Listener No 4348, Quads: A Setter’s Blog by Shark
Posted by Listen With Others on 21 June 2015
I have had a number of requests for a setter’s blog, particularly several intrigued to “know how you approached the grid construction”. The problem is that after a few years I have forgotten exactly how I managed to construct this puzzle, but I won’t forget the blood and sweat produced to get there – that will be firmly etched on my memory. Just glimpse at the spreadsheet that I created whilst attempting to set the clues trying to make sure I didn’t make a wrong move (even that wasn’t foolproof).
What I can tell you about the construction is that there were no fancy software programs used in this puzzle; I am not that computer savvy having no programming experience at all. The idea for the puzzle came out of nowhere (probably at awe solving the amazing grid transformations produced by Kea to win his Ascot Gold Cups): but essentially I wondered if I could rotate a quadrant of the grid to a different place creating real words. At first I thought I could swap quadrants, but gave up on that idea as the puzzle would be over too quickly (those of you that have solved Shark puzzles are often faced with convoluted endgames). I worked on the upper left-hand corner and ANIMAL to LAMINA with MADRID to DIRDAM immediately came to mind. Because of rotational symmetry, whatever length went in one quadrant had to be the same in the other. So using six letter entries meant I didn’t have to worry about creating any further new words. I also knew I couldn’t use too many short words, despite their being easier to find, because the average word length would end up being too low. I noticed that ALLEMANDE would work with MADRID and that forced the bottom right-hand corner to be SQUAMELLA. ECARTES and RETRACE followed and it wasn’t long before I had most of my rotating quadrant filled.
Starting to add bars to the remainder of the grid made me realise that there were going to be around 40 entries and, if I could make it such there were 36 numbers, I could use these letters to refill the blank space. Searching through my files I have found an early spreadsheet, but interestingly I have no recollection at all of doing it! Some of the letters were already fixed because of the upper-left hand corner as well as a few crossing entries, so I must have created a spreadsheet to identify what letters were forced before filling the rest of the grid.
I refilled the top-left hand corner at the same time as I filled the rest of the grid and entered the bottom right-hand corner last as that had the most leeway. There are not a lot of 9-letter words that end with NDE and so using DEMIMONDE forced an M in the same place. A few more words in the refill were starting to be created and several “A”’s were appearing as duplicates. This was very fitting as I intended to make the solver refill the grid as a setting exercise and so ACADEMIA would make a perfect finale to the puzzle. As I went along I was saving the files as “quad” but ended up calling the puzzle “setting school” (see more on that later).
However, would it be fair to the solver to refill it?
There were two nagging doubts. Firstly, was the refill unique and secondly, could solvers actually refill it without losing their sanity. I asked Chalicea to have a go. So she sat for days, literally days, with 36 scrabble tiles and the blank corner of the grid with only the crossing partial entries as shown.
At the same time I heard from Nod that Elap was a whizz with programming and he might be able to tell me if my solution was unique. He created a programme and left it running, working through 100 million possible fills for the above corner with the set letters, and came up with 5 solutions. However each of them had a word re-used from the original (apart from icterus, which was the only entry I was forced to re-use). As I had used ELEA, a proper noun (even though it is in Chambers under Eleatic), this proved that there was no fill using real words. There may well be another possible refill using proper nouns, but I had no way of proving that. Given that Chalicea was still struggling, made up my mind that I had to give the solution fill somewhere in the clues.
On several occasions I have used cluing gimmicks that relate somehow to the theme. This time, as I needed so many letters to get a workable instruction, meaning two from each clue, I used movable definitions. Now I also had to include the refill letters somehow (originally I called them ‘staggered letters’ until Shane suggested I change this to ‘successive letters’). It was Artix who stated after test solving the puzzle, that what I had achieved with the 36 letters in the grid would slip by several solvers. He suggested I should take out the last clue’s hidden letter especially as that letter corresponds to an unch in the grid. I therefore re-clued it with the question mark at position 36. The 6 un-thematic clues coincided with the 6 entries starting from the same 6 cells in the grid as thematic entries and so these were used to guide the solver to the re-fill shortcut. From reading the feedback from JEG, I think the fact the 36 letters are the starting letters in the grid (and therefore the 6 un-thematic clues coincide with these) has slipped by several solvers.
Artix also came up with the adjusted title changing it from “setting school” to “quads” as it tied in with the process of the movements of quadrants and ACADEMIA (given that quads or quadrangles are seen in places of academia). I kicked myself as I had called it this all along, but was fixated on the fact that I wanted this to be a setting exercise for the solver. That aspect didn’t matter anymore.
Finally, I must thank all my test solvers who battled through the puzzle, giving much needed preamble suggestions. The preamble, as I have often found when setting and solving, can make or break a puzzle. Noticing some of the comments that have arrived from JEG, many solvers found this tricky, but rest assured, hours of consideration went into every aspect of this puzzle to make it as fair as I could make it. This took almost a year of setting, tweaking, vetting, adjusting and then over two years before further editorial changes were required, including several clue shortenings due to lack of newspaper space, so I am pleased this puzzle has seen the light of day and has gone down well with all the solvers who gave it a go.