A Christmas Chess Puzzle

Frederic Friedel writes "Here's a nice little chess puzzle I got from Grandmaster John Nunn many years ago. It looks incredibly simple, but even the strongest players in world have been stumped by it. The problem can be stated in one simple line: A game begins with 1.e4 and ends in the fifth move with knight takes rook mate. What are the moves? If you want to read a couple of stories on it, go to Chessbase. There is a very special prize to be won if you are able to solve it -- a book signed by some of the world's top chess players, testifying that the winner is The Greatest. "

Update: 12/25 11:50 by michael : Well, I thought I figured it out, but I was wrong.

1. e4     b8-c6
2. a4     b4
3. a1-a3  c2
4. a3-d3  b4
5. g7-e2  d3++
-->

Just to clear up some confusion below, the condition is simply that a knight makes the last move of the game, which is a capture of a rook on move five (either color), and this results in checkmate for the other king. Either the knight or some other piece could be giving the check. One poster below reasons that black would be the one giving the checkmate - this is very sound reasoning. :) You just have to think outside the box.

Comments

Reminds me

[Gambit Thirty-Two]

Reminds me of when Chessmaster 6500 beat me in 5 moves when it was set on level novice.


I'm a REALLY bad chess player.

CORRECT LINK

[Imperator]

http://www.chessbase.com/puzzle/puzzle.h tm

Simple chess engine

[Imperator]

Is there a simple engine that I can download? All you'd need to do is brute-force the possibilities. I might be very wrong, but 5 moves doesn't sound like it would take very long for a fast computer. Just grep through the results for the end move you want.

Re:Simple chess engine

[Rayban]

Have you ever seen the tree for chess moves? The number of possible board combinations after 5 moves is way, way more than a trillion or so. I wish I could have the number here.

What I'm curious about, however, is if it could be possible to do some sort of backwards extrapolation. Here's a bit of an idea:

(8:51pm - restate my assumptions ;))

1. There are a finite number of squares for a knight to land on in 5 moves.
2. The rook must be able to make it to one of these squares in 5 moves, so the knight can take it.
3. The king must be able to make it to a square accessible from one of these squares for a mate to occur.

Keep in mind these assumptions, as well as the fact that you might be able to castle, and you can reduce your workload dramatically.

Now... Does the program state that white knight takes black rook, or vice versa?

Re:Simple chess engine

[tpck]

Good God, don't do that. :)

And its probably more than you think. :)

But honestly, computers are killing Chess. :( Soon it will be possible to have a computer brute force EVERY SINGLE MOVE in a game and beat a human opponent everytime. Personally, I think this sucks. I'm all for advances in computer science, but Chess is a HUMAN game. It should be played BY HUMANS against one another.

I'm personally going to work through it, I don't care how long it takes. :)

I do ask that if you, or anyone else, brute forces the answer, that you don't reveal it or try and win the prize. That's just not fair, really. Actually, I don't care about the prize, but don't spoil the riddle for everyone who wants to work through it -- thats like giving away the ending to a good book/movie, or telling the final score to someone who has taped a sports game. Discovery through diligence is good. Don't kill that, please.

Re:Simple chess engine

[Restil]

All possible chess games out to 5 moves (including both sides) would make cracking DES look like counting to 10.

Also, the real trick is that the knight must MATE the king, which means that first, the king can't be able to move to a safe space. It either needs to be trapped or all other available spaces need to be guarded. And secondly, the knight can not be captured by any peices from the other side.

-Restil

Re:Simple chess engine

[miahrogers]

dang, think different for once.

all you need to fool a computer is a problem that CAN'T be solved with logic. like a riddle. Riddles are solved by defying logic.

Re:Simple chess engine

[rhyac]

True, the tree is huge, but if you have a halfway decent tree pruning algorithm, you could eliminate the vast majority of the tree.

For example (disclaimer: I'm not a chess person - I hate the game), if the goal is a checkmate in five moves with the knight, then... moving the bishop to one position, and then right back to where it started.. Well, that whole subtree can be killed right off the start. It's obviously (well, unless there's some amazingly subtle chess strategy going on here) a waste of two moves.

That is how (as far as I am aware) chess games work. they create a tree of moves, and say 'okay, this is the goal..'. Then they prune the tree to lop off the bits that don't show much promise in attaining the goal, and try to brute-force the rest in search of an optimal solution. if it can't find a solution at a given depth in the tree, then it chooses the most 'promising' path, and selects it's move accordingly.

Correct me if I'm wrong. (Or if this is an insanely outdated technique that modern chess theorists just laugh at).

Re:Simple chess engine

[Anonymous Coward]

Your assumption number 3 is flawed. It never said that the king must be mated BY the knight. In fact, I'll bet that the actual solution involves an "exposed" mate where the knight takes the rook, and exposes another piece, say the queen, that actually "mates" the king.

Re:Simple chess engine

[[Sn]]

You may reduce the workload dramatically, but you still have too much workload for even the most powerful of computers to compute in a reasonable amount of time, because the problem is still not "trivial" (a term my professor used to use to describe easy, ie polynomial or in some cases simply sub exponential time). I have worked on a number of similar problems (ie, given 4 particular pieces, give the one shortest winning sequence). This search tree is incredibly huge unless you utilize some of the heuristic ideas that algorithm genius' (as I call them) have determined to be most efficient for games such as chess, but that still requires that you know how to judge a good chess board, which is sadly something that I suck horribly at. If I remember correctly, the best algorithm I could come up with took approximately 4 minutes on a Pentium II 400, 128M RAM to guarantee the winning moves for a very limited board state. Given that the problem is not a limited state (in that any algorithm still must consider a much larger number of pieces, I dont think we would see a computer calculatable solution for this problem any time in the near future). This is why games such as the chessmaster series make us algorithm design guys drool. [Sn]

Re:Simple chess engine

[erasmus_]

The movie Pi, a must-see if you haven't already. But since it rings a bell for you, obviously you have :)

Re:Simple chess engine

[Pascal Q. Porcupine]

Pi (the movie).
---
"'Is not a quine' is not a quine" is a quine.

Re:Simple chess engine

[vectro]

Well, gnuchess is a Free Software chess program. You can search for it on freshmeat. I am presently working on an optimized brute-force solution to this.

Re:Simple chess engine

[Fourthstring]

Even forgetting for a moment the error in #3 (which someone else has pointed out), the three assumptions (plus the castling one) are incomplete. There are many rules in chess, you have forgotten one of them.

Re:Simple chess engine

[NI3]

I don't think you 'll find an engine on the market that can solve this problem in a reasonable amount of time. However, it looks to me that you could write one for this particular problem. The main decision you have to take is how deep you are going to use brute force before you filter the results.

Old Wolf wrote: "There are 119,060,324 different 6ply (3 moves) chess games". So abouth 6 million that start with e2-e4.
In the vast majority of these games it will not be possible for a knight to capture a rook in two moves. Checking for this condition can be done very fast ( 3,812,256 different positions for the four pieces,which are either valid or invalid: an array small enough to be kept in RAM).

You're left with a number of valid positions ( anyone an idea how many?) to consider. Depending on who mates, you'll have between 40 and 3000 possible moves to check (rough estimates).

My conclusion: the number of positions to examine will be in the order of 10**9. Writing the program will probably take you more time than running it.

Re:Simple chess engine

[dborod]

Quite correct. In fact, I found a solution in 10 minutes to this problem using this strategy of knight taking rook exposes the king to mate by the queen.

One problem though, this mate takes place on move 6.

Re:Isn't this supposed to be a tech news site?

[Yebyen]

Oops I'm wrong this can be tech-related... see the post before me :-) This post is not off topic, because it is an addendum. Moderators: If a reply to a post has something to do with the original post, can it really be off topic?

Re:Isn't this supposed to be a tech news site?

[Dopefish]

Ahhh....but, chess is the oldest technology, using the oldest technology.

Or...something deep n' cheesy like that. Woooo...

Re:Isn't this supposed to be a tech news site?

[aaron p. matthews]

yeah, this really isn't NEWS because this is an old puzzle, but it's new to most of us. And furthermore, what 'nerd' doesn't play chess??

white or black

[Bilibala]

Doesn't say if the 5th move is going to be mated by black or white.... guess it'll be easier if it's black mate in 5

Re:white or black

[Validus]

Ok I might be compleatly off but I thought that White must be first and therfor 1W 2B 3W 4B 5W so the fith move must be white. Now this is far to simplitiscic(sp?) to be right so I must have over looked somthing or read somthing wrong in any event could some one tell me why the fith move could be black. Thanks

Re:white or black

[malaclypse]

Each turn has a white part and a black part, in chess notation. So, the first turn consists of a black move and a white move. The puzzle, then, either has nine individual moves (five white and four black) or ten moves (five white and five black).

