## Word Sudoku

For fans of Killer Sudoku, Samurai Sudoku and other variants

### Word Sudoku

Here is my first attempt at creating the type of word Sudoku hybrid I suggested in an earlier post on an unrelated topic -- one that is NOT solvable by Sudoku logic alone, but will require wordplay as well. Up until now, all other published letter Sudokus I've seen were solvable even if you replaced the letter with numbers. This hybrid seems ripe for experimentation.

Also, I've set the clues in an obvious pattern that though not symmetrical, does haver *order*. I've seen puzzles in Japanese magazines with every possible symmetry as well as many that were not symmetrical but had a shape or pattern such as a letter or symbol. "I picked a pattern of different sized triangles for no particular reason." he said, for no particular reason.

This is a standard 9x9 Sudoku except that it is NOT solvable without this additional stipulation: One of the rows or columns contains a 9 letter English word that can be found on webster.com. Since the word will have one of each of the nine letters in the puzzle, it shouldn't be too hard to discover the word and fit it in. Overall, this is an easy level puzzle.

Here it is without the box lines so that the pattern is apparent:
Code: Select all
`R G L . . A I T . I S . . . . N . . E . . . . . . . G . . . . . . . G N . . . . . . . . . . . . . . . . . I L E S I R . . N A . T . A . . S . R . . I . . T G E L `

And again, as normal:
Code: Select all
` R G L | . . A | I T .  I S . | . . . | N . .  E . . | . . . | . . G -------+-------+------ . . . | . . . | . G N  . . . | . . . | . . .  . . . | . . . | . . I -------+-------+------ L E S | I R . | . N A  . T . | A . . | S . R  . . I | . . T | G E L `
tso

Posts: 798
Joined: 22 June 2005

Cute!

As you've probably seen, I've got some word sudokus on my site (http://sudoku.top-notch.co.uk/wordoku.asp). I chose to put the word in one of the long diagonals; I think this makes the puzzle much more interesting, because you are then no longer restricted to words that use nine letters once and once only. This means you can use words such as "statistics" (5 unique letters) and introduce a few "spare" letters that don't appear in the diagonal thus making your puzzle much less vulnerable to cracking by simply finding an anagram (by brainpower or computer).

The ones on the site are, however, solvable by pure sudoku methods. Knowledge of the word isn't necessary to solve them. Of course, figuring out some of the cells from the word is always an option, and it does make things easier (or more enjoyable) for people who are more accustomed to a crossword mindset than the logic of sudokus. I think my puzzles are actually very easy to solve, especially as the sudoku itself generally doesn't require any of the "fiendish-or-harder" techniques.

My puzzles can of course be solved by any of the many solvers out there, so long as the user is prepared to do some simple swapping of letters for numbers. One of the big advantages of your approach is that the puzzle becomes much more invulnerable to solver programs.

If we combine hiding-the-keyword-in-the-diagonal and puzzle-not-solvable-by-sudoku-logic-alone then we get a puzzle that's resistant (but not invulnerable, unfortunately) to both sudoku-solver and anagram-finder programs. (That is until someone adds a dictionary to their solver-program )

Here's an example. The puzzle has multiple solutions, but only one gives a word running top-left to bottom-right. There is one spare letter involved which was enough to defeat this anagram finder.

Code: Select all
`. . . | . A . | S . .N . O | L . T | . . G. T . | D . . | . . .------+-------+------O . . | . . . | . N .. . . | I . D | . . .. L . | . . . | . . A------+-------+------. . . | . . S | . G .A . . | T . O | N . I. . G | . D . | . . .`

EDIT:
Actually, this puzzle is much better with two additional clues given:
Code: Select all
`. . . | . A . | S . .N . O | L . T | . . G. T . | D . . | . . .------+-------+------O . . | . . . | . N T. . . | I . D | . . .G L . | . . . | . . A------+-------+------. . . | . . S | . G .A . . | T . O | N . I. . G | . D . | . . .`

tso wrote:The best I can do is CATALOGUE and COAGULATE -- but even with these inserted into the puzzle, the puzzles still have multiple solutions. Do I have the wrong words, or did you not put in enough clues?

Those are the words I had in mind, so in that respect the puzzle works. You are quite right though, I set insufficient clues, and there are a lot of other solutions (albeit that don't result in words). I'm finding it a pretty tall order to make a two-word puzzle that has two and only two solutions!

