How good do chess engines play chess?

That situation happened with a specific robot. They designed it with 6 legs and it was supposed to walk sort of like a spider. However, instead of making the motors move a specific way, they made the motors move any direction and let the computer control how to move.

Instead of moving like a spider, it simply took 4 of its legs and stuck them straight in the air, and then crawled like a centipede with the remaining two.

The movement was much faster and easier than the one the designers had originally intended.

xman720,

Relatively speaking, you have a point. But in reality we all know that the ability to parse and analyze every possible branch within the constraints of a predetermined set of variables has nothing to do with creativity. It's logic. Logic is linear, thus predictable. Provided that there was enough time and effort, a human could do the same; but no one is crazy enough to do that. Since we won't do it, we build tools instead. There's nothing creative about a tool doing what it was designed to do.

Creativity is not just about thinking of something that no one thought before, it's also about the way in which one thinks. I'm sure you're aware that machines don't actually think...

The above examples of machine learning have nothing to do with creativity or "thinking outside the box". Those machines were doing exactly what they were designed to do.

...and so do humans do exactly what they were "designed" to do...what's your point?

There is a very poor understanding how computers work and what they are capable of in this thread, and even among engine "developers", many of whom just copy someone else's source code and then tweak and fine tune.  Engines still rely on valuations derived from human construction and experimentation, and they rely on opening books seeded by human play.  That is the only reason humans are still able to even understand engines games with great effort...engines are being constrained and still somewhat "leashed" to human play.

Someday, someone is going to start constructing an opening book purely based on brute force engine calculations 50 ply deep for every reasonable opening variation...and then. more importantly, somebody is going to eventually realize that engines could be fed pre-parsed results of thousands and millions of engine and super-GM games that allow the engine to derive their own free-floating valuations for material and positions without human input, including the developers and/or their current titled chess consultants, and those valuations will evolve over the course of time, even for something as short as a match.

This bootstrapping approach will produce the same effect as a runaway AI that becomes self-aware in some ridiculous Hollywood movie...the engines will take off, feeding off each other's valuations and play, and humans will be left far behind with the anchor of their own chess history and basic human timidity and caution.

That is simply wrong.  

All the top programs employ some form of "pruning" ---->> Pruning

Engines that show they are calculating at 20 or 30 ply, are not evaluating 100% of the moves for each ply, they are eliminating 99%+ of the lines they come across within a few ply and only following the very best candidates out to the end.

This particular aspect is not that different from human processing; the computer makes "assumptions" about a position and the follow up of that line and discards (sets aside and stops examining, to be more precise) the entire line it if it looks bad enough early on in the calculations.  That's somewhat oversimplified, as an engine can have rules about re-examining a discarded line based on what other analysis turns up (long transpositions, etc.)...

@nartreb: who polices the police? You are like 10 years old? Otherwise you might have heard of IA (Internal Affairs). Just some more mumbo-jumbo in this "generous" thread.

The same as yours, actually.

Let's replace computer-biased wishfull thinking with actual expertise/knowledge, for a change. I'm no authority on the subject, but it's easy to see who knows what they're talking about and who doesn't.

Yes, it is easy.  The retired in his 40s guy with a full career as a software development director/manager (who taught himself programming at 13) is the one who knows what he's talking about.  You, on other hand, just admitted that you don't know what you are talking about.  

Ummm...thanks?  People don't usually undermine their own arguments, I'm just not that used to it ;)...

I wasn't arguing anything. I was actually agreeing with you. I could tell that you have some knowledge on the subject.

You're welcome.

Also not being an authority on the subject doesn't necessarily mean one doesn't know what one is talking about. I've had a limited experience with programming and currently nurture an interest in AI. I keep myself informed on the big picture, just a curious guy.

So far the openings created by humans are not that bad. Chess engines have more trouble evaluating openings than humans. Making use of opening books improves their play significantly, just as their perfect memorization of previous games played.

Opening books also improves our play significantly.

I consider humans cheaters because we always play with our opening books.

Engines are bad at openings and Endings. But Modern Engines are developed  and are being developed.

xman70

Computers very often have opening books programmed into them. So does that make theem "cheaters"?? 

Humans are way, way, way worse than computers at openings and endings.

Think about it: If we are so good at openings, why do we spend hours and hours studying just one move of an opening, spend years on basic opening theory, and read entire books dedicated to just single variations of single openings?

Computers can figure out all that stuff at least decently on the spot.

If all of your chess knowledge was wiped, you would play 1: g4 and then 2: Nh3.

Case closed, compueters are far better than humans at openings.

Chess engines play chess better than you speak English.

Mathematicians rely nowadays for a lot of proofs on computer programs, see

https://en.wikipedia.org/wiki/Computer-assisted_proof.

The majority of the calculators on the other hand are not good in math.

the computer engine are so good.. it whoopped every single human on earth.. i hope this ans ur question

@guardianx9: no, it does not, but thanks for the try.

If you want to improve yourself, then is it good to understand how players with a better skill think. What they see, how they interpret a position.

Furthermore is every aspect of a chess engines created by humans. And with respect to chess games, combinations, algorithms: they are all created by chess players that are now totally overrun by the programs that they helped creating.

The chess engines, that can't think like we can, win everytime from the best human chess players on earth. They show us that when we use our own algorithms better than we are currently doing, that we can improve ourselves a lot.

That is why the question is so important. How can we improve the use of our own algorithms? You might answer that chess engines have more brute force. That is correct, but the brain of a human functions differently. It is in some respect much more efficient, because it understands not only the algorithm, but is also capable of understanding the position by her/himself. A chess engine lacks that last possibility, which is why they need the brute force.

Show me a human that can beat an engine.. ... Yep that is Right they cant.. Game over