Cheers,
-K

Damn damn damn

[tpck]

Damn damn damn. I can do it in *SIX* moves. I try again. :) I'd help if it said who wins. My bets on Black. :)

What about Go?

[Kaufmann]

Yes, I know, chess is the "game of kings" and all... but I've recently learned Go, and no one else seems to play it. Does anyone else on Slashdot play this game?

Re:What about Go?

[CausticPuppy]

No, but I got a Diabolo for Christmas this morning, and I'm making it my new hobby.

Re:What about Go?

[m3000]

For the unenlightened, what is Go?

Re:What about Go?

[Kiwi]

The world champion Emmanual Lasker said that Go was a superior game to chess.

- Sam

Re:What about Go?

[thurinn]

Emmanuel Lasker was the CHESS world champion for a good 27 years or so. So, it's more like Bill Gates saying Linux is better than Windows.

Re:What about Go?

[neko the frog]

try looking into igs (international go server)
igs is kindof like a go equivalent of gamespy, except for the fact that it predates gamespy by at least four years ;) clients are available for most platforms, and at any given time there's about a hundred games going on involving everyone from 30 kyu beginners to like 9 dan masters.

here's their url if you're interested:
http://igs.joyjoy.net/

Re:What about Go?

[QuMa]

Yup, I play a bit. No, I'm not good, but I do play it. For a computer player get gnugo (requires CGOBAN IIRC, both oss).

Chess is a lonely game...

[Bumbling_Booby]

Knight --> fridge --> bourbon --> mate. It can be done in 4.

Scholar's Mate

[ajs]

I assume we are talking about Scholar's Mate, which is very similar to Fool's Mate. The two are described on this page in some detail.

The answer, from that page, is "1.e4 e5 2.Bc4 Bc5 3.Qh5 Nf6 4.Qxf7#"

Re:Scholar's Mate

[Compuser]

In your solution the last move is made by
a queen, not a knight as required in a
puzzle. Also, you only have 4 moves, not 5.
I suppose, this puzzle is not about the
quickest mate, but a very specific one.

Re:Scholar's Mate

[tpck]

Uhm, NO. :)

A game begins with 1.e4 and ends in the fifth move with knight takes rook mate.

This is the puzzle. KNIGHT TAKES ROOK MATE. That's what makes it hard. There are million ways to mate in 5 moves, but the usual way in which it is done is what makes this hard.

Please read the summary next time. Alot of people seem to be making this mistake...

Re:Scholar's Mate

[ajs]

Yes, it seems that I missed the crucial bit about the knight. I'll have to work on this one....

6 moves is easy

[/]

1. e2-e4 | b7-b6
2. f1-a6 | c8xa6
3. g1-f3 | g8-f6
4. g2-g3 | f6-d5
5. h1-g1 | d5-e3
6. g1-g2 | e3xg2++

It's whittling it down to just 5 that's tough.

Re:6 moves is easy

[tpck]

Heh, I've been doing it a completly different way. Or rather, ways. I've got half a dozen ways with each black and white. But I can't do it in 5. This is nuts. I might post them later, if someone else doesn't first.

BTW, thats pretty wack notation :)

Re:6 moves is easy

[cadelor]

Another way in 6:
1. e4 f6
2. Nf3 Kf7
3. Nh4 Qe8
4. Qg4 a6
5. Ng6 a5
6. NH8#

ok black needs to be really stupid but....

now more coffee to see if I can get it in 5 :)

Al

move five?

[illusionm8]

that means ther can only be FOUR more moves after the 1.e4. looks impossible

Re:move five?

[hellish]

By 'move' I believe they mean 'turn', where both white and black move. So the game would be completed on either the 9th(white) or 10th(black) 'move'.

Re:What about Go? (OT)

[jflynn]

I've played Go some. When I was going to school at Berkeley, and when there was a lunchtime Go club at a place I worked. Generally though, it's hard to find people to play Go in the US, I agree.

It's a more forgiving game for people like me who get things right the second time -- usually. :) It's also much, much harder to brute force.

rook mate?

[miahrogers]

what is rook mate and how does a knight take rook mate?

Re:rook mate?

[CausticPuppy]

That means that on the last move, a knight takes opponent's rook, and then it's checkmate.

Chess will NEVER EVER be solved by brute force

[jab]

Let's say a chess game has on average 60 moves by each player, who on average chooses from 10 possibilities. That's 10^120 possible games. Let's say you have a petahertz computer that can evalutate a move per clock. 10^15 moves per second. Let's say every proton in the universe was such a computer and they all worked in parallel. That's 10^80 computers, according to physicist's estimates. Solution time is 10^120 / (10^80 * 10^15) = 10^35 seconds. Since a year only has pi * 10^7 seconds, it would take way over 10^27 years to solve, or about 10^17 times the lifetime of the universe. Chess WILL NEVER be solved by brute force.

Re:Chess will NEVER EVER be solved by brute force

[Imperator]

Actually, rumor has it Transmeta is working on subatomic Beowulf clusters.

Point taken. :)

Re:Chess will NEVER EVER be solved by brute force

[jab]

Oops, should have used "preview" -- it's really
10^7 times the lifetime of the universe to purely brute force chess, with
a superfast, universe sized, parallel per proton computer.
Same conclusion, though.

Re:Chess will NEVER EVER be solved by brute force

[babbage]

Ever seen the math on the game of Go? Game board is 19x19, black & white players can go more or less anywhere. That gives 350+ possible moves for the first dozen rounds or so. Games tend to last many times longer than that of course. You can do the math yourself.

As fast as the chess possiblities rise, they are constrained by the possible movements of the pieces, the smaller board size, etc. Go rises orders of magnitude faster, and no computer program has to date (to my knowledge, mostly parrotting things I've skimmed here) been able to play even on a beginner's level. Fascinating stuff.

Not that this helps solve the Fermat's theorem of chess or anything.

Re:Chess will NEVER EVER be solved by brute force

[Chacham]

I'm not sure if this makes any sense, but I was always wondering if this would work.

Get a really, really large hard drive.

1. Map every possible combination of the chess board. That means every square a pawn, until every possible combination is saved as its own matrix.
2. Eliminate every impossible board, that is impossible because of by pieces or color. I.E. more than 9 queens or two bishops on the same colors while all eight pawns are still there.
3. Evaluate each matrix and record it's possible:
   • children via legal moves.
   • parents that can get here via legal moves
   • if the parent is the first board or if it is trusted I.E. it has the marking being connected to a parent or the first board.
4. Remove all boards that do not have trusted parents.

Step one may be relatively easy, in that once a matrix is designed, it just needs to be filled with every possibility.

Step two can be part of 1 or 3.

I wanted to try this for tic-tac-toe, which could then be extended to a board larger than 3x3.

----------------

Re:Chess will NEVER EVER be solved by brute force

[UnknownSoldier]

> Chess WILL NEVER be solved by brute force.

Don't say NEVER.

Sure, impractical at this time, but I CAN imagine a future, say in few thousand years, where it is doable.

Unfortunately that doesn't help us now.

Cheers

Re:Chess will NEVER EVER be solved by brute force

[jab]

NEVER. NEVER. NEVER. Death and Taxes will long have
disappeared and chess will NEVER be solved by pure brute force.

You think that's bad...

[Kaht]

I once played a chess game that, when you lost, the computer laughed at you...
And if you won, the board suddenly turned around, so you lost... and the computer laughed...
At least, I'm -pretty- sure this happened.

The more I think about it, the less it seems it really happened.

Re:You think that's bad...

[cheese63]

I had this handheld game, double dragon. i sucked real bad at it. once the batteries went bad while i was on a real high level too. i got so pissed off i smashed the fucking thing with a hammer. moral of this story, don't make offtopic troll posts on christmas, unless you have a real low karma like me.

Riddles

[Super_Frosty]

This riddle CAN be solved by a computer:

There once was a girl from Peru,
who filled her panties with glue.
She said with a grin,
they paid to get in,
and they'll pay to get out again, too!

Oh, wait, that's a limerick.

Re:chess

[Harry]

Check out the black bishop dude.
Nice try though....

-harry

Disappointed...

[Patzer]

I'm pretty disappointed that ChessBase couldn't make up their own puzzle at least...
I'm almost positive I have this puzzle in a book in my house. (but then again, I have a *lot* of chess books ;-) )

Oh well...

For those interested, Hugh Courtney's annual Christmas Puzzles in the british magazine "Chess" are especially enteraining ;-)

Better link

[/]

Here, assuming you only speak English.

Re: does knight have to have the checkmate?

[illusionm8]

the knight can take the rook last, but the queen can make the checkmate? if you read the page, at the bottom, the guy who didn't figure it out argued that it wasnt stated properly. ..."When I told him the solution on the phone I could hear Mikhail Botvinnik gasp in the background. And Garry, who was convinced I had stated the problem incorrectly, couldn't believe that he and his students had missed it."...

