Humans v Houdini chess engine (Elo 3300)

You're saying that computers aren't good at board games? Have you forgotten checkers???

Oh, the great master speaks!

Here's a nice challenge for you guys. Can you beat Houdini in this setup?

I can't, thats for sure..

Or try this one: very tough to win with white as well...

Let's make it happen!

Let's make it happen!

Let's make it happen!

well, show me proof then :) A 2000+ rated person will indeed fix it- but Id like to see it though. And if you win try the one in the post before that.

Let's make it happen!

:D ok, thats nice! But Im not sure how I can set the game up here on chess.com. When I start a game I can select hundreds of positions, but this one with the king in front of pawns is porbably not one of them. Is there another way?

And which one would you like to play? King on d4 or e5?

Kd4: 1. e4 Kc2
Ke5: 1. e4 e6 

What a fascinating thread.  I have read all 353 posts.  I find it amazing that we don't have an answer to the original question after several years of "advanced computers and software".  Some have pointed out that the Super GMs don't have any motivation to play CC games against Houdini.  I would have thought some GM would have been willing to put the question to rest and prove they can beat a modern computer and Houdini.  Even if uncompensated it would be great bragging rights for the GM.  I suspect that the GMs know they would not be able to beat the machine though.

If compensation if what is preventing us from discovering the answer to this burning question, then let's fix that.  I personally would donate $10, even $20 to satisfy my curiosity.  I think I read that chess.com has 1.6 million members.  Some others would also be willing to donate I am sure.  We could also extend the plea for donations outside chess.com.  We could create a cash prize to attract a GM to play CC chess with Houdini.  We might have to require a lie detector test after the game to satisfy (maybe) the doubters that the GM didn't cheat if they happen to win.

This question technically is easy to answer, but still so very elusive.  Tons of opinions and no concrete evidence. 

There is no way of proving that the GM did not use an engine to assist. Even if the GM did not,ones would still doubt it was genuine.

Hello! People!

I'm new here and this is my first post. This has been a fascinating thread to follow and I would especially like to thank "FirebrandX" for your very insightful contributions. I used to play (2012 CFC) in the early nineties but life and other such things made me move on.

Sorry if this is OT, I'm not even really sure what the original topic was anymore...

What I have observed (I'm sure many others too) is that chess has traditionally been about beating the other "person". There exists in chess a long history of "preparing for a specific opponent", "anti-computer strategy", and psychological aspects of gaining the upper hand. As opposed to search for truth (best possible play). It seems pretty clear to me that "centaur" chess (as described by FirebrandX at least) is about finding best play. This is evident in a couple of Judith Polgars quotes:

"Chess is 30 to 40 percent psychology"

"I can't confuse the computer"

I used to truly enjoy studying beautiful winning combinations. But now I kind of think these combinations belong in the footnotes to the "mistaken" moves that led to them.

Also, I think that it is a MYTH that chess equates to human intellect. Kasparov bought into this whole heartedly and it was impossible for him to believe that he had been beaten by a computer. What surprised?
him was that IBM was playing to WIN. They probably consulted a team of psychologists (I would have at least) to plan the event. "Possibly?" they deliberately dumped the first game to psyche him out then played at full strength for the rest of the match (OOoohh another conspiracy... Those damn Freemasons and their secret hand shakes :-) ) Either way Kasparov was totally psyched out and humbled publicly. Hence... They must be cheating... In light of computer chess domination today, this seems unlikely.

Remember the "ICC Fischer hoaxes"? Even Short was 99% sure it was really Fischer. It was pretty clear to me that this was some guy with a computer (I'm in the IT field). What really interested me the most about this
was the preposterous openings that "Fischer" played yet won every game. (Granted these were 3 min games, however Short was one of the worlds best players at that time). This has kind of led me to question human
understanding of opening theory.

What I would be interested in seeing is the strongest engine (I think Houdini 2, though not sure) playing on ordinary hardware (quad, hex core) WITHOUT an opening book or tablebases at classic time controls,
perhaps as a competitor in a strong tournament.

This would potentially be very embarrassing. Has this ever been done? I would love to play over the games. When I used to play I found that I could play pretty much anything based on general opening principles and
solid moves (not obviously detrimental) and still expect to win against an opponent 500 points lower than myself. Granted I was only rated 2000 that means an opponent of 1500, so I'm sure this could steered into
a seperate discussion but I suspect the same situation is true for a 2800 rated person against a 2300 rated opponent. While these are generalisations I would love to put this to the test. This also would eliminate those "better prepared opening book" and "engame tablebases" arguments (excuses???).

While this proposition cuts both ways... the computer has no opening knowledge except for some "general principles" tweaked algorithms (if even that). The human is also out of book pretty quickly and has to play
for real much earlier in the game with the same old "burden of proof" and the clock ticking.

Thanks for reading my long winded post... these are just some of my reflections.

I apologise if I have double posted this but when I first posted it did not appear in the thread after about 5 or 6 hours so I reposted. Good thing I saved most of it in a file! Cheers!

You should probably try doing this before saying it, because I don't think it likely you'll have much success... :/

I put this into Houdini 2.0c, gives +1.5 to black, and considers rf3 as one of many good moves. Am I missing something? Is this position winning for white?

Yes, I presumed it was winning for black , I was wondering whether you were supposed to be surprising me by saying that this endgame isn't winning for black, which it is and the engine believes it is... the difference between +1.5 and +3 is the ply, shock horror; longer you leave the engine on in the position the more accurate it will get, I don't see what you've proved.

There are a few positions that engines still don't understand but they'll very rarely (if ever) show up in actual games and if you'd care to try to get an engine to convert into a lost endgame (per your "Trick any chess engine into a losing endgame, and it will go there like a puppy!") then you'll find that it's really not that simple... or probable... or possible.

I'm sure it will increase. I think you're missing what we're saying here, it's not important that an engine initially misevaluates a position, it'll get there after just a minute or so's "thought". If you're interested 53...Rc4 54.Rh5 Rxc5 55.Rxc5+ Kxc5 56.Ke4 is the initial mainline, after which on mv56, analysis gives +3.

The cheesykid still has a point.

@Lasarus33,

That is an intriguing idea about letting the engines play without an opening book.  I would be interested to hear if anyone has any experience with this.  I can only think it would be an advantage to the computer and a handicap to the human, but I am not an engine expert.

Why, thank you for this very flattering comparison, but IM pfren is way better than I am, especially in endgames