Each of the four 2x2 blank "cornerstones" can be completed with a bi-gram that creates a new 5-letter word. The clues for the 3-letter words don't necessarily also clue the corresponding 5-letter word, but they at least mention it or refer to it in some way. When the "cornerstones" are placed in their respective corners of the internal 5x5 grid, a "magic" word square (whose entries read the same across and down) begins to form. The only missing entry is the central one, and the solver is left to deduce that LEGOS fits in that spot and is the meta answer. I knew there were some other words (ABUTS, STALE) that could fit in the word square, but I did not account for STILE, which could yield LEVIS. That does not answer the prompt as well, but I was bummed to see another solution to the word square was possible.
Sometimes, I get ideas for metas from things that I try that don't work on other puzzles. This puzzle was born from failed attempts to solve "Build Your Brand" in the WSJ recently. I spent most of that weekend trying to "build" my "brand" by taking blocks of letters of various shapes and sizes from different parts of the completed grid and combining them to try to make a brand name. Don't ask me how I got stuck on that path, but I was 100% sure it was right and that the answer would end up being LINCOLN LOGS or LEGO or ERECTOR SET or something. I eventually did break out of that and end up solving ADOBE, but before that weekend was over, I had started work on this puzzle.
I lost track of the number of iterations this went through before this final version. Went through about 4 different word squares. Originally, the 5x5 box was separate from the grid. The 2x2 squares were not in the corners, but scattered throughout the grid. As it evolved, they moved to the corners, but just as regular entries in a square-shaped grid. In that version, the clues for the "cornerstone" words tried to double-clue the words in the word square. I'm still not convinced that I ended up with the best possible execution of this idea, but people seemed to have fun with it, which was the idea. It's supposed to be a fun one, not necessarily a hard one.
Thanks to everyone who solved. Quick count of PMs looks like about 32 correct entries.
New MMM tomorrow!
