New York Times, Friday, January 22, 2021

Themeless puzzles are typically judged on how "fresh" the fill is, how much the entries excite the solver. I wrote this puzzle with the goal of trying to maximize the amount of fresh fill. That may sound like an obvious goal, but most themeless puzzles actually don't try to do that. Often, constructors start with a few "seeds", entries that they really love, and then go from there. Basically, it's a question of quality vs. quantity, and most constructors put the emphasis on the former. But for this puzzle, I wanted to go for the other extreme.

I knew of an excellent Times puzzle by David Steinberg that seemed like a good place to start, as he managed to pack in an amazing amount of fresh fill. I decided to try to see what I could do with the same grid pattern. It's one of the easiest patterns of black squares you can imagine, primarily because of the lack of long words. That's a mixed blessing because the long answers are usually where most of the sparkle in a themeless can be found. You really have to make the most of the short fill. Even then, some solvers might still be disappointed with the lack of marquee entries.

Before I began constructing, there were some changes that needed to be made to my word list and my construction program. First, I looked through entries in my word list that had been marked as fresh and noted the ones that I liked especially. Because of the grid pattern, I only had to worry about entries that were 7 letters or fewer, which made the task much more manageable. Then I told my program to try to maximize the number of those entries. Finally, due to the super-symmetry of the grid pattern, dupes were an even bigger problem than usual, so I wrote some code which mostly eliminated them.

After all the hours I put in to get to that stage, the actual construction was a breeze. One coincidence I found amusing was that I had two Apple products in the grid (iPad Pro and Mac) plus a company (Mott's) known for its apple products. You might notice that Mac is no longer in the puzzle, as the original crossing MCU/MAC was changed to MSU/MAS. I've watched a lot of Marvel movies, so MCU was actually on my list of fresh fill. It just goes to show that tastes differ.

There remains one aspect of crossword constructing that really can't be automated: cluing. It's an area I've always struggled with, and this puzzle was no exception. In fact, when it was accepted, the Times team requested that I reclue it. I think my second set of clues was an improvement, but as always the final version is considerably better than anything I could come up with, thanks to the editorial team.

Notoriously difficult themeless layout, featuring 7-letter entries — 36(!) of them. It's not a difficult grid to fill in, but doing so with color is another story. Looking at the whole set of possible 7-letter entries, the ratio of sizzling ones (two-worders like LIP RING or evocative one-worders like GEYSERS) to neutral ones (DISMAYS, LARAMIE) is low — I'd estimate roughly 1 sizzler for every 4 neutrals. It's tough to avoid so much blah filler.

(8+ letter entries have more possibilities in forming multi-word phrases, so you might get 1 sizzler for every 3 neutrals. Much easier to select colorful entries!)

Given these technical issues, an average grid might have only 20% of its 7-letter slots converted to great material. And given the competition in themelesses these days — especially in 72 worders, the easiest of all themeless tasks — I'm not impressed unless that figure is upwards of 40-50%.

That's all a long-winded way of explaining why the constructor in me enjoyed this one so much. BALL HOG, ER NURSE, GET LOST, HASHTAG, HOME GYM, IPAD PRO, the list went on and on. Well over 50% — bravo! All the word list groundwork clearly paid dividends.

The solver in me enjoyed the cluing even more. I've been critical of Daniel's puzzles, some of which have felt drab in their cluing. Much improvement today, for example, OPEN BAR a place for "free spirits." TRIVETS as "Pot supporters," too — neither needing a giveaway question mark! Excellent work.

My one big hesitation before giving this the POW!, though: the AGONIST / ARGOT region. The gridwork isn't the problem, but the cluing made the area near unsolvable. I'd much rather get a dictionary-ish biochem clue for AGONIST (it's a commonplace word in the pharma R&D industry). Given the region's toughness, you have to clue ARGOT in an easier way than [Cant].

I spent so many frustrating minutes feeling like Daniel was my antAGONIST that I initially disqualified this puzzle for POW! consideration. After a day of reflection, though, I decided that was petty.

All in all, careful gridwork selecting for color and a big step up in cluing fun earn Daniel his first POW!