An easy-to-use chess engine?

I don't think I'm asking for terribly much... I simply want a program that can look at a game setup for a short amount of time (a few seconds to a few minutes) and then explain what the best move would be in that situation.

Is there any program that is understandable (I find that Arena and Tarrasch are ridiculously complicated and do not justify this with a good tutorial), something that I can have working within minutes?

Thanks.

Houdini or Rybka?

I don't think there are any engines that would explain anything, though it will tell you which move the engine thinks is best based on their algroethem.

If you want explanation, you need a coach.

Yes a coach is good

By explain I mean a justification... the line it "has in mind" would do this perfectly well.

Houdini and Rybka are chess engines, but they lack a GUI, as far as I can tell.

I know of hundreds of engines, and none of these  engines "explain" the reason for the moves it chooses.   They don't work like that. 

As far as programs (GUIs) are concerned, Tarrasch, as an interface, is about as simple and understandable as it gets.  Another simple GUI is Kvetka (http://kvetka.org/en/)

I think the problem is that I don't understand the output of these chess engines... when I ask Arena to analyze a position, I get output like this:

8/22     00:00     105,182542,882     +0.31     c5xd4 e3xd4 d6-d5 f2-f4 d5xc4 Nd2xc4 b6-b5

I'm not sure what to make of it, though. I think the 8/22 is how many plies the computer looked ahead as compared to maximum depth (though this seems bubious, since that "22" would change sometimes (although this might be an indication of some kind of quiescent search)), but the 00:00 and 105,...,882 don't make any sense to me. I think +0.31 is the heuristic value of the position evaluated, and the moves following it is the setup contemplated. Is this correct?

I think I should have allowed myself to calm down before making this thread, but nonetheless, I am a little bothered by the archaic way Arena analyses a position.

I see something similar when I run the Stockfish app on my iphone in analysis mode.  At the bottom of the board, it states what it believes to be the best move for the side to move, then follows that up with what the best sequence of moves would be for each side. This might be what you are looking for.  However, more often than not, going over these lines often does not offer me any closure in terms of why it is superior to other moves unless it results in something obvious like a win of material or a very dominating position.

This is why scarcely use chess engines for anything other than analyzing some very difficult chess puzzles.  

Well, I guess it depends upon what you want to do.

If you are just interested in chess as a game, and you are interested in knowing the best line, you can ignore the

8/22     00:00     105,182542,882

part of it.  That's just the engine information on the number of nodes or plys it's searching, the time it has spent searching, etc., etc.   If you are just interested in chess as a game, you can go straight to what it believes is the best line, which is

c5xd4 e3xd4 d6-d5 f2-f4 d5xc4 Nd2xc4 b6-b5.

But if you are a programmer or you are interested in the chess engines themselves and how quickly one engine might find a key move or a line, compared to how quickly another engine might find it, than all of this other engine output, besides the line/move, them becomes meaningful to you.

That makes sense.

So how do I determine what it thinks the best line is? (I suppose this is the ultimate question :P)

Also, do chess engines evaluate the position for the player moving, or simply for white? This is to say, if it's black's turn to move, should he go for the move with the greatest value, or the least value?

You don't look at that output for an explanation :p.   What you do is play the why not this or that game.  It says Nf3 is best, but you think there will be problems when black pawn storms the kingside, or opens the queenside, or trades it off into an advantageous endgame, or whatever... so you take up the opposing side and by trying to prove the engine wrong with a challenging line, you learn why its suggestion works.  It could be tactically justified (so now you notice certain weaknesses you didn't before) or e.g. your endgame is not advantageous because your king is too far back.

It wont give real word explanations, but it will speak to you in the lines you choose to explore with it.  It also helps to have its top 3 or more options open or to analyse with multiple engines.  So, for example, when you reach a position where the five top moves have an evaluation difference of .1 between #1 and #5 you know that there are no forcing liens, and the feel for the position / idea behind your plan is more important than the next 1 move.

