Listen With Others

Should I ever be called upon to choose a suitable soundtrack for my puzzle setting career, it will certainly not feature Edith Piaf warbling “Je ne regrette rien”, my tendency being more towards the Frank Sinatra school of regrets. Oh yes, during that period between submitting a puzzle for editorial consideration and reflecting on its reception by solvers, I’ve had a few — but as you are about to learn, they are not too few to mention (and anyway, Frank, surely the barrier to comprehensive regret-mentioning would be superabundance rather than paucity?).

The most vexing of them fall into three broad categories. Worst of all, perhaps, are the times when it has become clear from the online forums (which I check repeatedly on publication day with hope and trepidation in unequal measure) either that the endgame required a mental leap which was inadequately signposted or that the finishing line could not be confidently identified. Regardless of the merits of the rest of the puzzle, such faults inevitably leave solver dissatisfied and setter rueful.

Next on the list come those instances where, upon solving the initial proof, I have discovered issues (usually manifested as some form of ambiguity) which seemed to me to render the puzzle unsatisfactory. In my early days of setting, I would tell myself “It’ll be all right, no-one else will notice.” Experience has taught me that if the only reason I can sidestep the hole is because I dug it myself, others will undoubtedly fall into it, and they won’t enjoy the experience; now, if there’s time to fix the problem I will, but sometimes the proximity of the publication date means that it just isn’t feasible (I guess that using test solvers might help here, but I’m too set in my ways to start now). Result: self-flagellation of the mental kind.

Lastly there are those puzzles which at a late stage I realise simply don’t do justice to the theme — “I should have included that name in the grid, or done the thing as a Carte Blanche, or spent longer developing the idea.” Even if the puzzle is reasonably well received, it still feels like a good opportunity wasted. And when you struggle like I do to come up with original themes, that is a cause for considerable regret.

If I manage to avoid these pitfalls, the resulting puzzle may still not be up to much but at least I will look upon it with a degree of authorial affection.

With Get Weaving, my starting point was the idea of intertwining two normal (but meaningless) clues to produce a composite with a relatively sensible surface reading. I then looked for a theme with a weaving connection, and the Arachne/Minerva story seemed to fit the bill perfectly, in particular the potential ARACHNE->A SPIDER transformation. Using interwoven answers to make up the weft would allow me (I hoped) to introduce the combatants into the grid without their names needing to span bars. I realised that the grid-fill for the solver would not be trivial, so I wanted the endgame to be simple and unambiguous.

The challenge of populating the grid was compounded by the need to include in a single block six pairs of words which varied only in their third letter, as required for the RACHNE/SPIDER change. Half of the 24 across solutions ended up being words which I would describe as ‘unfamiliar’ (to normal people, if not to those of us with multiple editions of Chambers on our shelves), but I didn’t feel that I was going to be able to reduce the ratio without jeopardising my last lingering trace of sanity.

For the interwoven clues there could be no redundant words, while the down clues needed to be readily blind-solvable. In an attempt to make things slightly easier for the solver, I determined that the first letter entered in a column would always belong to the first solution, and that where two across solutions were the same length the first word in the clue would belong to the first entry. Ultimately it was decided not to include this information in the preamble, although some solvers apparently worked it out for themselves.

An extra (self-imposed) requirement was that the unwoven across clues had to be made as sound as possible: I couldn’t see the editors wanting to rewrite them, and I certainly had no wish to do it myself! In the event, writing these clues didn’t in itself prove inordinately difficult; the main issue was that producing half-decent surface readings meant that the separated clues were quite a bit trickier than I would have ideally liked. The high proportion of uncommon words contributed significantly to that problem, a pair such as ESSENE and TEASEL being unlikely bedfellows.

I knew that the finished puzzle wasn’t easy, and I was concerned that the down solutions might be of very little help in the solving process. I was reassured by the fact that both vetters managed to battle their way through it, and when I came to tackle the proof myself I found that solving the down clues was in fact the key to cracking the puzzle. I can’t recall ever completing a barred crossword without doing a certain amount of ‘reverse engineering’ (and I see no reason why a clue cannot legitimately support a hypothesis as well as give rise to one), but it was clear that the across clues here were going to require an unusually generous dollop of it.

I wasn’t surprised that the response on the forums was mixed – those who see the grid-fill essentially as the means to reach the endgame were always going to be disappointed, and on top of that the solving process here could be viewed as either an exacting challenge or a disagreeable slog. Those who described the puzzle as ‘an old-fashioned Listener’ didn’t venture to add whether that was a good or a bad thing!

And as for regrets? None on the three counts listed at the start, though I do wish that I had been able to make the individual interwoven clues easier to solve, or at least to parse. But hey, I did it my way…

Paddock, March 2021