Re:chess

[Rupes]

>1. e2-e4 | g7-g6
>
>2. g1-e2 | g8-f6
>
>3. e2-f4 | h8-g8
>
>4. f4-h5 | g8-g7
>
>5. h5xg7+++
>
>(mate)

Sorry, this doesn't work - black's black bishop
takes the knight.

wait wait wait!

[illusionm8]

the knight can take the rook last, but the queen can make the checkmate? like the knight takes the rook and moves out of the way for a queen to make the checkmate. get what im saying?

if you read the page, at the bottom, the guy who didn't figure it out argued that it wasnt stated properly.

..."When I told him the solution on the phone I could hear Mikhail Botvinnik gasp in the background. And Garry, who was convinced I had stated the problem incorrectly, couldn't believe that he and his students had missed it."...

Re:wait wait wait!

[Crixus]

the knight can take the rook last, but the queen can make the checkmate? like the knight takes the rook and moves out of the way for a queen to make the checkmate. get what im saying?

Discovered check(mate).

Re:wait wait wait!

[tpck]

I assume that the Queen can make the mate, so long as the knight takes the rook. But I haven't found a way to make it work with the Queen in less than 6 moves. :(

I think I found the solution though. Or, rather, I think I know how to solve. Working on it now. :)

Verify this solution for me...

[Chagrin]

Apologies for my lack of knowledge of chess notation...

h7 - h6 (black king's rook's pawn)
g1 - f3 (white king's knight)
g7 - g6 (black king's knight's pawn)
f3 - e5 (white king's knight)
f7 - f6 (black king's bishop's pawn)
e5 - g6 (white king's knight)
a7 - a6 (black queen's rook's pawn - irrelevant)
d1 - h5 (white queen)
a6 - a5 (black queen's rook's pawn - irrelevant)
g6 - h8 (white king's knight takes rook)

I think that satisfies all of the requirements.

Re:Verify this solution for me...

[jellicle]

It's a very good solution. except that that is White's sixth move, not fifth.
--
Michael Sims-michael at slashdot.org

Brute force solution

[Qeyser]

A chess solution may be possible if you consider that most algorithms aimed at this problem are only going to consider reasonable moves and plausible board positions. Kind of like "guided" brute force.

Checkers on the other hand, - I've read somewhere that there are over 10^120 plausible board positions: orders of magnitude greater than the estimated number of particles in the universe ; )

-q

Uhhh... no?

[tpck]

The spoiler that's been posted is not a solution? I won't repeat it here, just in case, but as far as I can see, black checks white at one point, and white never moves out of check. If this is 'thinking outside the box' and this is the actual solution, this puzzle is useless. :) Breaking the rules to solve the puzzle is, in my opinion, cheating. Did I miss something?!

Close, but incorrect

[tpck]

Mate is given on whites 6th move, so this isn't a valid solution. Close though. :)

I think i got it

[Anonymous Cowpoop]

I must be missing something... maybe the pieces are wrong... I dont understand the letter-number thing... Move the right knight 2 forward and 1 left; 2 right and 1 forward; 2 forward and 1 left; 2 forward and 1 left. Gee that was really easy... could someone tell me how the board is lettered-numbered?

Chess Board Numbering

[Zaffle]

From left to right, when sitting on Whites side of the table, is A-H. From the white side to the black side, is 1-8.
So D1 is Whites Queen.

When stating a move, the result is sometimes stated by itself, when its obvious what happened.
(eg: e4, theres only one piece that can make that move, the pawn at e2).

For more info, check out your local chess faq :)

Possible Solution

[asmussen]

How about:

1 E2-E4 D7-D6
2 D1-E2 C8-G4
3 E2-E3 B8-C6
4 A2-A3 C6-D4
5 A3-A4 D4-C2++

Any problems with this one? The above spoiler does not work as white fails to move out of check at one point.

Re:Possible Solution

[tpck]

Sorry, but at no point does the knight take the rook, right?

Rules say: A game begins with 1.e4 and ends in the fifth move with knight takes rook mate..

You've got Knight takes Pawn mate.

Re:Possible Solution

[Rupes]

>How about:
>
>1 E2-E4 D7-D6
>2 D1-E2 C8-G4
>3 E2-E3 B8-C6
>4 A2-A3 C6-D4
>5 A3-A4 D4-C2++
>
>Any problems with this one? The above spoiler >does not work as white fails to move out of check >at one point.

I believe that a knight must actually take a rook, which invalidates this answer...

Re:Possible Solution

[jellicle]

But it doesn't end with knight takes rook.
--
Michael Sims-michael at slashdot.org

Re:Possible Solution

[matthewo]

You have to capture a rook with the last move involving the knight.

Re:Possible Solution

[asmussen]

Whoops. That's what I get for not paying close enough attention... :)

Re:Possible Solution

[illusionm8]

the condition is to take a rook on the last move. you didn't.

OMG.

[tpck]

I just had a very disturbing thought.

What if WHITE Knight takes BLACK Rook, and this causes mate for WHITE? Surely its not that. But it is said: And Garry, who was convinced I had stated the problem incorrectly, couldn't believe that he and his students had missed it.

If its some silly little trick in the wording, I'm going to be anoyed. Very much so. Grr. Of course, in the meantime, I'm not getting anywhere with this thing. 101 ways to mate in 6, but none in 5. Sigh.

I'm thinking maybe you need to promote a Pawn? Just a though. Anyone else out there actually working on this?

Re:OMG.

[illusionm8]

promoting a pawn? wouldn't that invlove much more than 5 moves?

Re:OMG.

[kaputsk]

I think one of the interesting things about puzzles like this is that you rarely find irrelevant moves. I've seen a few posts here that suggest things like 3. a4 .. 4. a5, ie. just pushing a pawn forward for lack of a better move. Usually in chess puzzles, every move is given for a reason.

And by the way, check out 1. e4 Nc6 2. a4 Nb4 3. Ra3 Nxc2 4. Rd3 Nb4 5. Ne2 Nxd3# given (albeit slightly incorrectly written) from an Anonymous Coward above.. It looks alright.. ?

Re:OMG.

[tpck]

But so would everything else... hehe. I dunno, I'm lost. :)

Re:OMG.

[kaputsk]

err.. to myself.. sorry, rook to d3 is not okay.. sigh. I think it's bed time.

Re:OMG.

[illusionm8]

plus, you can't put yourself into check, let alone checkmate

Re:OMG.

[glwillia]

I just had a very disturbing thought.

What if WHITE Knight takes BLACK Rook, and this causes mate for WHITE? Surely its not that. But it is said: And Garry, who was convinced I had stated the problem incorrectly, couldn't believe that he and his
students had missed it

Impossible. For a player to make a move which is checkmate for himself, he would have to make a move which would put himself in check. This is forbidden by the rules of chess, and thus not a legal move.

Eh? Looks nice

[kaputsk]

I think you mean: 1. e4 Nc6 2. a4 Nb4 3. Ra3 Nxc2 4. Rd3 Nb4 5. Ne2 Nxd3# Eh?

Re:Eh? Looks nice

[Kymermosst]

Nope. That should be:

1. e4 Nc6 2. a4 Nb4 3. Ra3 Nxc2+ 4. Rd3 Nb4 5. Ne2 Nxd3#

Notice black's 3rd move, Nxc2+, puts the white king in check. White's 4th move, Rd3, is illegal, as the king would remain in check.

Re:nope

[illusionm8]

your 3rd move. you cannot go from a3-d3 because you are ignoring the check that the knight has on the king.

Re:Black eh? with an odd # of moves and white star

[tpck]

Heh, actually, I've since changed my mind. I just thought it might be black, cuz white is too obvious...

Re:Isn't this supposed to be a tech news site?

[cheese63]

you can be moderated to offtopic posting anything. just read some of my posts...

um, heres a good question

[Rhydant]

whats this 'nice little prize' that guy is tlaking about? for all we know it could be something crappy. im sure someone here can git it right, you just have to think about it. maybe try working backwards or something. but im not gonna try because this confuses me. good luck y'all

Re:um, heres a BAD question

[illusionm8]

read closer

Zillions?

[MikeFM]

If anyone is a chess person could you make a quick variation of this puzzle that can be played on Zillions (http://www.zillions-of-games.com) please. I don't follow chess lingo but would be interested in playing. Coding games in Zillions is really easy so it should be simple. BTW anybody who likes board games be sure to try to convince them to port Zillions to Linux. They aen't unwilling but need some motivation. :)

Re: im thinking something to do with castleing

[illusionm8]

maybe the king/rook castles and the knight takes that rook, while moving out of the way for a queen?