-----------------------------------------------------------------
http://sudoku.top-notch.co.uk
Last edited by TN_Mike on Wed Jul 20, 2005 1:00 pm, edited 1 time in total.
TN_Mike

Posts: 7
Joined: 18 July 2005

Just solved the top one - bit mean choosing nine letters with so many anagrams, but very good nevertheless.
PaulIQ164

Posts: 533
Joined: 16 July 2005

TN_Mike wrote:Here's an example. The puzzle has multiple solutions, but only one gives a word running top-left to bottom-right. There is one spare letter involved which was enough to defeat this anagram finder.

Code: Select all
`. . . | . A . | S . .N . O | L . T | . . G. T . | D . . | . . .------+-------+------O . . | . . . | . N .. . . | I . D | . . .. L . | . . . | . . A------+-------+------. . . | . . S | . G .A . . | T . O | N . I. . G | . D . | . . .`

EDIT:
Actually, this puzzle is much better with two additional clues given:
Code: Select all
`. . . | . A . | S . .N . O | L . T | . . G. T . | D . . | . . .------+-------+------O . . | . . . | . N T. . . | I . D | . . .G L . | . . . | . . A------+-------+------. . . | . . S | . G .A . . | T . O | N . I. . G | . D . | . . .`

Good one! I can't believe I didn't see it sooner. It's much *easier* with the additional two clues, but not necessarily better. Creating puzzles that have a good balance of wordplay and logic will take finesse. I imagine being forced to use logic first, come to a dead end with enough additional information so that it would be possible but still a challenge to figure out the word play portion of the puzzle, but that finding out the word or words doesn't automatically finish off the puzzle.

Actually, even with the extra clues, there are two solutions -- r4c2, r4c3, r8c2 and r8c3 are either DSDS or SDSD. Of course, if the composer stipulates that the solution to the puzzle is the WORD, not the completed grid, this could be done intentionally! I've solved other logic puzzles -- not Sudoku, but similar genre -- where the solver was asked to find the contents of a particular cell rather than complete the entire grid -- and in fact, the entire grid could NOT be completed because there would have been multiple solutions -- and this fact worked to make the solving more difficult.

Anyway, I repeat that this is a very fertile and mostly unexplored area.
tso

Posts: 798
Joined: 22 June 2005

tso wrote:I imagine being forced to use logic first, come to a dead end with enough additional information so that it would be possible but still a challenge to figure out the word play portion of the puzzle, but that finding out the word or words doesn't automatically finish off the puzzle.

I have put online five word sudokus that attempt to do what you describe. They should have a unique solution that can't be arrived at without deducing the hidden words (plural! - I put two words in each, just because I think that adds more interest). Once the words are deduced, normal "logic" methods will complete the grid, and give the unique solution.

The puzzles are at http://sudoku.top-notch.co.uk/twodoku.asp

If you try to feed these puzzles into a solver program, you'll get an error or "multiple solutions" or whatever message. In an attempt to thwart those who would use a solver program, I've deliberately tried to set these puzzles in such a way as to give a high number of multiple solutions when fed to T&E solvers.

(tso - I have given you a credit on the page, if you would rather remain anonymous or want me to change the text, send me a private message.)

-----------------------------------------------------------------
http://sudoku.top-notch.co.uk
TN_Mike

Posts: 7
Joined: 18 July 2005

Fantastic. I'll give them a try. I like the fact you chose to used interesting atypical layouts as well.
tso

Posts: 798
Joined: 22 June 2005

When I suggested this, I didn't even consider placing the words on the diagonals, thinking that using 9 letter words in which all letters were unique was somehow better, especially if the grid could be created with one letter missing, forcing the solver to find an anagram of with one unknown letter.

It seems obvious that , your method leaves mine in the dust. It looks like the difficulty levels could vary far more then in a standard sudoku, especially using rare words and those with multiple duplicate letters. One, two, three or more words could be embedded -- just so long that IF the solver came up with the right words -- the puzzle has a unique solution. They probably can be made as logic puzzles with a little word play -- or word puzzles with a little logic -- or balanced between the two. The variety of puzzles is also great, as the number, length and placement of words are all variable. Crossword or better yet, Cryptic crossword style clues could be given as well.

In the end, I don't suspect the popularity of this form will be influenced much by the degree to which they are resistant to computers solvers, but it's always nice to be able to make them that way for online competitions.

(When solving the first one, it ternz out that I'm a teribal speler. I found the word on the / quickly, but because I thought it was spelled with 8 letters, I dismissed it.)
tso

Posts: 798
Joined: 22 June 2005