Hi folks,
As promised, some info on how I set the Sudoku crossword puzzle smash.
This was a puzzle which started out from an 'I wonder if...?' moment, having realised that a 9x9 Sudoku with one row and column between each of the panels, plus a one-cell border round the edge, would be a 13x13 square. I found a random Sudoku online and used that to fill in the OTTFFSSENs, then used a Listener grid from a puzzle a while back to give me a sensible set of bars, and started trying to fill it. I knew I wanted to have some fully barred-off squares to make the Sudoku element more interesting, so I popped them in quite early on. Otherwise it was then a case of fiddle/try filling/repeat for a bit.
The grid-fill was done with the wonderful QXW. I used my standard 50,000 word dictionary but it quickly became apparent that that wouldn't be enough so I expanded it to a fuller word list. I would dearly have loved to get SKELETON STAFF in there (being the only 13-letter phrase I could find which contains OTTFFSSEN) but it didn't work out. QXW has wonderful permutation options, including making the word entered normally, backwards, cycled, cycled backwards and jumbled; without this I wouldn't have been able to get things sorted. This actually wasn't a bad one to code in QXW; I've had much harder (including an eightsome reels in 3D -
https://3dcalendarpuzzles.co.uk/wp-content/uploads/2024-11-NOV-PUZ-SOUP-AGC-futxyjOu.pdf - and a 3D alphabetical jigsaw (with rhyming couplet clues, natch) -
https://3dcalendarpuzzles.co.uk/wp-content/uploads/2023-11-NOV-PUZ-SOUP-qV0qZ19M.pdf).
Then to testing. Dear me, the testing. The trouble with this layout is that if you have pairs of letters at the corners of a rectangle which is otherwise undivided by bars (eg FS at the top corners and SF at the bottom corners) they can be switched to give a valid solution to the sudoku, and it's SO easy to miss one. And as soon as any barring is changed then the whole thing needs to be resolved from scratch to see if it's introduced any other anomalies. So... lots and lots and lots and lots of starting from a blank grid and seeing if it worked and getting it wrong. My thanks to Andie Johnson and John Henderson for being early victims of it not working properly.
And then, eventually, it all fell out. One final word tweak for a rectangle I'd not noticed, and it was done. I set it three years ago, sat on it for a while, and eventually decided to try the Listener. Thanks to the editors for having me, and for their careful and thoughtful amending of some clues to rein in some of my wilder flights of fancy.
I am personally absolutely delighted with this puzzle. It's an example of my favourite type of puzzle: where you know from the start exactly what you have to do, but it looks impossible, and then you realise that it isn't and you're going to have to do it because you can't let the damn thing beat you. (See also my vowelless puzzle,
https://uploads.guim.co.uk/2019/10/31/Genius147_v3.pdf, which I think is my favourite ever of my own puzzles - the simplicity of the instructions, ' ll vwls hv bn rmvd frm bth cls nd sltns' sits in counterpoint to the difficulty of the puzzle - and the glorious, glorious moment of realisation for Simon Anthony, as he understands that the Miracle Sudoku isn't a hoax:
https://www.youtube.com/watch?v=yKf9aUIxdb4 .)
I do hope you enjoyed solving it - the comments are positive, for which many thanks, and I am quietly very proud that some of you think it might be up there for your puzzle of the year - that means a lot. I look forward to annoying you further in future.
Hamish/Smidgers/Soup
PS I'm also very open to people getting in touch with me if they have something they'd like coded in QXW - I am getting better at using it and understand how some constraints work, and find filling grids like this really very fun.