#161: Dragon of Pearl and Stone

I took part in Puzzlers Club’s Secret Santa again, and this year I even managed to share the puzzle I made here in a timely manner. I tried hard not to overscope again, so this time I decided I would only make a single puzzle, but then this was the puzzle. Oops.

Anyway, I’m going to ramble about the construction process a bit, feel free to scroll down for the actual puzzle.

My santee this year was ADGOD and their wishlist included shading, loops, unusual topologies, dragons and an example puzzle. So that’s exactly what they got. I had the general idea for this puzzle pretty quickly, but I needed to pick the specific rulesets. Originally it was going to be two shading genres with a loop through all remaining cells on each side. But I realised that there would be some inherent non-uniqueness whenever the loop crosses to the other side in a corner (since both possible crossings are equivalent). I considered a few solutions but ultimately decided that the nicest option would be to make one side an actual loop genre with some constraint on the unvisited/shaded cells. To resolve the ambiguous corner crossings, I needed some ruleset that could force a turn or a straight (possibly via segment lengths) to disambiguate these corners and the choice for the loop genre quickly fell on Masyu. I also knew that ADGOD had played a lot of Islands of Insight, so the natural choice for the shading half was Isowatari, which had the added benefit of making both sides use the same type of clue.

The Isowatari rules are extremely strong, especially with the extra requirement of fitting a loop in the unshaded cells. I picked N=5 because I figured it would give me a little more wiggle room for resolving the puzzle near the end. For the Masyu side, the shading constraint was essentially between “shaded groups are rectangular” and “shaded groups are not rectangular” and I just picked the former to see how it goes and never ended up trying the other option. When I first started constructing the puzzle, I placing all clues on both halves symmetrically (or antisymmetrically), and I actually got pretty far without running into any issues. But I also quickly realised that I was mostly looking for any pair of clues that would work instead of placing each clue to create the best possible solve path. So I started over without trying to aim for any visual theme beyond the grid (except for the eyes and no other black circles on the heads).

Construction went very smoothly overall, though I was concerned that the Isowatari just wouldn’t work out at the end. Indeed, before I even got to the middle section, I decided to bash Isowatari half to see how constrained it really was, and it turned out there were already no solutions (all lookaheads failing much further along the grids). Instead of backtracking, I fixed it by making some minor tweaks to the grid shape that wouldn’t be noticeable, making note of a solution that was valid now and just worked towards it. On the flipside, with the rules for this section being so strong, I could now get away with resolving the entire midsection without any Isowatari clues (while still maintaining a fair solve path).

After the first few testsolves, I only made a small tweak where a random black circle on the Masyu side created an unintended break-in that didn’t really lead anywhere, but gave a lot more progress than I would’ve liked. Otherwise, the puzzle worked as intended and seemed to have roughly the difficulty level I was going for.