I’d had the idea of doing a Colorado puzzle for quite a while, intending to identify the grid with the rectangular shape of the state, and to use ‘COLOR ADO’ as a highlighting instruction. The problem was that I couldn’t find a famous person or other source of thematic material which related to Colorado, so I put the idea aside for a while. It was only much later that it occurred to me that I had neglected Wyoming, and that I could still make use of Colorado as a neighbouring state. When I discovered Pollock was from Wyoming ideas fell in place very quickly. First, being famous for his drip paintings, I (perhaps naively) thought that I could depict pretty much any streaky random set of colours and represent this as a Pollock (more on this later). Secondly, I soon saw the possibilities of CODY as (a) the basis for a seminal clue (COD=POLLOCK) and (b) the way to indicate the interpretation of the grid as a state. Later I noticed that I could force the solver to show they understood this by having CODA as another entry. This would require, of course, that (across) clues could not be numbered or in clue order, and this put paid to an earlier idea in which I would relate Jackson Pollock’s numeric designations for his paintings to clue numbers (for example Summertime is designated as 9A) .
As a matter of course, I checked ODQ and found the quotation which would allow me to confirm the theme by applying the quotation to some paintings (I understand that this quotation is unfortunately not in some earlier editions). I still wanted some additional way of specifying the Wyoming interpretation of the grid. I couldn’t do this by a misprint message, as these were going to be for instruction and the painting names; after a fair amount of playing around with different ideas, the sequence of paintings culminating in [W]INTERPHONE[Y] emerged as the winning idea.
When I set about creating the grid, I wanted it to be as accurate as possible geographically, and I based the 11×15 grid on the true aspect ratio of Wyoming. I then placed Cody as accurately as possible within the grid. (Rather gratifyingly, I notice that a poster on the AnswerBank message board did check out the cartographic accuracy). I had a bit of flexibility for the length of the painting instruction and consequently the number of down clues. The number of across clues was fixed by the painting names. I wanted to get as many ADO’s in as possible in a reasonably streaky sort of way, but still have a pleasing grid with good unching. In general, when I am constructing a grid, the ideal is to have a Radixean unching model in which the number of unchecked letters is a non-decreasing function of the length of the entry; this has to be balanced with having good words (not too many inflections), no excessively long connected set of bars, and reasonable average entry length. Usually it is not possible to do this with the other thematic constraints (in this case CODY, CODA, INTERPHONE, a fixed number of across entries, and a healthy smattering of A’s D’s, and O’s) , and something has to give. But on this occasion, and after a lot of playing around in the excellent Sympathy package, I came up with a final Radixean grid with a pleasing 7:8 ratio of foreground to backward cells, and with the correct number of clues.
By necessity, across clues could not be numbered, so I gave these in alphabetic order of answer. I made the down clues in standard clue order to make things a bit easier for the solver to get started, but I didn’t put clues numbers in the down clues/entries so as to maintain the aesthetic effect of the painting. I decided on RED/GREEN/BLUE as suitably seasonal colours.
In Mango puzzles (Mango consisting of Steve Mann, Roddy Forman, and myself), we always try to use thematically coherent devices for spelling out messages. In Shackleton puzzles I have taken a different tack, preferring instead to primarily use the ‘misprint in the definition’ device. It’s reasonably tricky for the setter as most of the time just finding a misprinted definition with the right correction is difficult enough; trying to mould it into a sensible and entertaining clue is very time consuming. However, I think that this type of clue is hopefully fair enjoyable as there is the double satisfaction of solving the clue and spotting the misprint, and the misprint is not in some arbitrary place in the clue, but locates the definition. I usually take a very long time to compose such clues, doing it over several weeks, and constantly reviewing ideas, fine tuning clues, discarding old ideas if necessary (never becoming too stuck on a particular idea, however promising ), and trying to make the surfaces as reasonable as possible. Sometimes finding a misprint for a particular answer seems completely intractable, but I’ve found that if you’re patient and throw enough ideas up in the air eventually one of them will land nicely. I don’t generally worry about ease or difficulty of the clue; just try to find the best clue for the word – this will generally yield a good spectrum of difficulties.
The puzzle was originally called ‘Forgery’ (I’m relieved I didn’t stick with that title since, as one person has pointed out, ’fake’ (= pretend piece) is more accurate than ‘forgery’ (=counterfeit). But I had also been toying with using ‘Much Ado About Nothing’ as the title. I had slight reservations about this for two reasons. First that it might increase the possibility of the solver submitting a correct solution without understanding everything (by assuming that ADO from the title was the correct order for RED/GREEN/BLUE). The second reason was that I didn’t want to suggest that I was being disparaging about Pollock’s paintings (in fact one correspondent, tongue-in-cheek, took the title as reassurance to quell his concerns about the somewhat random nature of the final painting). However, in the end, the title seemed too good not to use.
Most of the feedback has been very positive (‘an entertaining load of Pollocks’) with solvers appreciating the sequence of PDMs; some have very been generous in their praise, and I thank them for that. A handful of solvers have objected to the final result as being not very Pollockesque which is a fair enough comment, though I did try to make it as streaky as possible, and perhaps if you look at it from a distance it might be a closer representation. It would have been a more reasonable representation if solvers had been allowed some artistic license to ‘paint’ streaks rather than ‘color’ cells, but that would have created a big problem of interpretation for John Green, and I didn’t raise this possibility with the editors. A final thought here: for Pollock, painting was very much in the action and process of creating the work of art rather than the end result, and hopefully that mirrors most solvers’ experience in creating their Pollock.