As some of you have noticed, over the last few months I’ve become more and more interested in so-called Japanese pencil puzzles and so I’ve had in mind creating a Battleships-type puzzle for a while. A couple of things had prevented me from taking it forward:

• No idea as to how to integrate a battleships puzzle into a “theme” for a crossword. I don’t really think that producing a crossword which leads to another type of puzzle just for its own sake is quite enough. To put it another way, there should be a reason for introducing the second puzzle.
• Trying to decide how hard to make the battleship puzzle and whether any unusual property could be used to make it more interesting to solve. Many battleships puzzles can have quite interesting logic which I wanted to force solvers to find. The difficulty is then to allow solvers to find the logic WITHOUT guessing. In terms of an effective solving technique a lot of battleship puzzles can be “forced” out by trial and error. This is not what I wanted at all so I wanted to make sure the underlying puzzle was difficult to solve by guessing.

While browsing the internet I came across such a puzzle which used the property that all of the ships were in just two rows and two columns. The high numbers given for these rows/columns allowed the solver to deduce that the four intersections of these rows/columns must be ships. Importantly this puzzle was also awkward to solve using trial and error so I thought it might be suitable for use in a crossword.

The next step was to create some sort of theme. The most obvious thing to do (and therefore what I ended up doing!) was to give the letters of a quotation/instruction in the battleships so I did some scanning of ODQ. Finding LITTLE SHIPS OF ENGLAND was pure luck but it fitted so well with the battleships idea that I had to use it. The whole Dunkirk thing appealed to me too. Solvers may recall war themes from other Pieman puzzles – the Churchill quote comes to mind.

A big change in my setting practices is that I often try to write the preamble before I create the grid nowadays – this followed something I read by Radix suggesting that this was what he did. I remember at the time being flabbergasted because the preamble was often cobbled together last in Pieman puzzles but I started to try it myself and found it imposes (sometimes necessary) discipline on the setter at an early stage. It forces the setter to refine the process he will go through to create the finished product rather than just jumping into the pool and hoping he will be able to reach the other end without getting cramp.

Anyway the preamble was sketched out now that I had met the two conditions above for making the puzzle ie I had the right sort of idea for the battleships puzzle and the right reason to use a battleships puzzle as the basis for a crossword. Creating the grid wasn’t too difficult from memory. 20 fixed cells is not too high, even in a 10×10 grid and I probably used Sympathy to help with the fill as it’s brilliant for puzzles with this sort of constraint. [Note to Ross – if Sympathy could ever be adapted to allow for misprinted letters which still create real words, this would be AMAZING!]. I’d thought it would be nice to make solvers explicitly note the Dunkirk reference but there was no way to get DUNKIRK into the grid as well. However DUNCE was possible so it went in too and the sketched preamble adjusted accordingly.

I knew I needed to provide the numbers for the battleships somehow and decided on one (well actually two!) of the hackneyed ways. I was also conscious of criticism (well I took it as criticism…) in recent feedback suggesting that Pieman clues had become too easy recently so I made a decision to produce a properly Piemanic set. Initial feedback suggests this was a reasonably tough set of clues so hopefully there won’t be too many solvers complaining they were too easy.

The clues are always the slowest part of a puzzle for me to put together. Sometimes I can write 10 clues in about half an hour (very rarely quicker) but sometimes one clue can take half an hour on its own. I try to reassure myself when this happens that Sabre takes on average one day per clue on his Listeners so I’m still skimping!

All told the above probably took about three days from start to finish to put together and then it was time for test solving. I expect that part of Pieman’s style has evolved from getting fed up with watching a set of clues which has taken a day to put together be destroyed in half an hour of Magoo solving. I seem to remember Mark finding the clues in Homers pretty tough, which was gratifying. Indeed both the Mr T clue (ANITA) and the clue for THE (in English “dis”) which I’d thought would be early solves turned out to be much harder. The battleships puzzle (which Magoo was familiar with) didn’t take too long although I was gratified to hear that he’d tried to guess it first and crack it by brute force but this hadn’t worked. I’d also been a bit concerned that solvers who got stuck on the battleships puzzle might realize there was likely to be a message in the grid and use the high numbers in rows to try and deduce it that way. Mark hadn’t even considered this so I hoped others wouldn’t either.

The title was easy from the Homer reference in the quote and the hint at what the boats actually did. The use of boat throughout the preamble was also deliberate as I didn’t want people who hadn’t seen these puzzles before trying to track down computer solvers of Battleship puzzles on the internet.

Anyway I think Mark found the puzzle between D and E in terms of difficulty but, on the basis that there was no monstrous “leap” to get to the solution, we plumped for D. I would finally add that the ELAND possibility was completely unintentional and we are awaiting a decision from our marker on whether it may be acceptable as an alternative to DUNCE.

Posted on Monday, October 1st, 2007 at 2:35 pm under Setting the scene.
Leave a response