New York Times, Thursday, April 18, 2019

I'm curious to see how this puzzle will be presented in print and in the NYT app. I shaded CATERPILLAR gray in my submission, and I hope this feature will be preserved. It's visually appealing for two reasons.

First, the final grid art looks better with a head/body (in gray) to go with the wings (depicted by the connect-the-dots feature). Second, the gray CATERPILLAR also resembles a caterpillar, making its transformation into a butterfly more remarkable.

Unusually, there's almost nothing in common between my first submission and the final grid. Originally, I had CATERPILLAR in the same central vertical slot, but I used right/left symmetry for the grid, which made the pattern on the wings symmetric and much prettier. CATERPILLAR's bottom R intersected METAMORPHOSIS.

Unfortunately, a central vertical theme entry in a puzzle with left-right symmetry requires using either (1) two flanking vertical entries at least as long as the central entry (which both strains the fill and makes it unclear what's thematic and what's not); or (2) a one-letter word somewhere in the middle of the puzzle. I chose the latter.

Given the aesthetic considerations at play. I hoped for an exception to the NYT's prohibition on words less than three letters long. Unfortunately, Will and crew didn't go for:

1. "I" clued as a one-letter word,
2. the consecutively-presented phrase YES/I/CAN, or
3. checking the singleton letter by making it part of the alphabetical list of circled letters.

Reluctantly, I abandoned my dreams of left-right symmetry and looked for options using standard rotational symmetry. After some brainstorming, I found the final configuration of BUTTERFLY and CHRYSALIS symmetrically intersecting CATERPILLAR. Sometimes the crossword gods are kind.

One final note: I'm super bummed that my clue for PANDEMIC didn't survive the editorial process. I clued it as the [best-selling cooperative board game]. If you're into games and looking for something new, I highly recommend it.

Liz Gorski used to do so many beautiful "connect the dots" visual puzzles that Jim wrote a JavaScript function to automatically connect the circles in alphabetical order and animate the fades: it's called "GorskiDraw." Take a moment and enjoy the pretty picture below.

(Liz has stopped submitting to the NYT, instead focusing on her successful subscription service. Please consider subscribing! It's great to see her entrepreneurship pay off. Makes me consider going that route, too.)

As a solver, I don't usually appreciate fortuitous themer interconnect — at least, I usually appreciate it less than the constructor. But today, I had a moment of admiration when I realized how nicely BUTTERFLY and CHRYSALIS interlocked with CATERPILLAR.

Jim pointed out afterward though that the ordering was odd. Why BUTTERYFLY -> CATERPILLAR -> CHRYSALIS? What are we, Benjamin Button-terfly?

It's a great observation. Hitting BUTTERFLY so early in my solve gave away the game, too.

This could have been fixed by "flipping" the grid on the NW to SE diagonal, so that CATERPILLAR ran horizontally across the middle, with CHRYSALIS in the NE corner (vertically), and BUTTERFLY in the SW corner (also vertically). That would have been perfect.

Connect the dots puzzles are so tough to construct. I had a tough time working with one that didn't have to be symmetrical, and that nearly broke me.

I imagine Alex tweaked his positions over and over, with one side working out but the other not cooperating. Unfortunately, a lop-sided BUTTERFLY … well, it just wouldn't fly, would it? Alex also had the luxury of starting / ending the pattern anywhere — and going either clockwise or counterclockwise — but still, it's no joke.

Overall, the grid is decent. A bit of toughness in entries like AVANTI, BONTON, PINKOES (not PINKOS?), but that's expected in corners as big as the NW / SE. I might have considered putting a black square at the E of ÉCLAIR to facilitate filling, but that would have cut that corner off but good. So I think he made good decisions all around.

It's not Gorski. But it is Gorski-esque.

There's a technical reason why Will Shortz would choose not to shade the central CATERPILLAR as Mr. Eaton-Salners wanted, at least for digital distribution, and it has to do with software limitations.

Shaded squares aren't supported by Across Lite at all. When the NYT uses them, circles are encoded in the grid, and publishers get notified that those circles should be interpreted as shades. That breaks down when grids have both circles and shaded squares.

XWord Info has no such limitation, so we added the shaded squares here to realize the constructor's intention.