My thoughts

[Adar]

1. It can't be done with castling (this is fairly obvious- no way to mate with knight at the end.)

2. The mate is to the king side (the side with no queen- think of the way a knight would have to move...)

3. It's not a discover mate. That's because b1/b8 to a1/a8 takes at least four moves- which doesn't leave enough time for a relevant queen or bishop move by either side (black must move a pawn.)

4. The mate does not involve moving any rooks. There are only four positions from which a knight can mate a king on e1/e8; there isn't enough time to move a rook somewhere relevant and make sure it's not protected by any other pieces.

5. The mate is apparently by black. I say this because, if it's by white, black only has four moves- but a black mate gives black five moves, which should be the least amount you need (assuming e2-e4 is irrelevant- it's a red herring.)

That's what I was able to figure out...good luck :)

Re:My thoughts

[illusionm8]

the mate doesn't have to be by the knight as far as i know. that would eliminate some of your ideas

Re:My thoughts

[David A. Madore]

(I know we're not supposed to talk of moderation on slashdot, but, would somebody please moderate Adar's post up? It's about the only interesting one in this whole discussion so far. And would the people who propose ``solutions'' please check them using xboard (run in ``edit game'' mode) before posting them?)

I agree with these conclusions. And it's infuriating. Castling seemed such an ingenious solution for getting the king and the rook in place; but it just takes too much time. And a discover mate seemed so elegant; but it just doesn't seem to work.

I also tried the following approach: move the white king's knight to E2 so that the king be completely surrounded by its own pieces, and try to mate from F3 (with the black queen's knight). Ingenious, doesn't it? And as usual, it ``almost'' works; only (1)moving the E2 pawn to E4, (2)getting the white knight in E2, (3) getting the white G pawn out of the way, (4)getting the white H pawn out of the way (so the rook can get out), and (5-6) getting the rook in F3 so the black knight can take it — all that takes 6 moves for white. :-(

Mating in G2 doesn't seem to work much better. And moving the white king seems to screw everything up.

Rats.

Re:My thoughts

[bons]

1. It could be a discover mate: The knight would have to take the rook on the knight's third move. The rook's protection would therefore not matter.

2. There is a very good reason for e2-e4. It is NOT a red herring. It is there to add challenge to the puzzle. I've come across a number of solutions if only white had not made that move. i.e. the move is necessary in order to limit the solutions.

3. The mate could easily be by white. I say this because e-2 e-4 allows the queen and bishop to move to advantageous positions to for #1 (above). In addition, black's position at the start of the game is much more vulnerable and he has the time to really put the rook in a stupid place.

The questions I face are:
1. Where can a knight get to in two moves that blocks c potential check (for a discovered check)
2. What protected position can a knight even get to to check the king by turn 5, (never mind mate)?

Arrrgh!
If I could get the dang rook to C-2 in three moves it would be easy. The black knight can get to c-2 in three moves and bishop takes queen at g4 involves 1 more move for white and 2 for black.
Taking an unmoved rook requires 4 moves for the knight. No time to bring out anything for a discovered check. and the only place to get the king in check on the kings side from knight take rook leaves the king one too many places to run. (I can't find a way to get the kind somewhere on the queen's side to result in an unmoved rook check).
Ok. I started this line of thought a half hour ago and I've come up with 7 failed solutions... I can feel my karma dropping.

Re:My thoughts

[malaclypse]

This thread seems to be where the analysis is.
I've spent an hour going over this problem, and of course I've found a number of six-move solutions and five-and-a-half move solutions. I found one five move solution, but gasp!, it requires that I leave white's king in check over a turn and thus rather badly violates the rules. It is fairly clever, though, imho:

(I'm doing this without looking at a board, I apologize if I make an error)

1. e4 Nf6
2. Be2 Ng4
3. Kf1 Nxh2+
4. >
5. Qe1 Ng4
6. Rh1 Nxh1#

(White move #4 is, of course, illegal, as white's king is in check).

Here're my contributions to your analysis:

• I've considered discovered attacks, but it seems impossible for this to work in less than six moves if you capture on h1/h8 or a1/a8.
• I agree that castling will not work (95% certainty).
• I have not been able to determine yet if it is black or white that mates -- although there are ten ply to work with if black mates and only nine if white mates, it may be easier to maneuver black's king into a smothered position with his pawn still at e7.
• Moving rooks may very well be the key, especially to h2.
• It seems obvious that this is some gem of a smothered mate, and the attacking colour surely uses no more than two pieces. It may be that the attacking colour uses only the knight.

Well, back to the board. Good luck.
Cheers,
-K

Cooperative Effort

[tpck]

Anyone wanna get together and do some realtime chatting on the subject? We've got a couple people here at least, bouncing messages back and forth, sorta, but I think we could go a lot faster if we could get a realtime discussion going. :)

Remember: All bugs are shallow to many eyes. Same thing goes for Chess problems. :)

Here, I got an idea. Anyone who wants to discuss the problem, I'll start a room called #chess on slashnet.org. come one come all. :)

Solution

[matroid]

I have discovered a truly marvellous solution to this problem, which however this textbox is not large enough to contain.

(Now if somebody else actually publishes the solution it will at least be named after me.)


"He who takes credit for everything, is bound to get credit for something."
-My Dad

Re:Solution

[ixjzv]

you are alluding to FLT?

The Answer

[Runna^Muck]

One of the players has to pee like a racehorse and concedes the game. Just a guess, however. I could be wrong.

Re:The Answer

[Brento]

That's it! That's the answer! You cheated and read it in a book, didn't cha?

Re:Black eh? with an odd # of moves and white star

[BlaisePascal]

Except that's not how moves are counted in chess. A "move" in chess counts both white and black, so a mate in the fifth move could end with either white or black moving:

1. White Black 2. White Black 3. White Black 4. White Black 5. White

or

1. White Black 2. White Black 3. White Black 4. White Black 5. White Black

no, look again

[/]

The king is on e1, so there's no way in heck he can get over to g1. As for e2, that's covered by the bishop on a6. He is truly checkmated, for all the good it does in this contest.

On black's fifth move, and knight takes rook.

[Chagrin]

E2 - E4 (the starting move required by white)
D7 - D6 (black queen's pawn)
F2 - F4 (white king's pawn)
G8 - F6 (black king's knight)
E1 - F2 (white king)
F6 - G4 (black king's knight - check)
F2 - G3 (white king)
G4 - F2 (black king's knight)
G1 - F3 (white king's knight)
F2 - H1 (black king's knight takes white king's rook - checkmate)

Re:On black's fifth move, and knight takes rook.

[Chagrin]

Aw dammit! I give up.

Re:On black's fifth move, and knight takes rook.

[Vrallis]

G3-H4 (white king)

Not checkmate. Try double-checking your board--unless you made a mistake in your transcript of moves above. This is probably a good possibility considering the line 'F2-F4 (white king's pawn)' is a paradox--F2 is white king's bishops pawn.

2 moves (fools mate)

[dw]

That's a variation of fools mate:

1. f4 e6
2. g4 qh4++

but alas, it doesn't involve a night or rook :)

Re:Isn't this supposed to be a tech news site?

[Hemos]

Hehe. If you want to locate tech news on Christmas, be my guest.

A really frightening thought

[Ledge Kindred]

Someone's eventually going to come out and tell us the solution to this, right? I mean, because if not, I may never get another decent night's sleep for the rest of my life...

-=-=-=-=-

Do you know ... ?

[Taco Cowboy]

This is going to be a little bit OT, but this needs to be said anyway ...

Do you know that both the game of Chess and GO originated from the East - from China?

The Chess game was introduced to the Western world (Europe and Middle East) by the Mongolian invaders, whose has learned the game of Chess from the Chinese.

The GO game was introduced to the Western world via the Japanese, whom in turn learned it from the Chinese.

Sorry for this not-so-on-topic message, but I think you might want to learn something this Christmas day anyway. :)

Merry Christmas !!

Re:Do you know ... ?

[Kaufmann]

IIRC, chess actually originated in India. Go is Chinese.

#chess on slashnet.org

[illusionm8]

anyone who wants to chat about this or wants to share ideas, come to:

#chess
slashnet.org

Re:dont count on it (nt)

[illusionm8]

dont count on it

Interesting trivia about Chess and GO

[Taco Cowboy]

This message is going to be a little bit Off-Topic, please do not get too offended ...

Do you know that both the games of Chess and GO originated from the East - from China?

The Chess game was introduced to the Western world (Europe and Middle East) by the Mongolian invaders, whose has learned the game of Chess from the Chinese.

The GO game was introduced to the Western world via the Japanese, whom in turn learned it from the Chinese.

Sorry for this not-so-on-topic message, but I think you might want to learn something this Christmas day anyway. :)

Merry Christmas !!

Re:My thoughts (and mine)

[Zaffle]

Another thing to note is the knight cannot take the rook in under 3 moves. This only leaves 2 moves to play with.