Or at the very least you'll be justified in positions where a certain move just looks so fundamentally correct you wonder why the computer doesn't play it.  Well maybe the eval difference is only .02, but if you're only looking at the 1st choice you won't know your move is equally as good.  I say equally as good because a .02 eval difference means nothing.  As the depth increases AND as the position progresses the eval swings are much greater than such a small number.  Not to mention the program isn't taking into account the ease of play or practical chances of a position.

tl;dr  --  you have to work with the program to milk info out of it.  If you want something more interactive get a membership and hop on chess mentor here.

No, its best line is irrelevant unless you immediatly understand it after seeing it.  Otherwise you have to work with it one move at a time.

The eval is for the position, regardless of which side it's calculating a move for.  So negative numbers mean it thinks black is better by that much, and positive number are white is better by that much.

If your GUI and engine is listing just one PV (Principal Variation) then best line according to that engine is always the line that is given.  In your example above it's the line beginning with c5xd4.

Many engines are able to give more than one line.  Just set your PV to "2" in your GUI, to see the the top two moves/variations.

All engines evaluate the bet line for both sides.  Meaning, the engine will look for the best move regardless of whose turn it is.  To clarify further, if it's White's move it looks for the best move, and then if that move were played it then looks for Black's best move, and then White's best move from the resulting position, and so on and so on.

your kinda on the wrong path,.. computers dont really tell you what the best move is only what their mathmatical comparison thinks it might be hence the evaluation part. 
its fairly meaningless for humans if an evaluation is +3.14 or +2.96 the difference is neglegible so whats important is if YOU can play the position out. Why you hear a lot of comments like "thats a computer move" or "only a computer can play like that" because some positions only computers can hold together the tactics in the lines they offer. This is one of the reasons computer cheaters are so easy to pick out. They play lines the computer offers that vary littl in the long run but avoid some hidden, meaningless, tactical idea 4 moves deep that humans wouldnt even bother with.

For now just use your own brain and then use the computer as a faster way to varify what your brain is telling you. if the evaluation changes more than say 1.00 then be concerned if your move is .5 different than the move offered by the computer who cares....

after games I look at the ideas a computers offer but dismiss most of them as long as mine is viable. i do blunder checks and missed chances and maybe if something is really interesting ill check it out but always keep the monsters in a box in check. 

Here's a position of mine from a recent OTB tournament game:

1. 19 [+0.78]  24.... c4 25.Nh4 cxb3 26.c4 c5 27.Ng6 cxd4 28.Qe4 Rxg6 29.hxg6+ Bxg6 30.Rh7+ Bxh7 31.Qxh7+ Ke8 32.Qh8+ Ke7 33.Qg7+ Ke8 34.Qxf6 Be7

2. 19 [+2.43]  24.... Bxf4 25.dxc5 e5 26.Rdf1 Bf5 27.Nd4 Bh7 28.Qf3 Rb4 29.Ne2 Bf5 30.Nxf4 Rxf4 31.Qe3 Qd5 32.h6 Bh7 33.Rhg1 Rg6 34.Rxf4 exf4 35.Qf2 Rxg1+

I used "Scid vs. PC" in this example, to determine this best line.


The following clip is cut and pasted from WinBoard.  As you can see, the analysis output is slightly different. 

17    +0.69    254.5M    0:29.05   
c4 Nh4 cxb3 c4 c5 Ng6 cxd4 Qe4 Bc5 Rhg1 Bxg6 hxg6+ Kf8 Rde1 Re8 Qh1 Ke7 f5 e5 Qh7+ Kd8 Qxd7+ Kxd7 Rh1 Kd6 Rh7 Rc8

17    +2.33    349.3M    0:38.16   
Bxf4 dxc5 Qd5 Rhf1 Bf5 h6 Rg3 Nd2 Bg4 Qh2 e5 Ne4 Bxd1 Rxd1 Rg5 Nxg5+ fxg5 Qh5+ Kf6 Re1 Rh8 Qh3 Kg6 Rh1 Kf6

(The engine was thinking for different amounts of time when I clipped the analysis, so comparing the lines is meaningless.)

