1-two-dimensional-knot

2-daily-high-and-low

3-self-strengthening

4-song-dance

5-polished-stems

6-the-meat-of-the-matter

7-shes-an-astronyaut

8-game-stu-diorama

9-wheres-that-pokemon

10-round-numbers

11-creeeature-feeeature

12-ooh-ooh

13-eddy

14-you-fool-i-have-70-alt-accounts

15-the-walls-are-screaming

16-check-that-right-now

17-hear-and-everywh-ere

18-ooae-oe-oy

19-latin-square

20-sound-cloud

21-markupright

22-the-ipod-mini-mini

23-flip-overtime

24-anatomy-notes

25-mug-shot

26-diatonically-increasing

27-push-for-directions

28-alleles

29-is-this

30-offscreen-thwomp

31-octo-expansion

32-black-mesa

33-tetris-tutorial

34-the-rebus-mini

35-length-x-height

36-illusion-of-choice

37-foreboding-message

38-im-not-skilled-enough-to-make-a-completely-empty-7x7

39-hows-the-weather-up-there

SPOILERS

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

okay let's talk about it!

if you're solving this puzzle, you should quickly notice that the numbering of the clues doesn't match the numbering in the puzzle. if you naively try to answer the clues and place their answers where the current numbering says they should go, you'll quickly find that things don't fit.

that's because the entire grid needs to be changed. the puzzle contains two revealers that both independently describe what must be done to change the grid. those are:

[bit of lag... or a hint for how to fix this puzzle's grid] FRAME DROP

[recovers from a shuffle... or a different but equivalent hint for how to fix this puzzle's grid] SQUARES UP

basically, the border on the outside of the grid (the "frame") needs to be shifted downward ("dropped") relative to the content inside the grid, such that there's extra space at the bottom and less space at the top. when performed, that shift looks like this:

and equivalently, you can also think of it as the black cells in the grid (the "squares") rising (going "up").

you can then compare the numbering before and after and see that the placement of them has changed in such a way that the clues now match.

of course, while solving you won't know to do this, so piecing together what the grid actually looks like is similar to solving a diagramless crossword: a puzzle where you're not given a grid at all and have to assemble the answers to clues yourself into a working arrangement. eventually you'll learn that the original grid is almost right, and then understand what exactly needs to be changed about it.

the final solved grid looks like this! note that now there's an extra row at the bottom of the grid for the final 9-letter spanner STEEL TRAP.

the mechanic of this crossword means that it basically can't be built in crosshare - you would need to be able to write in black cells or write outside the grid to fully express the intent of a solution. and because the puzzle provides two frames of reference for the mechanic (you can drop the frame, or you can move the squares up), i wanted a solving interface that didn't necessarily bias one way or the other.

so i built one! the google sheets solver is mechanically designed to recognize two different solutions to the puzzle: one where the squares are kept constant and the frame is moved, and one where the frame is kept constant and the squares are moved. if you solved with recoloring anything, you would get solution grids that look like this:

the solver works by tracking which of these two grids it thinks you're making and adapting the checker / revealer to match against that one. there's a little table of diffs that you can find in the spoiler zone which does this tracking :)

of course in testing, the frame-drop solution (where the squares are kept in-place) is a much more common approach - it seems easier / less-friction to do compared to shifting the more complex, separated shapes that the squares form. but the solver leaves the door open to both ideas, and that was important to me when setting it up. the crosshare solver, which can only feasibly pick the frame-drop solution to represent, ends up going for that one too.

i had the initial idea to do this crossword over a year ago, but the idea fell firmly into place when i realized that two revealers for the idea were possible. at the beginning i was only thinking in terms of the answer FRAME DROP, and it was hard to justify a full midi-sized puzzle with just that. once i found the extra perspective and elegance of adding SQUARES UP to the puzzle, the idea finally felt complete and the grid became constrained enough that i could fill it reasonably knowing that i had settled on something i liked the most.

the gimmick places some interesting restrictions on the square placement in the grid! there's two factors at play:

the shift makes it so that a row appears at the bottom of the grid and a row also disappears at the top of the grid. that disappearing row can't have a black square in it - otherwise, it would be a bit edge-case-y: does it also disappear with its row? or does it hug the top of the frame as if it's a wall? i wanted to avoid this concern.

additionally, if the grid starts out in a way where obvious conventions of crossword grids were violated (like, if there's a 2-letter entry or an unchecked square in the unmodified grid), it would tip off solvers too early that there's something funky about it. the square arrangement should suggest, in absence of a gimmick, a perfectly normal, non-suspicious setup.

those two factors basically meant that the top of the grid had to be a triple stack of 9s. the bottom of the grid, however, was free to be divided, and i took that opportunity. in the end it produces this interesting play where, after the shift, there's different convention-breaking at the top and the bottom of the grid: three 2-letter entries appear at the top of the grid, and an unchecked square appears at the bottom.

and of course given all of that i'm happy to be able to land two bonus 11-letter entries running vertically :)