Castling takes 4 moves minimum. That only leaves 1 move to move the rook into a position that the other knight can then take it, and put the king into mate (discover or otherwise).

This leads me to conclude that castling isn't an option.

At first (and 2nd, 3rd, 4th... nth) glance, it seems impossible. There just aren't enough moves. And anyone who has tried this has come to the conclusion, "I can do it in six".

Grand Masters have been stumped. So this means one of two things.

it ISN'T classical. And its the sort of move which never happens in chess.

and/or Theres a trick in the interpretation of the instructions.

One question I have, is a matter of wording. and ends in the fifth move with knight takes rook mate. Is it possible that the knight it white, and the mate is done by black?

Technically, it'd still be move 5. And in move a knight could attack a rook.

Only problem with this is, it takes a knight 4 moves to attack a rook (assuming the rook doesn't move, since I'd assume the other player would be moving ass to get into a mate possition).

Notation (OT)

[TrbleClef]

How is the notation "wack"? It seems kind of normal to me...

Re:Notation (OT)

[MattMann]

the notation is less whacked, wacky, and wack if it's
1. e4 | b6
2. Ba6 | Bxa6
...

So how many ...

[sporty]

So how many slashdotters does it take to screw in..er.. solve a chess puzzle? Let us take this in steps...

1. Does it involve MS? If so, 75% of the people will hate the MS led answer, regardless of how it came about.
2. FreeBSD vs Linux? 10% will get so caught up on what OS Netscape/Lynx/Opera was running on to post the ideas to each other, they will never get anywhere. The MS people are exluded above. Mac users, who uses a mac? (running away)
3. .005% to the 5 first posters.
4. 14% will somehow discuss moderation or how checkers and go have similar problems. Thus starting new threads.
5. The rest .995 will have discussions on topic, but their message threads will be spaced apart by the rest of the banter.

There was hope..

---

Re:My thoughts (and mine)

[Adar]

Can't be done, for one reason: you can't put yourself in check.



Nice try, though.

Bleah. This thing took three hours of my life in which I could have been doing something productive, like playing Everquest...time to let someone else have a go.



Merry Christmas to all, and to all a good night :)

How literal?

[kaputsk]

The problem can be stated in one simple line: A game begins with 1.e4 and ends in the fifth move with knight takes rook mate. What are the moves?

If this is readable as "Knight takes rook," "Mate," then:

1. e4 Nc6
2. Qf3 Na5
3. e5 Nb3
4. e6 Nxa1
5. exf7#

There were mentions that there might be a bit of word-play going on here.. I wonder if it comes down to that?

Re:My thoughts (and mine)

[hellish]

It says 'ends in the fifth move with...'

The way I interpret that, it means that knight taking rook is the last move, since the game ends 'with that move'.

You could be talking about the knight moving from a position that was blocking a check(mate), but that wouldn't be a legal move.

If it's a "word trick" then this works...

[Anonymous Coward]

Try this

1. e2-e4 | g8-f6
2. g1-f3 | f6-e4
3. f3-g5 | e7-e6
4. g5-f7 | d8-h4
5. f7-h8 | h4-f2+++

This ends in "knight takes rook, mate"

Re:If it's a "word trick" then this works...

[wyv]

if your solution is the correct one, then:

1) congrats.. that's pretty smart ;>
2) curse this question for being worded so poorly.


If the answer uses "word games" in a devious way, then a pox on chessbase. I am reading the question to mean that on either white or black's turn during the fifth set of moves (5. XxX | XxX), there is a move in which a knight takes a rook, and this capture in itself is the mating move.

How are other people interpreting this question?

Hmm, looking at the slashdot page...

[Zaffle]

...where there are 2 articles: "Science: 50 Year Old Quantum Physics Problem Solved" and "A Christmas Chess Puzzle" next to each other and having tried this problem myself, leads me to daydream this:

Slashdot, 50 years from now...

Slashdot: 50 Year Old Chess Puzzle solved

Finaly 50 years after the chess puzzle was posted, an Anonymous Coward has posted a correct and working solution to the elusive chess puzzle.

This puzzle was first introduced 50 years ago at Christmas time, as something for the readers to chew on. CmdrTaco, Hermos, and thousands of other slashdot readers have not been to sleep since.

The article was upto its 10 millionth comment when Hermos finally archieved it on its dedicated 50terabyte drive.

Re:Hmm, looking at the slashdot page...

[malaclypse]

That ends with NxB, not NxR as the problem requires.

-K

Go: The executive summary

[Our Man In Redmond]

The brief version (for those who don't care to click on the link the previous AC provided):

Go is a game played worldwide, but has the strongest "community" in the Orient, where there are Go professionals and professional Go commentators and writers (especially in Japan). The rules are fairly simple but unfortunately not simple enough to reproduce here (especially since I'm doing this from memory). Very briefly, it's played on the intersections of a 19x19 grid of lines with pieces called stones. Players alternate placing stones on the grid, attempting to capture as much territory as possible by making it impossible for the opposing player to place uncapturable stones inside the territory. A stone or group of stones is captured if it is completely surrounded by enemy stones, so if a group can't be surrounded it can't be captured. Captured stones count against a player at the end of the game, so efficiency is paramount, both in securing territory and in trying to attack it.

I remember reading a summary of a book written over 25 years ago comparing chess and go in the context of Eastern vs. Western military philosophy (this was toward the end of the Vietnam war). The author's thesis was that in chess, the object is to capture a particular piece, and a player can sacrifice as many of his pieces as necessary to capture the king. In go, the goal is not to capture particular pieces (in fact, every go stone is just as powerful as every other -- it's how groups of stones are deployed that make them weak or powerful), but to capture territory, and as I mentioned above, the more efficient you are at it, the better go player you'll be,
--

On Slashdot and Chess

[Asparfame]

With so many people complaining about "decreasing quality" of slashdot stories recently, I must say that this is a great one! Chess and slashdot go very well together.

Just look up (or down depending on the moderation) and see that the comments are of a supreme quality, and the discussions are very intellectual. Not too much name-calling or bashing here.

How many people like the idea of having more chess stories on slashdot? Let's just maybe get an informal poll here. Maybe we could have chess (or otherwise) puzzles linked to more often. Solving puzzles is certainly a trait shared by many nerds.

Let's make it a habit. Who's with me?

Re:On Slashdot and Chess

[tpck]

Chess RULES. I don't know about meshing Slashdot and Chess tho. Mebbe a Chess box? But I wouldn't want to see regular Chess stories posted. There just isn't enough relevant Chess news. :)

I think a puzzle section period would rock. Have a mix of all different sorts of puzzles. Mebbe and art section as well. Diversification is cool. :)

The point of chess problems

[Old Wolf]

These figures from a chess-computer programmer:
There are 4,865,609 different 5ply chess games
There are 119,060,324 different 6ply (3 moves) chess games

Now, continue on this scale of increasing..
calculating all 10ply positions would take a long time.

Now, for my real point. Doing it by brute force misses the point. The enjoyment of chess problems is the effort and lateral thinking expended in solving them. I don't want to see a solution - I want to work it out for myself. (Big bad nasties to those who have posted lines on slashdot - I've tried to not read them).

For chess aficionadoes, here is a related problem by Sam Lloyd (which a computer would solve quickly, incidentally, but which I enjoyed finding):
1. f3 ???
2. Kf2 ???
3. Kg3 ???
4. Kh4 ??? mate
Fill in the four missing black moves

Word trick? I don't think so.

[assonfire]

Nunn seems to be the chess world's Puzzlin' Grand-pappy. Given his status, the "puzzle" here almost certainly lies on the chess board, not in the wording - if so, it wouldn't be at all pleasing to solve from a chess standpoint, which is the whole point of these things.


Though I admire your lateral thinking, I'm afraid this isn't "it". Or I'm wrong. ;-)

(OT) Re:Brute force solution

[natek]

A chess solution may be possible if you consider that most algorithms aimed at this problem are only going to consider reasonable moves and plausible board positions. Kind of like "guided" brute force.

Checkers on the other hand, - I've read somewhere that there are over 10^120 plausible board positions: orders of magnitude greater than the estimated number of particles in the universe ; )

Huh? Checkers has far fewer board positions--There are only 32 usable squares and 12 identical pieces per side, as opposed to the 64 squares of chess with 16 pieces per side with many different functions. Someone computed that there are ~5e20, or ~2^69 legal checkers positions, however many of those are unobtainable by legal play. This is a vast number even for a computer, however it is certainly possible that checkers will be solved in the near future. I'm not sure if an alpha-beta search is applicable to solutions of a game, but if it is, that would cut the workload down to ~2^35 positions, which seems almost piddling...

Chess, on the other hand, has an astronomical number of positions and will never be solved by brute-forcing with conventional computers. Should quantum computation ever become a reality that may change.

Of course this problem is many many magnitudes easier than solving chess, though still not quite pocket-calculator material.

