I like a meta-puzzle. I think any set of individual puzzles can be immediately improved by the addition of an over-arching meta-puzzle that links them together. This requires co-ordination, but a group of editors can manage that. So the annual Treasure Hunt is our contribution to meta-puzzling, and I’m keen to see it continue.

I’ve always thought that finding the end of the Treasure Hunt should be every solver’s aim, and I slightly regret the fact that it doesn’t count towards the annual statistics. The result is that lots of solvers seem to ignore it, which I think is a pity. This year’s puzzle was designed to be easier than usual, with a view to encouraging more entries.

XKCD is a web comic which is occasionally brilliant. I think the Hofstadter joke is a particularly good one, and I thought that it would make an ideal solution for a very meta Treasure Hunt. It just required each setter to hand out one letter, and the number puzzle to hand out one number. Then I could write a puzzle that finishes with HOFSTADTER, and we would have the location, the name, and the number of the cartoon. The answer would be the self-referential acronym. Simple.

I just needed to come up with an idea that would be meta enough for the theme. Godel, Escher, Bach seemed like the obvious place to start, though maybe I Am a Strange Loop would have been more appropriate, had I read it. To be honest, I haven’t read all of Godel, Escher, Bach, but I get the impression that Hofstadter talks about self-referential or recursive things, and I wanted to capture some of that. I’m a programmer, so I hoped that I could produce some sort of self-modifying computer program, which could be applied to words, and which people could readily code if they wanted to. The problem was coming up with something that can be explained unambiguously in a paragraph.

In the system I came up with, each letter can be thought of as an instruction, which has an effect on all the future instructions. This is self-referential enough, and (fortunately) it turns out that the code maps all the letters of a 25-letter alphabet to the same alphabet in a one-to-one relationship. My son tells me that this is called bijective. Jason says it would be better termed cyclic. This is set theory terminology so I don’t understand it. I was pleased to discover that reversing the code is straightforward, and surprised to discover that running it 500 times always produces the original input, regardless of length. Nobody has yet explained why this is, but apparently it would be simple to prove if any mathematician wants to try.

I was aware that running the code was not all that fun, which counts against the puzzle. Error-prone and admin-heavy rote procedures aren’t the most popular forms of solving entertainment. Maybe it’s one of those cases where the exercise of writing the program is actually more rewarding than running it by hand, though the standard solving approach was manual. Certainly very few people, if any, implemented the inverse of the code, which would have significantly eased a few of the clues.

AJ was responsible for the subtle IS META acronym in the puzzle instructions, and AJ is normally responsible for the much more sophisticated Treasure Hunt puzzles. Now you’ve had a chance to see how rewarding these Treasure Hunts are, hopefully you’ll be inspired to give it a go next year when we’re back to normal difficulty.

Posted on Thursday, January 9th, 2025 at 5:35 pm under Setting the scene.
Leave a response