Arimaa Forum (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
Arimaa >> Site Discussion >> Puzzle p45
(Message started by: loafes on Apr 16th, 2014, 1:27am)

Title: Puzzle p45
Post by loafes on Apr 16th, 2014, 1:27am
Hi guys, dont have time to introduce myself but just learnt about this game last night and have literlly only learned the moves hours ago so could easily be missing something.

anyway I went to do some puzzles and the 12th puzzle down (p45) it said silver to move and win in 1. I came up with silver to rabbit e6 then gold rabbit to f6 winning?

but then after hitting the answer tab it showed gold dog to g8 silver elephant to g7 silver rabbit to g6 gold rabbit to f6. Was there something wrong with my soloution, perhaps some rule I'm unaware of?

(like I said I just learnt the moves, I'll learn notation later.)

Title: Re: Puzzle p45
Post by supersamu on Apr 16th, 2014, 6:30am
Your solution is correct.
Capturing the last rabbit of the opponent wins the game.
The puzzle probably was added to the database to underline the fact that it is sometimes possible to capture the last rabbit of the opponent.
Trying to get a rabbit to goal is sometimes slower.

Title: Re: Puzzle p45
Post by loafes on Apr 16th, 2014, 7:08am
OK that's good to know, though it begs the question why the puzzle solution isn't the quickest win. Especially since it mentions at the bottom that it's from a championship game.

Title: Re: Puzzle p45
Post by 722caasi on Apr 29th, 2014, 8:47am
The reason I wrote the question was because the silver player, chessandgo, played another move that did not win. Therefore, I gave the win-in-one puzzle. Since the puzzle is "win in one move", and not "win in the fewest number of steps", both your answer and mine are equally valid.

Title: Re: Puzzle p45
Post by Fritzlein on Apr 29th, 2014, 11:51am
It's an essentially universal convention in the puzzle solving community that a puzzle should have only one correct answer, whether it is a crossword, dinner party seating chart, or a "win in X" chess position.  It is a sensible convention for exactly the reason illustrated in this thread: multiple correct answers create confusion.

While I understand that puzzles posed by real life usually don't have unique solutions, if we flout the universal convention in our puzzle collection, we are going to confuse and annoy people who expect anything entitled "puzzle" to have a unique answer.  Therefore, my preference would be to remove all puzzles with multiple solutions from our puzzle collection.  Not that it is pointless to study positions in which there is more than one correct move; on the contrary such study can be very fruitful.  It's just that such positions shouldn't be presented as "puzzles" to an unsuspecting audience.

Title: Re: Puzzle p45
Post by browni3141 on Apr 29th, 2014, 4:39pm

on 04/29/14 at 11:51:44, Fritzlein wrote:
It's an essentially universal convention in the puzzle solving community that a puzzle should have only one correct answer, whether it is a crossword, dinner party seating chart, or a "win in X" chess position.  It is a sensible convention for exactly the reason illustrated in this thread: multiple correct answers create confusion.

While I understand that puzzles posed by real life usually don't have unique solutions, if we flout the universal convention in our puzzle collection, we are going to confuse and annoy people who expect anything entitled "puzzle" to have a unique answer.  Therefore, my preference would be to remove all puzzles with multiple solutions from our puzzle collection.  Not that it is pointless to study positions in which there is more than one correct move; on the contrary such study can be very fruitful.  It's just that such positions shouldn't be presented as "puzzles" to an unsuspecting audience.

Normally, I would agree with you, but it is much harder to ensure uniqueness in Arimaa. I think many puzzle definitions even require each move in the winning line to be unique?

Anyway, I think we would lose a lot of good puzzles now and in the future if we restrict ourselves to unique solutions. I suggest restricting puzzles such that each problem in the set is conceptually identical. This is a little bit subjective, of course.
I really do not like problem 45, as there are many different solutions which don't resemble one another.
However, I would not call problem 98 flawed. The two solutions are nearly identical, and I would call them conceptually identical.

Edit: I have seen "real" chess puzzles with multiple solutions.

Title: Re: Puzzle p45
Post by browni3141 on May 3rd, 2014, 3:51pm

on 04/29/14 at 11:51:44, Fritzlein wrote:
It's an essentially universal convention in the puzzle solving community that a puzzle should have only one correct answer, whether it is a crossword, dinner party seating chart, or a "win in X" chess position.