Nate

Re:Black eh? with an odd # of moves and white star

[dodobh]

A *move* is 1> White Black . What you have said is usually termed as a ply (Ply == 1/2 move)

The solution

[Enfinity]

I think I finally got it.

1. e2-e4, g8-f6
2. f1-e2, f6-g4
3. e1-f1, g4xh2
4. d1-e1, h2-g4
5. h1-h2, g4xh2+++

Enfinity

Oops

[Enfinity]

Oops, move 4 is illegeal. Sorry.

Third assumption incorrect

[Andrew Cady]

3. The king must be able to make it to a square accessible from one of these squares [where the knight might be] for a mate to occur.

There could be a clear path from a rook to the king, except for the knight blocking it. The knight could then move out of this space putting the king in check from the rook, and then kill the rook that would have been able to move into the path himself, thus securing the checkmate.

Re:Isn't this supposed to be a tech news site?

[Syberghost]

Last time I checked, this was "News for Nerds", not "Tech News for Nerds".

Whether you love chess or hate it, you have to admit it's a passtime in which many nerds (and geeks too, for that matter) engage, and is thus "Stuff That Matters" for a large contingent of our colleagues.

So, in summary, Hemos; you go, girl. Keep posting that chess shit.


Note to the clueless; Hemo is not female. I am aware of this. Get off the computer and do something else for a while. Watch Springer or something.

Re:Isn't this supposed to be a tech news site?

[Syberghost]

Grrr. Hemos is not female, either.

Bad Syberghost, bad.

Thank you for wasting my day... :)

[bons]

Now I'm going insane trying to solve this.
Since I can't find my chess set, I had to dig one up on NoNAGS under their games pile (EGChess).
I couldn't find a decent java game that didn't require two different computers. (sorry)
Onthe brute force side, given that we have a limited number of moves and a limited possible number of moves in a turn, this should fall well into the brute force category. Hmmm. 9 moves maximum Black only has 20 possible moves. 8 moves after that, White has 28 possible moves. 7 moves left. At that rate it should be possible to brute force the dang thing in a reasonable period of time with enough decent resources.

But I'm going to sit here and pound my head on a virtual chessboard. I figure I'll find the answer just after everyone else.

Coming close (turn 6)

[ajs]

1. e4 f5 2. Ne2 g5 3. Nf4 h5 4. Ng6 a5 5. Qxh5 a4 6. Nxh8#

Ok, so you see that I've wasted a black move on 4. ...a5.

Can that move be used to re-position the rook so that this works? I was thinking earlier that the solution had to begin with 1. e4 h5 2. Qxh5 Rxh5, with white's bishop providing the checkmate, but I just don't see how that works, even if black's rook moves to the center....

Hope this helps.

Aussie fix???

[Issue9mm]

Okay, if I'm thinking "outside the box enough", I think this might satisfy the solution.

A game begins with 1.e4 and ends in the fifth move with knight takes rook mate.

That is the problem. Now, it's possible, that with a slight bit of imagination (which I have an abundance of), the sentence could be read as such: "A game begins with 1.e4, and ends in the fifth move with knight taking rook, Mate."

Just re-read the sentence in your best Aussie impression, and you've got it. Doesn't matter the sequence of moves, long as the fifth move ends in knight taking rook...

Too far outside the box??? I dunno... Never was all that great at chess...

Not mate

[David Price]

6. Bxd3

stalemate?

[BigBadaboom]

Stupid question perhaps.. (I'm no expert on chess terminology), but does the term 'mate' necessarily refer to checkmate? Could the game end, instead, with a stalemate?

Re:stalemate?

[mikera]

You couldn't get a stalemate (no available moves) this early in the game.... you could not have taken enough pieces. But 'mate' is conventionally used to mean checkmate by pretty much every chess player I've encountered. I've never seen an official definition though.

Re:stalemate?

[UnknownSoldier]

> but does the term 'mate' necessarily refer to checkmate?

Yes. Allways.


> Could the game end, instead, with a stalemate?
Not this game. Stalemate refers to where either player can't mate the other. i.e. 2 kings.

For draws, there are also
a) repeat-3 move, and
b) 50 move pawn rule.

Cheers

Re:stalemate?

[dsb3]

> Stalemate refers to where either player can't mate the other. i.e. 2 kings. It also applies if the player whose turn it is in unable to make a legal move.

apologies to Murray head (and you guys...)

[bons]

(the American:)
My room. Pure geek setting
And the coder don't know that the coder is getting
The worst damn puzzle in the chess world.
One that can't even be solved by Yul Brenner.

Time files - doesn't seem an hour
Been working on this puzzle, haven't even had a shower
No hope - don't you know that when you
work on this puzzle you need more than a small clue

It's an exposed check... or a castle... or a fools mate... or... or ... I don't know!

(Slashdot:)
One night of puzzles makes a hard man humble
Checkmate in six moves but the man says five
The knight moves slowly and I don't see a castle
No matter what I do, the king survives
Solving this dang puzzle just my take my life.

(The American:)
One move is very like another
Neither gets me closer to a quick kill brother

(Slashdot:)
It's a pain, it's a challenge, it's really such a pity
That you're trying to solve it since your rank's so shi77y.

(The American:)
Whaddya mean? You shoulda seen me in high school you rotton S.O.B.s!

(Slashdot:)
Linux, Brute force, need a hack.
This is worse than a Cal-tech Stack.

(The American:)
Get real! You're talking to a slacker
It's been too long since I was a hacker.
I get my kicks from coding legos, Slashdot!

(Slashdot:)
One night of puzzles makes a hard man humble
Checkmate in six moves but the man says five
The knight moves slowly and I don't see a castle
No matter what I do, the king survives
I can't solve this puzzle. I should ask my wife.

(The American:)
Slashdot's gonna be the witness
To the ultimate test of cerebral fitness
This grips me more than would a
unfair patent or bogus lawsuit

And thank god I'm only submitting a reply, not moderating it.

I don't see you guys getting
The kind of Karma I'm contemplating
I'd let you help, I would invite you.
If I could fing a solution on Yahoo.

So you better go back to your linux machines, your hacks, your compilers.

(Slashdot:)
One lousy puzzle and you're drinking vodka
The puzzle's locked inside your brain
It's like some bizarre scene from Kafka
And even whiskey wouldn't dull the pain.
This simple puzzle has driven you insane

One night of puzzles makes a hard man humble
Checkmate in six moves but the man says five
The knight moves slowly and I don't see a castle
No matter what I do, the king survives
Solving this dang puzzle just my take my life.

Apologies to Benny Andersson and Bjorn Ulvaeus, both of ABBA fame, and renown lyricist Tim Rice

yeah, but try it

[/]

For the queen to get involved it takes 2 extraneous moves if you're black, which you just don't have to spend if you're getting the knight over there to claim the rook.

If you're white, then you can do an uncovered mate in 6 like this, but that doesn't solve the puzzle:
(white) | (black)
1. e2-e4 | g7-g5
2. d1-h5 | a7-a6
3. g1-f3 | b7-b6
4. f3-e5 | c7-c6
5. e5xf7 | c6-c5
6. f7xh8++

If this is right, I'll buy a hat and eat it

[Uri]

1. e2-e4 | c7-c5
2. g1-e2 | b8-c6
3. a2-a4 | c6-b4
4. a1-a2 | d8-a5
5. d2-d4 | b4-a2

Presto ?!

Re:If this is right, I'll buy a hat and eat it

[wyv]

sorry, but no.

you're forgetting 6. d1-d2 or 6. c1-d2 or 6. b1-d2.

good thoughts though.

Think like a computer scientist...

[zimbu]

0. e2-e4 | b7-b6 1. f1-a6 | c8xa6 2. g1-f3 | g8-f6 3. g2-g3 | f6-d5 4. h1-g1 | d5-e3 5. g1-g2 | e3xg2++ There, now its on the fifth move :)

Good idea, but no

[/]

Look at all the pieces on the board. For there to be a stalemate, all of one player's pieces can't move. It just can't happen after 5 moves.

black has to take white

[surfsalot]

black has to take white based on the number of moves... 5 moves, white moved first, black has to move next. Meaning that black makes moves 1,3, and 5, white makes moves 2 and 4.

I need to do my laundry
Please send $3 to:
Jon Allen
p.o. box 308142

Re:Whats the point??

[Caspian]

Yes, and all that RC5 cracking "proves" is that dividing a brute-force problem into a lot of little slices and giving each slice to a computer will eventually find an answer (actually it proves nothing, proof by example is not a proof). If you had given just the keyspace chunk containing the right answer to just one computer, you still would have found the answer, without every other computer. :)

So what's your point? Sure, all it proves is that distributed work, uh, WORKS. Because it DOES. Sure, it's not as efficient as simply seeking out the One Person Who Actually Does Know The Answer, but it's a quick and dirty way to solve problems that sometimes-- as with RC5 cracking-- is the only feasible way to do it without a huge time/money budget.

