Do engines lack something?

Isn't it fairly obvious they lack depth? If computers could calculate 300 ply deep in a reasonable time scale (which I believe they never will), they wouldn't need fluffy qualities like true positional understanding or long range planning.

In a sense do chess engines have a mind of their own?  How does an engine outperform a human in a human programmed thing.  I can understand how computers are FASTER at a huge number of things than humans but how is it that a computer can out calculate past its programmed limits, of which a human had to set.

Computers arent my thing so forgive me for any generalizations or whatever but it seems crazy to me.

Something I read or heard somewhere was; if you can't explain it in the simplist of terms, then you do not fully understand it yourself.

So can someone explain how computers are so good. 

Thanks in advance.

LOL! #tigerprowl5

Engines need the syzygy six man end game table bases to form clever plans. This literally involves every legal position with different combinations of six men and requires several terabyte external drives. Much more space than the former nalimov four man table bases. They also need opening books optimized for the variables in their code or style of play. Those who write the books cover novelties and anomalies intended to gain advantages over other programs. It's generally accepted that human databases shouldn't go into the books. The books are actually worth buying and are updated every year for competition. It was probably some of these main line novelties that Fabiano Caruana memorized because the eventual advantage would be very hard to detect even for a GM. Also different programs are better at different parts of the game. The tree beard based engine is a genius at the end game while rybka is great in the opening stage and others like stockfish are better with the middle game. It's interesting to point out that popular programs for sale have actually stolen their code from open source programs. rybka you knew of but houdini has stolen its code from robolito and ivanhoe. They make combinations of the variables and some engine variants stole code from developers who stole code and so on. The hardest opponent would probably be to switch engines at different stages of the game and have different opening books and the full table bases.

Chess software all have basically the same goal to win a game and I think that the gain of a single pawn is what is programmed first. After all a pawn gain with the elite is often enough to win a game. The pawn win represents an advantge of about 1/2 point in the program. So do not be surprised that software prioritize pawn gain over positional formation.

The only comments I remember for computer end games by top chess players was from Natalia Pogonina saying that they are not reliable in many instances.I read this on her site where she talked about tournament preparation.

In any event the best book as a starter is Silman's complete Endgame Course from beginner to master, published in 2007. I have it and often refer to it. Well organized, over 500 pages and at a cost of about $25 it's a super deal, it covers all situations with tests and solutions at the end of each of the 9 chapters.

MrKornkid,

I'm not sure if this will be useful to you or not but I will try.

Computers do NOT have a mind at all.  Computers actually do NOT calculate past their programmed limits at all.

Computers are, in the most literal sense, machines.  I don't just mean because you plug it in and turn it on it's a machine.  I mean at the level of billions of tiny transistors on circuit boards, given positive and negative charges according to a plan created by human engineers and human programmers, and producing a result which has meaning to humans (but no meaning at all to the computer that produced the result).

Here is a video of a toy mechanical computer which can count, add, subtract, multiply, and divide just by having balls roll around on a table.

http://www.youtube.com/watch?v=fLuvopVjAWg

Or, if that link doesn't work, just search for "DigiComp II" on Youtube.

It may help you understand because, to a shocking degree, this is exactly what "real" computers are doing except instead of balls rolling through wooden gates, they use electrons and transistor gates.  In fact, I'd say this thing actually is a "real" computer, it's just not electronic.

So, I don't think you'd say this machine knows how to add, subtract, multiply, divide, at all.  It knows nothing.  It does what it's set up to do. However, it might possibly get the right answer to a math problem that a human gets wrong.  Computers playing chess are the same.

Does that help at all?

Wow, that's a really good analogy.

Someone said that engines are playing like 1800 rated player given 1000000 years for each move

MrKornKid wrote:

How does an engine outperform a human in a human programmed thing. I can understand how computers are FASTER at a huge number of things than humans but how is it that a computer can out calculate past its programmed limits, of which a human had to set.

____________________

Let me ask you, can humans build a machine that can travel faster than the fastest human? Of course they can. A car is faster than a human, just like a computer plays chess better than a human.