There are many examples of puzzles where solutions are not perfectly unique, such as Sokoban, N-queens, Rubix cubes. Perhaps you mean something much more specific by the "puzzle solving community" which would exclude people that enjoy these types of puzzles, but it isn't clear what.

Title: Re: Puzzle p45
Post by 722caasi on May 15th, 2014, 1:46am

on 04/29/14 at 11:51:44, Fritzlein wrote:
It's an essentially universal convention in the puzzle solving community that a puzzle should have only one correct answer, whether it is a crossword, dinner party seating chart, or a "win in X" chess position.  It is a sensible convention for exactly the reason illustrated in this thread: multiple correct answers create confusion.

While I understand that puzzles posed by real life usually don't have unique solutions, if we flout the universal convention in our puzzle collection, we are going to confuse and annoy people who expect anything entitled "puzzle" to have a unique answer.  Therefore, my preference would be to remove all puzzles with multiple solutions from our puzzle collection.  


Fritzlein, this is an impractical requirement if we want the puzzles to come from real games, as this one did. This requirement is an artifact of chess having so few options per turn, resulting in real games having a unique best line.

Title: Re: Puzzle p45
Post by Fritzlein on May 15th, 2014, 10:23am

on 05/15/14 at 01:46:24, 722caasi wrote:
Fritzlein, this is an impractical requirement if we want the puzzles to come from real games, as this one did. This requirement is an artifact of chess having so few options per turn, resulting in real games having a unique best line.

Clearly, the more restrictive our requirements are for what qualifies as a "puzzle", the fewer positions will qualify.  That's the downside of having standards.  That said, there are plenty of positions from real games where the "goal in X" move is unique.  Indeed, aaaa has collected thousands of them.  Thus obeying the standard puzzle convention wouldn't mean our collection has no puzzles from actual games.

Browni, I'm thinking, for example, of the World Puzzle Championships, which have about fifteen different types of puzzle every year, and teams competing from all over the world.  As far as I know, no matter which country is organizing the competition in a particular year, and no matter what puzzle types they choose to be part of the competition, every puzzle in every year in every country has a unique solution.

By the way, the convention from chess is that only the "key move", i.e. the first move, has to be unique.  The continuation will obviously depend on how the opponent responds, and it is common that a sub-optimal defense allows multiple ways for the winning side to finish it off.

Again, we can have "puzzle" mean whatever we want it to mean.  It is just that departing from the standard meaning is going to confuse people.  It would be hard to argue, "No, we aren't going to confuse anyone," given the recent posts by people who were confused.

Title: Re: Puzzle p45
Post by browni3141 on May 15th, 2014, 12:49pm

on 05/15/14 at 10:23:16, Fritzlein wrote:
By the way, the convention from chess is that only the "key move", i.e. the first move, has to be unique.  The continuation will obviously depend on how the opponent responds, and it is common that a sub-optimal defense allows multiple ways for the winning side to finish it off.

I had meant to specify the main line, and I thought I had. It is considered a good quality of a puzzle (depending on the type) if for any optimal black move, there is a unique optimal white move, otherwise the puzzle is called a dual. Having one optimal white moves possible against sub-optimal black defenses is not as important.
Here is a 20 move example puzzle:
http://gameknot.com/chess-puzzle.pl?pz=3584
If you want more examples, just pick some random chess compositions of a "good type" by legitimate composers. A bad type would be "white to play and win (eventually)" because many of those will be "dual," but it's not fair to describe them as such.
Refer to this page: http://www.saunalahti.fi/~stniekat/pccc/codex.htm#c3
In helpmates, intended multiple solutions are often tolerated, or considered , such as in this puzzle:
http://www.reocities.com/parastrike/0312parastrikepz.html

Title: Re: Puzzle p45
Post by clyring on May 15th, 2014, 3:46pm
I don't think this confusion is inherent to having puzzles with multiple solutions, but rather stems from the fact that other solutions are not mentioned in the puzzle answer text.



Arimaa Forum » Powered by YaBB 1 Gold - SP 1.3.1!
YaBB © 2000-2003. All Rights Reserved.