No way

[mangu]

pick one:
6. b1-c3
6. b1-d2
6. d1-d2
6. c1-d2
6. c2-c3
6. b2-b4

It works

[Anonymous Coward]

1. e4 | Nf6
2. f3 | Nxe4
3. Qe2 | Ng3
4. Qxe7+ | Qxe7+
5. Kf2 | NxF1#

Re:It works

[EricWright]

Wow! That takes an incredible amount of stupid play on white's part. Twice, white passes up a pawn for knight exchange, and on move 4, sacrifices the queen for a pawn. The fool's mate is something white can completely miss until it is too late (the first time, of course). This is something else!

Eric

Re:It works

[Winged Cat]

It's called a "helpmate". Basically, the losing side plays so poorly that it appears to be actively, deliberately helping its opponent to win. It's a bit of a psychological leap to tackle problems where you assume helpmates (basically, you control both sides to get the desired end) vs. handling "normal" play where you assume your opponent won't play to lose. (At least, it was a leap for me.)

Ambiguous language

[/]

If yours is the solution, then it's because the English language is ambiguous when it comes to the word "with": either it means "by means of" (in which case your solution fails) or it means "accompanied by" (in which case your solution succeeds, although perhaps comment #145 is better in that the knight-takes-rook move also occurs during the 5th move).

I suspect it's not supposed to be a word-play like that, though, because there are several possible solutions that make use of that trick, and this puzzle probably only has one brilliant solution. But maybe I have too much faith in John Nunn.

Bi-directional Search Tree

[x mani x]

Probably the best way to brute force a solution to this problem is to use a bi-directional search tree.

Imagine two points in a space, one is the starting state of the chess game, and one the ending state. You branch outwards from both points, where the first state branches forward through game states, and the other branches backward through game states. Whenever either of these two trees touch (i.e. they share a common node), we have found a solution.

Given that the path length between the two points cannot exceed 5 (5 moves), it would be safe to limit the height of each tree to 5, or if you're clever, ceiling(5/2) (i'm not too sure about this part, data structures people should speak up about this).

AFAIK, using this search method should not strain a high-end vector processor in the least; it might even be doable on a high-end workstation. Given that the branching factor of a chess game is completely insane, a large memory would be useful to store each tree through every iteration.

The one thing that bothers me is that people a lot smarter than me must have already tried this. And failed, since no one has posted a solution. Maybe there is no solution?

Please correct me!

Re:What about Go? (OT)

[Kaufmann]

And I'm not even /in/ the US... :P

Looks right! (except for a typo)

[DanTilkin]

Black's fifth move is Nxh1#, not Nxh1#. Other than that, it seems to me to be right, and WinBoard agrees. Impressive!

Yeap GO is MORE complex to program !

[UnknownSoldier]

> Ever seen the math on the game of Go?

Yes. I was even going to implement a Go game.
And I quickly looked over GNU Go.

I would of thought that with only 2 pieces, GO would be easier to implement.

Uh-ah!

The strategy for GO is WAY more complex then chess. Even shrinking the board down to 9x9 or 11x11 is still difficult.

Someday (like in +2000 years) we'll have computers that will be powerfull enough to have the entire search tree in Ram. Until then, the human AI will have to do ;-)

Cheers

Simply Solved with the Help of Ti-82

[Spicy Bisquit]

Correct Solution is 1. e4 | Nf6 2. f3 | Nxe4 3. Qe2 | Ng3 4. Qxe7+ | Qxe7+ 5. Kf2 | NxF1# "make my taco salads pretty, jimmy!" - ucb

Re:Simply Solved with the Help of Ti-82

[mangu]

Seems right. Congratulations!

Re:Simply Solved with the Help of Ti-82

[bons]

Congrats.
I had figured out I needed the queen to do a verticle attack.
I had figured out that I needed the knight to take the rook and trap the king exactly that way.
I never throught about using White's Queen to get my dang pawn out of the way!

I bow to your superior wisdom.

Re:Simply Solved with the Help of Ti-82

[MattMann]

nicely done!

what, pray tell, did you use the Ti-82 for?

Re:Simply Solved with the Help of Ti-82

[Zaffle]

Well done, its a winner.

and btw - the entire slashdot world says one collective "Doh!".

I hope you sent your solution to the guy before you posted it here :).

Now, as to why the rest of the world couldn't figure it out. Personally, I was wayyyy of track in my last assumption. (Its ethier a trick of words, or something that never happens in normal play), and I think most people didn't find it because its valid play and its also very agressive of both sides. In almost all the solutions I worked on, I very rarely considered taking pieces. (except for Knight takes Rook :), I wouldn't rule it out ofcourse, but the way I was working, the less pieces you have, the less you have to block the king with.

The cunning in this solution is finding the location that the king is block by his own troops, and his troops can't save him. In the final resting place, the king could only move 4 ways. 3 of which were blocked by the black queen, and the other blocked by the black Knight.

But as I predicted, it didn't involve castling, and it did involve primarily the black knight kicking butt to get over to the other side.

Its a shame the puzzle has moved from the "dark corridors" of chess out into the public eye. I imagine now it will be difficult to use such a thing to stump your friends, but never-the-less, It will be fun to try.

I would be keenly interested to hear exactly how "Spicy Bisquit" did indeed find the solution, and how a Ti82 helped him.

Nxe4 was brilliant too. :) I think the best tip you could give someone who was trying to solve this, would be "White must sacrifice her queen". It would ofcourse, make it 10 times easier :)

Once again, congrats to the winner, and we are all keenly awaiting your next post telling us exactly how you did it.

Re:Simply Solved with the Help of Ti-82

[crayz]

http://slashdot.org/comments.pl?sid=99/12/25/20252 13&cid=207

That one 12:04 PM EST, this one 2:15 PM EST.

Hmmmm.

Re:Simply Solved with the Help of Ti-82

[/]

He used it for the part where he bullshitted you into thinking he didn't just steal it from the anonymous coward who posted it earlier.

A lesson in scientific theory.

[bons]

I think, if one was to study through the methods eveyone was using you would find a lot of flawed assumptions. I know I read and thought of a lot of lines of reasonings that made sense at the time, were very well thought out, but had absolutly no proof.
What assumption killed me? The assumption that I had to move a pawn to bring the queen into play. If I has challenged that assumption I might have had more Karma, or at least, a cool book.
It's nice to have my assumptions challenged a week before I find out whay Y2K really does.

The correct answer (AFAIK)

[Rayban]

This was posted earlier by Spicy Biscuit. I'm just writing it down in a different format for those who prefer it:

1. e2-e4 g8-f6
2. f2-f3 f6-e4
3. d1-e2 e4-g3
4. e2-e7 d8-e7
5. e1-f2 g3-h1++

Very ingenious solution :)

Re:The correct answer (AFAIK)

[epsilonboy]

perhaps i'm mistaken, but that should read

4. e2-e7 e8-e7

or i'm not thinking straight. but unless the black queen takes the white queen instead of the black king taking the white queen, move 5 is not checkmate, since the white king can escape with
6. f2-e2

otherwise, the solution appears sound.

Re:The correct answer (AFAIK)

[Rayban]

Nope... notice that the black queen invalidates that move. The King is definately trapped there...

Re:Frederic answering

[David A. Madore]

(Frederic: creating a Slashdot account takes only a few seconds. All you need is to remember a password.)

If we have by the end of February, I say we can brute force this. There are only eight or nine unknown plies; if we assume that there are around 20 possible moves per ply, which I think is a reasonable order of magnitude (``reasonable'' logarithmically, of course), this makes 500 billion combinations, and at a very very reasonable rate of 10^5 positions analyzed per second (a quite conservative estimate: most engines will do better than that on a modern computer, and they are evaluating, not just checking for mate) this takes 60 days. Put in two or three computers full time for two weeks and we have the answer for sure. And probably much before that, since we know, for example, that the eighth or ninth ply has to be ``knight takes rook''.

Does anyone know enough about the internals of either crafty or (the new 5.00) GNU chess so that we can extract their movegen and set them running? From the cursory glance I gave them, the source looks like a horribly tangled mess (less so in GNU chess than in crafty, but still rather unappealing).

Still, it would be more satisfactory to solve it ourselves. ``When you have eliminated the impossible, Watson,'' said Holmes, ``whatever remains, however improbable, must be the solution.''

Re:Isn't this supposed to be a tech news site?

[nathanh]

Whether you love chess or hate it, you have to admit it's a passtime in which many nerds (and geeks too, for that matter) engage

Do you think so? I don't see how that explains the god-awful chess skills being shown here then.

