusual to get a winning position against a human opponent where you might be one, even two pieces (or pawns) up and thus have an easy risk-free win just by either trading down calmly or restricting enemy counterplay patiently.

However, the computer will ALWAYS go for the fastest tactical kill if one exists. Meaning, even if you're up two whole pieces and could win totally risk free by taking more moves, if the tactic is there, the computer will recommend you sac your Q, R and R, to get that one super complicated winning line in which one false move means you go from totally won to totally lost, because the CPU never errs in these tactical situations. 

It took me awhile as a beginning player to learn that you can't trust the computer for best play in a lot of positions because of this, and you still have to use your judgment as to whether the CPU's line is playable by humans or simply too risky in terms of leaving you totally lost if you make even one slight calculational error.

There are other complicated positions where the computer evaluates improperly. One thing you have to ask is: is the computer not seeing the line 'cause there's something wrong with it? I'm not so sure. I think computers evaluation of positions is sometimes, if not mostly, radically wrong. I don't use computers for a reason, they won't help me prepare. I kind of prepare the way Tal might have. Inspiration. I let the world of chess guide me instead of consulting a silicon crapshoot.

It's funny 'cause I have been playing for five years and am already at the level of say a 1900. I beat 1700s, 1800s, at my club all the time in the last few weeks. I've been learning to balance chess with school and continue improving at a very fast rate. One day I'm going to be world champion class, and I really don't know about you but I'm getting there. But what is the use of computers even then? I am a very strong player and you should understand this before putting me down because of my rating.

Some elite GM recently said that engines don't play the best moves. They just don't make any mistakes. 

If you want to find the best moves, you still need to analyse a position for at least 20 minutes without moving any pieces. Choose some complex positions from games that are well annotated (Khmelnitsky's Chess Exam book or positions from Yusupov's series of books work well for this), and just go to town analysing. 

When 20 minutes are up, decide which move you would play, then write down all of the lines you analysed. 

When you're done, compare what you've done to the published analysis. Go back and find out whether you or the authors made mistakes. If you do this regularly, you will begin to be able to find the best moves.

Computers are just calculators that help to check your math.

You're right. But it doesn't even have to be that long. That's why intuition is so important. If you're keyed in to something greater than the mind then you are suddenly able to make crazy moves without even calculating.

Putting the bong down seems like an unnecessary step in the learning process.

The weirdest thing about me is that I don't believe that there is an objective reality for chess. It almost seems like there are too many possibilities for there to be an "objectively best move". I think there are criteria before which we find our moves. Those criteria seem different for everyone, even though some of them overlap to an ultimate truth.

Aquarius, if you are not willing to put in the work (20 minute exercises), you really will not improve much.

I am improving quite quickly already smyslov. sigh at the rating-oriented noobs here.

I agree. This is more what I was getting at!

I actually think that Stockfish plays objectively superior to any human on the planet. 

Just one example is the "Rybkamura" match where Nakamura + computer in a 10 hour match, was handily beaten by Stockfish. 

http://www.chess.com/news/stockfish-outlasts-nakamura-3634

Any non-GM who says that they can come up with objectively better moves than Stockfish doesn't understand at all how strong Stockfish is, and is grossly overestimating how strong they are relative to it. Back in Kasparov's day, it was true that GMs could outposition computers, but that gap has been long closed. 

I saw a quote not to far back from Roman Dzindichashvili something to degree that he was deeply impressed with Stockfish's ability to play pure positional positions accurately.  

Note as well that just because Stockfish says such line is objectively is the best, does NOT mean it is the best line for humans to play. In fact, it might be a terrible line for humans to play! There are some STockfish lines that it says is +2.0 so up about an exchange with perfect play, but for a human to play accurately enough to not get checkmated with obvious moves from the opponent is very, very difficult. (Some lines in the Sicilian Dragon are like that - white attacks with Q/R/minor piece against a lone K where one wrong move = instamate, but Stockfish says it's objectively better for black. Most GMs would say to not even go down that line in the first place.) 

Sorry, but can the whole objective strength of moves theory, its wrong. 1 is boundless and 0 is infinite, and once humankind figures that out, beating comps will be easy. I think its a lack of confidence against comps since the 1997 deep blue match, which wasn't even a computer by the way. The line that killed Kasparov's confidence in the second game was programmed in by Nick deFirmian and Joel Benjamin and Kasparov had no idea. Computers are not dangerous, and fearing them is why most GMs have so much trouble. 

Some people still like to think they can outthink the computers. Might have something to with them not understanding how computers think. For example if black has one extremely complex line that leads to draw and all the other moves in the position are completely busted, as far as computer is concerned the evalution still stands 0,00.

You are very funny! I like that. But you are also very mean. I don't like that. I am incredibly strong. If you do not believe me check my archive. 

You're the one who mentioned ratings. I didn't. 

Let me be clear, computers are far superior to humans in chess now. But they don't always play the best move. When given an option, they tend to play the most forcing move. Also, if there's a trivial difference between the quality of moves (for example, .01 difference), the engine will play whichever is higher at the time it stops its analysis. If the engine were allowed to run another 20 minutes, it may change that evaluation. 

A human and a computer together would be able to outplay the computer by itself. The reason Nakamura lost  the stockfish match, 3-1, was that he reached equal positions, but wasn't content to draw. He wanted more. It was only a short match and he got burned by his desire to win. 

If the match had been twelve games instead of 4 and if there had been a sizable prize for the winner, Naka probably wouldn't have pushed so hard to win from those positions.

+

Funny stuff right there. Delusional but funny.

Yes you picked my first loss. Why? Why couldn't you have picked a random win instead? Losses show us at our worst, and I was playing 15/10 then which I never play.

Also I'm on a 17 game winning streak in 45/45.

Man, why are there so many trolls on chess.com?