New York Times, Thursday, July 30, 2020

This puzzle's publication is gratifying, because it seemed close to impossible to make. I had the idea for it back in 2015, but only got about a third of the way down before realizing how insane the constraints were going to be.

Over the years, when scrolling through my files for half-baked ideas, I'd inevitably take another stab at filling this, only to give up after a little progress. Five versions (and five years) later, I'd made it two-thirds of the way down the grid, before a final flurry of inspiration in quarantine gave me a miraculous fill for the bottom section that even included a revealer (TWO TO ONE).

I'm sure astute solvers will notice some indulgences with the two-word phrases down there (I'm looking at you, GOOD TONE and LAND USERS), but hopefully they'll permit me a few stretchy ones.

Overall, while I'm not normally one to make a stunt crossword like this, it's fun to push the boundaries of the art form every now and then. Next time, though, I'll probably just ask someone with computer chops to write a script and save myself the five years!

I've been trying to construct something like this for ages — my concept is that all entries in a row could take the same clue — but every one of my attempts has been a miserable failure, the severe constraints making me gnash my teeth. I'm awed by Joel's success, almost every pair of Acrosses working to make a fine phrase. (The two Joel pointed out were the ones on my mind too!)

Even better, he worked in not just neutral phrases, but some excellent ones: INDIA INK, BEES KNEES, SKI SLOPE, WAIT A SEC, NEST EGGS. I love that Joel didn't say "good enough" and settle on only neutral pairings.

Best yet, the Downs didn't suffer much! Imagine how difficult it would be to figure out working pairs of Acrosses, all the while making sure the Downs are tractable. It's like some Sisyphean, Greek-myth punishment that Crucivera might hand down upon wicked crossword constructors.

Finally, to wrap everything up with TWO / TO ONE? Such a fresh product.

As a solver, I don't enjoy cross-referencing at all. I'm not sure I would have finished this one, if I wasn't so OCD. Jumping back and forth throughout the entire (electronic) solve injected too much annoyance, dragging down my astonishment of Joel's amazing creation. I fear that many solvers will have a similar reaction.

I'm hoping that the NYT app finds a way to circumvent this. It's a great example of a puzzle that could showcase the modernity of app-based solving — each pair of entries in the grid automatically treated as a single unit, so the solver doesn't have to manually move the cursor back and forth within that pair. For instance, if you were at the I of IRON and pushed the left arrow, the cursor would automatically move to the P of SCRAP (like it would if SCRAP IRON were one entry), instead of the solver having to manually move to it.

I went back and studied Joel's gridwork, a master class in walking that delicate balance between segmenting a grid for construction ease while still achieving adequate solving flow. Even then, I can only imagine how much trial and error was involved in working and reworking each of the tiny subsections.

Memorable puzzle that the constructor in me immensely admired.