hutt's puzzle dump

 Despite having suffered from a cold, I'm getting back into the swing of things. Last week, I picked up the Patchwork project again and made a little more progress in the second part. However, at the moment, I'm focusing most of my free time on something related to algorithmic puzzle solving: I made a particularly interesting independent discovery on Wednesday and I intend to type up an informal paper on the ramifications. I've been burning a lot of my time on solving the 2025 category puzzle, but by osmosis this has inspired me to get back into setting custom Connections. The first puzzle here was set on 2/11/26 as a better version of a puzzle I've sent previously in the blog. The second was made on 2/24/26. Just as a heads up, I've been told that the first is too hard and the second is kinda lame. I recommend trying to solve the second without looking things up, as a large part of its charm is how well the cards mesh with each other. It Just Wasn't in the Card...

The following puzzles were set and shared on 2/18/26.  I especially like how the clues feel local despite being able to influence large portions of the board. https://pzprxs.vercel.app/p?bhaibahan/7/7/q2i56g.q2g2.i2q https://pzprxs.vercel.app/p?bhaibahan/9/9/j1n2w1s59h1h41p2h3h1k1j

The following trio of puzzles were set and shared on 2/17/26, although I thought of the first one a few days earlier. Variant rule: All numbers are wrong. You can verify your solution by enabling "Check multiple errors": every circle should be red and all broken rules should relate to clue numbers. https://pzprxs.vercel.app/p?teri/v:/3/3/369l https://pzprxs.vercel.app/p?kurodoko/v:/5/5/729g9j47g7g54j9g523 https://pzprxs.vercel.app/p?oasis/v:/7/7/2h2h1h2k3h9h5k2h6h2k3h1h5h4