I failed to mention that the program I used (which is, btw, Arena as a GUI and SOS 5.1 (I think) as an engine (though the word "engine" is utterly meaningless)) did output more than just that one line... actually it outputted a myriad of them, about 50 in all.

For the record: I do understand how chess programs work. I've made them before. It's just that I've never had use someone else's, and the programs don't make any sense to me.

I don't understand why the program wouldn't have a best move(s). From a computational standpoint, the entire point of going through moves is to establish an estimation of how good a position is using minimax and eventually a heuristic function. An AI would then choose the position that yields the best estimation of how good a position is, so why can't I simply ask for a "what would you say are the five best moves here"?

Well sure, it's most recently calculated line would contain, by it's evaluation, what the next 5 moves should be.

I just don't see why that would be useful unless you go "aha! this vairation is to the point of the position, and I understand it completely" 

More likely you'll see something akin to it moving its rook to d1 on move 1, only to move it back from where it came on move 4 and then on move 5 sac a pawn... see what I mean :)

I'm not arguing that I'll be so enlightened as to see why it does something (because it itself doesn't see it either-- only sapience can allow us to take altitude from our problems), but I'd like to know what it recommends, and what lines it estimates will come out of those decisions.

Maybe I should copy-and-paste the output and someone can just point out what the program recommends?

Here are the five "best" moves, from this position:

Analysis by Critter 1.6a 64-bit:

1. +/-  (1.11): 12.Nd2 e5 13.0-0 0-0 14.Nc4 Bb7 15.Re1 Rad8 16.dxe5 dxe5 17.Nxe5 Rfe8 18.Nc4 Qxe1+ 19.Qxe1 Rxe1+ 20.Rxe1 Nc5 21.Bf1 Ba6 22.Be5 Ng4 23.Bc7 Rd7 24.Bg3 Nd3 25.Bxd3 Rxd3 26.Nd6 Kg7 27.Nf5+ Kg6

2. +/-  (0.92): 12.Qa4 0-0 13.Qc6 e5 14.0-0-0 Rb8 15.Rhe1 Bb7 16.Qa4 Rfd8 17.dxe5 Nc5 18.Qc2 Nxd3+ 19.Qxd3 dxe5 20.Rxe5 Rxd3 21.Rxe7 Rxd1+ 22.Kxd1 Bxf3+ 23.gxf3 Rd8+ 24.Ke2 a6 25.c4 Kg7 26.Be5 Kg6 27.h3 Ne8 28.Rb7 Kf5 29.Bc3

3. +/-  (0.80): 12.h4 g4 13.Nd2 Bb7 14.Nc4 Bxg2 15.Nxd6+ Kf8 16.Nc4 Ne8 17.Rg1 Bf3 18.Be2 Bd5 19.Qc2 Ndf6 20.0-0-0 Rc8 21.Qa4 Bxc4 22.Bxc4 Nd6 23.Be2 Rg8 24.Kb1 Nfe4

4. +/=  (0.63): 12.0-0 Bb7 13.Nd2 Nd5 14.Nc4 Nf4 15.Bxf4 gxf4 16.Qg4 d5 17.Qg7 Rf8 18.Ne5 Qf6 19.Qxf6 Nxf6 20.a4 Rg8 21.a5 Ng4 22.Bb5+ Ke7 23.a6 Bc8 24.Nc6+ Kd6 25.h3 Bd7 26.Nxa7 Rxa7 27.Bxd7 Kxd7 28.hxg4 Rxg4

5. +/=  (0.55): 12.a4 Bb7 13.0-0 Nh5 14.a5 Nxg3 15.fxg3 0-0 16.Nd2 Kg7 17.Qe2 Nf6 18.g4 Kg8 19.h3 Nd5 20.g3 f6 21.Bc2 Nc7 22.c4 d5 23.axb6 axb6

As you can see, Critter thinks Nd2, its first choice, is much better than a4, its 5th choice.  (+1.11 vs + .55)

This output came from Fritz 12, when I changed the PV to 5.