Listen With Others

I don’t think I can explain how the idea for Hungry came about. All I can say is that I had been reading some Poe around the same time that I had been contemplating creating some puzzles when the idea of a puzzle based on The Raven popped into my head. I guess it makes sense though – if I knew the process for coming up with a good idea, I would have good ideas all the time.

The plan was to create a puzzle with the following features:

• clues with a gimmick that would lead the solver to a line from the poem and a hint to a ‘titular character’.
• the words “raven” and “Pallas” appearing somewhere in the grid.
• an instruction to move the word “raven” to somewhere appropriate.

This felt like it had legs, but I was sure that there was a more interesting representation of the raven waiting to be utilised. The raven is described luridly in the poem as being “Ghastly, grim and ancient” and I wondered if the words GHASTLY, GRIM and ANCIENT could somehow be made to form a representation of a raven.

I started to play around with sheets of squared paper to try and find an arrangement of cells (henceforth referred to as the “birdblock”) that a) looked a bit like a bird and b) could have its outer cells populated with the words GHASTLY, GRIM and ANCIENT. It soon became apparent that for this to work there would need to be a fourth word. The latest iteration of the birdblock had three cells in a row at its base. I realised that if those were filled in with POE and one further cell were to be introduced elsewhere then there would be the correct number of outer cells for the other words. I found a place where the extra cell could be introduced, and stage 1 was complete.

With the outer form of the birdblock defined, I started to think about the grid. I wanted there to be enough space to enable the birdblock to be repositioned in such a way that meant no cells were shared between the before and after positions. Using a 14×13 grid I found a pair of positions for the birdblock (and for Pallas), and I could move on to constructing the rest of the grid.

I decided that I wanted all final entries to be real words (ignoring the spaces that would be left by the movement). This is something I always appreciate in other puzzles and so I felt compelled to make it a feature of this one even though it would increase the difficulty of the construction.

To make things easier, I decided to allow some of the initial grid entries to be jumbles of the corresponding answers. I wanted the clues for these to be normal, but I also needed to reserve a number of normal entries for:

(1)   gimmick clues that were to be associated with the line from the poem

(2)   gimmick clues that were to be associated with the hint to the ‘titular character’

(3)   gimmick clues that were to be associated with additional thematic items/words

For (1), I had decided that the (partial) line I wanted to use was “Tell me what thy lordly name is”. I liked this as it could be referred to simply as a ‘request’ in the preamble and it is also from the line that follows that of “Ghastly, grim and ancient raven…”, making it easy to refer to one of the lines as following or preceding the other in the preamble.

For (2), I wanted the hint to be Corvus Corax. Apart from its thematic relevance, I also liked the fact that the sequence of letters meant that it was unlikely to be guessable until most of the letters had been determined.

For (3), I wanted some clues to have jumbles of thematic words that would be required to be removed before solving. How many of these I would include would depend on the grid construction.

Given the requirements of (1), (2) and (3) there were only a few entries that could be jumbled, not without having a large number of total entries. I was also conscious of the fact that, to justify having jumbled entries in the grid, there would need to be an above average degree of checking.

I loaded two instances of QXW and positioned them side-by-side: one with a grid with the birdblock in the before position and one with the birdblock in the after position. I placed PALLAS in the same position in both.

The ANC _ _ T* of the birdblock suggested ANCESTOR as an entry. This would work well for the initial grid as removing ANCEST would leave OR. The ANCEST in the final grid would work if broken down as ANCES | T since ANCES is a stem for many words. Because I was operating with a pair of grids, the bar appearing after the R in ANCESTOR (and its symmetrical counterpart) needed to be introduced into the final grid and the bar appearing between the ANCES and T in the final grid needed to be introduced into the initial grid.

I won’t give an entry-by-entry account of how the before and after grids were finally formed (primarily because I don’t have an audit trail that would allow me to do this with any degree of truthfulness), but I hope the ANCESTOR example gives an insight into the dual mindset adopted during construction. I will however note the following:

• I didn’t know if it would end up being a real word or a jumble in the initial grid, but I wanted 36a to have a final entry of POETRY.
• I wanted the letter above the C of ANCIENT to be O to give the raven an eye (I’m glad to hear that some solvers took it upon themselves to colour this in).
• Whilst initially unplanned, the OUST of 44a became a permanent feature of later iterations of the grids as the OUS was perfectly placed with respect to the initial birdblock, linking nicely with the title (hungry = ravenous).
• Some of the jumbles were formed in such a way as to make it obvious where some of the birdblock cells (before and after) were, e.g., ENTIREGN and NGTRAIPSE. This was to try and make the inevitable grid stare when searching for the raven a bit fairer on the solver (apologies if I failed in this respect).

During the early phase of construction, I struggled to keep the number of jumbled entries down to a reasonable level. I was tempted to revert to programming but in the end, I persevered with my method of ping-ponging between the two grids. In the end the grid(s) had 54 entries which was higher than I had planned. The number of jumbled entries was 17 and the total letter count of TELL ME WHAT THY LORDLY NAME IS and CORVUS CORAX is 36 meaning 53 of the 54 entries were accounted for. This left one clue that could contain a jumble of a thematic string as per (3). To tie in with the request, I wanted this string to be NEVERMORE, which could be referred to as the ‘response’ to the request in the preamble.

At this point I was ready to write the clues. The setting guidelines state that a word count of 650 (or greater) would be in danger of getting the puzzle rejected. This placed a restriction on what clue gimmicks could be used. With 54 clues to write, gimmicks involving extra words were off the table. Although tougher to write, I decided to use misprints in the definition as the gimmick. It took a while and several rewrites, but I eventually produced a set of clues that I was happy with.

Writing the preamble was tougher than I had expected. As well as having to contend with the word count limit, I also struggled to give clear instructions as to how the words GHASTLY, GRIM, ANCIENT and POE were disposed in the initial grid.

I was otherwise happy with the puzzle and so I sent it to Shane, hoping that it wouldn’t be rejected. After several months I received an email from Roger confirming the puzzle’s publication date. The editors had made some amendments to the clues, either making some of the jumbled entry clues fairer or reducing the letter count of some of the longer ones. The main change was made to the preamble which, as well being made more concise, was now less opaque than the one I had submitted. Corvus Corax was now described as a sensible answer to the request rather than as a hint to a ‘titular character’ (although it still achieved that). Mercifully, the grid required no changes to be made to it. The changes made by the editors improved the puzzle greatly and I would like to take this opportunity to thank Shane and Roger for their efforts in this regard.

I would also like to thank everybody who took the time to give feedback. Hungry is my first published puzzle and I wasn’t sure what people would make of it. People seem to have enjoyed it and the kind comments have made the torture of the grid construction worthwhile.

Android