• People thinking it's fine to move a piece so as to put yourself into "check".
• People thinking it's fine to move a piece so as to put yourself into "mate"!
• People not knowing what a "move" is in the game of chess, thinking instead of a "ply".
• People not knowing what a "mate" means in chess, instead posting answers for "check".
• People thinking it's fine to remain in check while you move other pieces.
• People not knowing what "discovered check" means in the game of chess.
• People thinking it's possible to reach chess stalemate in 5 moves! Sheesh.

And I'm not even going to begin counting the number of people who can't even count up to 5, as they're posting answers involving 6 moves.

I'm hoping everybody had a bit too much wine with dinner, because if this is the quality of the "Linux Geek" then I'm scared. Very scared.

Perhaps it's people thinking along these lines:

"I know computers, you have to be smart to be good with computers, chess people are very smart, so I must be really good at chess too!"

I can't otherwise explain the number of plain dumb responses on this article.

PS: Thank you to the guy who posted the solution, I was going nuts and I knew I didn't have a hope of solving this problem. I'd not even thought of using the white queen to release the black queen. Very imaginative!

Clarification on the puzzle.

[matthewo]

I sent email to Frederic Friedel and he responded telling me the puzzle is *not* a word game. In other words the correct solution has the form: 1. e4 ??? 2. ??? ??? 3. ??? ??? 4. ??? ??? 5. NxR# OR 5. ??? NxR# (Depending on who wins). All moves are legal.

Bogus

[Woundweavr]

Looks like this was copied from an AC above. Has the same typo that the AC had, and is posted 2 hours later.

A similar yet equally delicious problem

[kiolbasa]

The game ends in checkmate in the fifth move with a pawn promotion to a knight. Enjoy.

How I solved it (not poster)

[2apenny]

Firstly, I am not the poster of the solution, but I did solve it independently last night. I'm in Indonesia on holiday. I just logged on this morning to see if I had it right. It took me about 1.5 hrs. Having a chess board might have helped :) Firstly, I reasoned it was probably black that gave mate as the problem did not state that it was white. I did spend some time searching for a white solution, but not much. I tried to block in the white king with its own pieces, but this takes too long. There were only a few ways to do this and none gave a valid mate. The move e2-e4 was not helpful for this. Castling seemed a waste of time. Took too long and the rook ended in a very inconvenient position. The main problem was to get the knight across fast enough. It was taking 3/4 moves. Then I tried to help out by bringing the rook to a more convienient place. Closer but still no luck. Then I reasoned that in such a difficult puzzle the move e2-e4 was probably important, but why? I didn't think it possible to mate with knight only so I tried bringing out another piece and trying a discovered check. But this takes two moves which was too long, but how else could it be done? I had ealier found it helpful to have the black night remove an annoyingly positioned white pawn, that lead to a "solution" that wasn't valid chess - white ignores check. Well what if white were to remove a black pawn... But now it took too long for white. If black were to take the white piece taking the pawn it would help, but it was still taking two moves to get the piece somewhere useful. What this puzzle really needed I thought were some queens whizzing about the board. At this point the e1-e8 line looked very suspicious. And armed with the idea of white removing a black pawn and the black knight removing an awkward white pawn it all fell into place. Of course it wasn't so straight forward, but that was generally how I did it. (I skip the part where I tried to figure out how long it would take to write and run a program to brute force it - a long time, and the frustating bit at the end where I forgot the solution) I'm not a good chess player. I think luck may have had something to do with it for me.

Re:How I solved it (not poster)

[MattMann]

Firstly, I reasoned it was probably black that gave mate as the problem did not state that it was white.

Much has been made here of trying to decide "white or black to mate" a priori. It's a puzzle! There're only 5 moves, it might be black, it might be white. Will that single extra tempo be "required"? Does e4 open a gap in the defense, or does it just open an attack against a truly unorthodox opening by black...

The only bit of cursory analysis that turns out to have bearing on the answer is "it takes 4 moves to get a knight all the way across the board, and e4 would "waste" the remaining tempo"... but a priori we don't know this will be the scenario.

Congratulations, though. Who am I to talk, I didn't solve it :)

Possible Solution if it's a trick question

[BagMan2]

I haven't played chess in 15 years and was never more than a casual player (the computer still beats me on level 1 most of the time); but, depending on how you look at the problem, I may have a solution. I am not convinced that you have to count the starting white move in your 5 moves. The problem seems to imply you have 5 moves from the stated starting position.

Other solutions I have noticed below have required that both sides take the 6th move in order to solve it. It would seem to me that if only white takes the 6th move, it could possibly be a solution since white will have only taken 5 moves from the stated starting position of the board. It all depends on whether that first white move that you have no control over counts or not. Anyways, here is my solution (forgive the notation, I am an amateur and it's the only notation I can figure out):

1w. e2-e4 (don't count this???)
1b. e7-e6 (black kings pawn forward one)
2w. e4-e5 (white pawn forward one)
2b. e8-e7 (black king forward one)
3w. b1-c3 (white knight)
3b. g8-f6 (black knight)
4w. c3-e4 (white knight)
4b. f6-e8 (black knight)
5w. e4-f6 (white knight)
5b. h8-g8 (black rook left one)
6w. f6-g8 (white knight takes rook, checkmate)

As you can see, if white doesn't have to count the mandatory move, the problem is solved. Given that an amateur like me figured it out this far, I suspect the problem requires that you count that first move, but perhaps somebody else can figure out a way to shorten what I have shown here by a half-move...

Re:Isn't this supposed to be a tech news site?

[Syberghost]

While I understand that you weren't actually responding to my post, just using it as an excuse to toot your own horn and engage in a polemic against people who you don't feel meet your standards for chess skills, I think you're doing a tremendous disservice to the community by discouraging people from exploring.

If someone makes a post that clearly indicates they're interested in something, but not expert level at it, a much more community-oriented response would be to gently guide them to a greater understanding of the subject.

I mean, the guy who posted the solution could have insulted *YOU* for not meeting his standards of chess skill.

Re:Isn't this supposed to be a tech news site?

[nathanh]

It has nothing to do with being "expert level", but understanding the basics should be a given, such as how pieces move and what the rules are, before blithering away.

You can hardly "explore" a chess puzzle if you don't know chess. This would be like trying to find a syntax bug in your code without knowing the language or how to invoke the compiler.

People would rightly complain if some ninny came onto comp.lang.c and tried to fix someone's bug with "have you tried putting line numbers in? I think that might help". Same thing here.

FYI, my post had nothing to do with my own chess skills (which are pitiful). This is nothing more than an underhanded personal attack: equivalent to crying "communist" to avoid the real argument.

Re:(OT) Re: Solving Checkers...

[CodeShark]

IIRC, Checkers has already been essentially solved, at least for Red. I remember reading a number of years ago (and would be interested in commentary either validating or contradicting it) that using branching logic -- at least compared to chess, there are relatively few opening moves -- computer programs had been designed that could not be beaten by "black" no matter what. In other words, the computer didn't need to solve all possibilities to win, just those which were possible and plausible in ordinary play.

The author of the article did not say whether there was any calculable way of using some obscure trap (maybe a sacrifice of several pieces which would open up the board) which would allow black to escape from the computer's move progression that would ultimately leading to a black victory or draw.

How I solved it...

[Winged Cat]

I waited for others to post the solution, then posed the question to some chess buff (but not really 'Net aware) friends of mine. I'll be merciful and give them the solution...in a year or two, after obsessing about this problem has driven them insane, shattering their will and priming them for mind control. ;)

No, I'm not serious. Except about taking the cheap (...er+faster=better) way out for getting the solution ('cause I already have enough problems to solve).

I've actually solved it.

[dannybones]

I worked on it for almost three hours, and had given up, when I had an idea. It didn't work, but I was in that state of mind where the inspirition was flowing, and the idea for the solution came to me. It was really slick. I stared and stared at the checkmate in disbelief, and began to almost get hysterical-- I had done in 3 hours what Kasparov and the other grand masters couldn't do at all. I checked the solution many times over. The pieces were set up correctly, the knight took the rook on the fifth move, and it really was checkmate- the king had no legal moves, and nothing could take the knight. But there it was. While I won't post the solution here until I get a confirmation back from ChessBase, trust me, it's the solution.

Here are a couple of minor hints:

- It's not a semantical trick- you can really solve this by playing through the five moves.

- Black wins

- Every move is essential-- there's no wasted half moves

- The knight does the mate-- no double or discovered checkmates

- No castling

- NO PAWN PROMOTION! >:

- No fancy or archaic rules were invoked-- it's just straitforward, legal chess!

Have Fun

Dan Rodgers

solution?

[davidmacq]

What about?

1.e2-e4 e7-e5 2.Ng1-f3 Qd8-h4 3.Nf3-g5 Bf8-c5 4.Ng5xf7 d7-d6 5.Nf7xh8 Qh4xf2# 0-1