Best chess engine at the moment

What's the best, and by this I mean the strongest, chess engine at the present moment? 

latest version of houdini or rybka. either one will meet any humans needs, for the next 20 years or so. seriouslys, they should just stop with that already, don't really need stronger engines at this point. it unecessarys. they are like 3500 elo or somethings! strongest human was 2800 somethings! I don't see the point in making them strongers really. really splitting hairs between the two, and the only times it makes a difference is if houdini and rybka are playing against each other on 20 cores or somethings, other than this either one will be mores than adequate for any humans analysis needs.

You'd want to have something that possibly doesn't give you the absolute best answer possibly accessible? I don't care how advanced the computer might be, if it's going to possibly send an inferior response relative to anohter more advanced one, then I don't want it. I would want the computer that provides me the absolute best answer possible, regardless of my personal ability. 

I certainly understand what youre saying, but do you really need to spend another couple hundred dollars every year to get a chess engine that has increased a few points?  And as omertatao said...chess engines are already 700 points stoonger than the hightest rated player in the world. 

Counter intuitively this can be the case with a very strong engine when it suggests the strongest line but requires 10 "only moves" to follow it up vs an objectively inferior but for practical purposes superior move, which gives a small advantage but offers no chance of counter play for the opponent.

That a tool is only as useful as the person's proficiency in using it is obvious, but thought I'd point it out.

Two things:

1. There is no perfect chess playing engine. No engine will always suggest the best moves

2. I'm pretty sure that you can't effectively compare engine and human ELO's because they are in completely different rating pools. A 2800 computer is not the same as a 2800 human (I think).

Best chess engine among other chess engines is Houdini 2.0 Pro (not free). Here is the website with more info http://www.cruxis.com/chess/houdini.htm   But I think a free Houdini 1.5a should suit you, since it's still stronger than it's competitors. 

I'm going to get this. Thanks.

Chess is, essentially, one long drawn out logic tree, so I would rather have the correct answer than an answer that is temporarily the superior move. Do you know what I mean?

Chess is, essentially, one long drawn out logic tree, so I would rather have the correct answer than an answer that is temporarily the superior move. Do you know what I mean?

 Different engines will have different weaknesses. You won't know when one is spitting out a better move than another, unless you are a really high level player. Engine A may be better than Engine B, but there are probably times where Engine B will spit out a better move. No engine will give you the correct answer 100% of the time. No engine will give you a better answer than all other engines 100% of the time.

What I mean is practice is different from theory.  Theoretically, sacrificing you knight to force a mate in 12 is the best move.  In practice though it's better to immediately win a rook for nothing and go on to win later.

In the sacrificial line you risk a misstep and may lose.  Winning a rook is objectively a lesser move but gives you nearly 100% chance of winning.

This is an obvoius example, in the course of computer analysis there are more frequent and subtle cases.

If you're just going to use it to find blunders though, you don't need anything past Fritz 5.32

In any case, it comes down to a tool only being as effective as the person using it.

You are absolutely right, and it is a fundamental flaw in the algorithms used by current engines that they don't recognize this. When they are presented with a position that leads to forced loss of a Rook in 20 ply, but not in 18, they are liable to sac that Rook on ply 1, where even an idiot would of course take it.

For 6 or 7 men on the board we already have 'perfect play' through tablebases. This is all good and fine when you present an engine playing by the tablebase with a won or lost position. (It will go onto win in the shortest number of moves, or put up the defence that holds out as long as possible). But when yu present it with a drawn position, it is a disaster. The 'perfect play' doesn't know how to make an effort to win, because it assumes perfect play from the opponent as well. So in KBPPKB, with unlike B, it makes no effort at all to even protect its Pawns or Bishop when you attack them, because it knows KKB is still a draw...

As Chess likely starts from a drawn position, an engine with perfect knowledge will not necessarily be a strong engine at all. Of course any mistake against it will be immediately fatal, but it might not know how to apply pressure to force errors. Without additional (but arbitrary) info to classify drawn positions as good or bad, it is likely to allow itself to be pushed towards the draw/loss boundary without offering any resistance at all, only started to fight back when it gets in danger of being pushed over that boundary. And that could be in positions that are so far removed from the draw/win boundary, that any reasonable player would not make mistakes big enough to allow the engine to jump into the win sector in one go. He would make mistakes, (if you can call them that) for sure, but they would just throw him somewhat farther back into the draw sector, after which the engine would lose interest, and allow him to sneak back towards the draw/loss boundary.

The problem is that game-theoreticall all drawn positions are indeed equal. They can only be classified as good or bad based on knowledge of the opponent's fallibility. E.g. trading Queens might be a very good strategy against a random mover, because only in the unlikely case he picks the recapture he would maintain the draw state. But against non-idiots it would only make the draw more likely.

Nothing is 100% in the world; and besides with chess believe me or not, the combinations are infinite (at least big enough for reaching even near; >10^123 if only consider 40 moves per game, more than that makes it much worse!) don't take it too serious! Houdini will do the job for our race.

both rybka and houdini would give you the same answer 99.9% of the time, if they differ is because their might be some programming in them that makes them prefer to win in different ways. there really isn't much difference bwteen them. what you are saying is kind of ridiculous. are yous contending for world champion title soon or somethings? those are the only people that might, and I stress might, care about these things, and that's only to make sure they are as prepared as possible for any home-preparation their opponent has done with their supercomputer. do you have a supercomputer? even if you had the strongest engine, why would this ever matter? your opponent is not going to play the response the strongest engine would. are you going to memorize thousands upon thousands of lines? impossible. are you planning to look at the computer everytime you have to make move? illegal. I just don't understand what yous is getting at here or how what you is saying is practical in anyway for a club player. are you actually a very high rated professional, top 100 in the world, or something? I don't understands.

it has to mean something when they say an engine is of certain strength. it is probably correct that an engine rating and a human rating are different in some way. but humans and engines play against each other all the time, computers don't only play other computers and humans only play other humans. all I know is the strongest engine around today, latest version of houdini or rybka, would destroy any human around most of the time. it is beyond the point where humans, even the strongest grandmasters can win a match against the strongest engine. this being the case, I don't really understand why they are always trying to make them stronger and stronger. at a certain point it ceases to be relevant in terms of a human playing a human in a standard time control game. a human being cannot memorize a million different lines. in my view engines should be used to check the validity and soundness of different lines and find new ideas, and the latest version of houdini or rybka would be more than adequate for this purpose. since you are going to be playing a human being, not houdini or rybka, having the strongest possible engine ever and this ending up giving you some kind of advantage in a standard tournament game is kind of slim I would think for a variety of reasons.

Houdini 2.0c x64