New York Times, Saturday, September 15, 2018

This puzzle is one of the success stories of the editing process, so I owe extra thanks to Will for this one, for having both high standards and patience with constructors.

Back in 2015, I'd built a puzzle around the same pair of 15-letter seeds (which may speak a little to their pairing: I was starting to write this puzzle just as Obergefell v. Hodges was being decided — plus I'm a computer science faculty member, so I was particularly drawn to the second of the two long entries). I had managed to include a bunch of other longer entries that I liked a lot (SPARKPLUG, HARPER LEE, ITS SO EASY, ...), but, as the rejection note from Will and Joel said, there were too many ugly short entries that "start to add up after a while to make for unpleasant solving." That's always a tendency that I have to fight when I'm filling puzzles: I tend to overvalue what I consider to be awesome long entries and fail to pay enough attention to the costly short junk that they induce.

I got similar feedback from some other editors at other venues, and so, begrudgingly, I put the puzzle away in my (overstuffed) filing drawer of rejections.

But I kept finding myself compelled by this pair of 15s, and so I pulled it back out a couple of years later. In short, I moved a couple of black squares around, refilled again from scratch, reclued it all, and submitted a new version. A good portion of the NW corner (my favorite portion of the 2015 version) stayed intact from attempt #1 to attempt #2, and OVINES survived (but migrated in the grid); save those pieces and the 15s, everything else was brand new.

And here's the result. Hope you enjoy it — and, even if you don't, I hope you appreciate that the version you didn't see was a lot worse!

How much I love a Saturday crossword depends on a couple of factors largely outside the control of the constructor or editor. The first criterion is whether I can finish, chiefly affected by the overlap of things the puzzle thinks are worth knowing and things I happen to know. The second is how amused I am by the clues, which says more about my sense of humor than anything else.

What makes today's puzzles satisfying (so, yes, I finished and with an amused smile) is that several common words get nicely diverting clues that at first seem impossible but nicely click into place. I know nothing about college football in 1890, but with a couple of letters in place, "Something STATE" seemed likely, leaving four letters to fill in. Utah or Iowa feels less probable than OHIO. I know little about Islam, but the first letter J pointed right away to JERUSALEM. I speak no Farsi, but STAN seems obvious. I don't play Risk, but SIAM fell easily. That's three fun facts I learned from one puzzle.

SELF-DRIVING CARS is a topical entry. "M" or "Z" is an impressive clue for MOVIE. The Edison clues are fun. Senate coverage isn't DOME at all, but TOGA. ASPS are cousins of garter snakes.

The equation x + 0 = x is indeed an AXIOM. Mathematics (also science, technology, etc.) depends on accepting such things on faith. If one of the famed Peano axioms is found to be untrue, your iPhone will suddenly stop working. If you have a Galaxy, it will catch fire.

When I started doing themelesses, it was all about the entries. A couple of sizzling ones like SELF DRIVING CARS were what drove (see what I did there) my final impression.

These days, it's more and more the cleverness of the clues. I came away entertained today, even though there weren't that many entries I could point out as marquee — just PIPSQUEAK, SAME-SEX MARRIAGE, CLEAN SLATE. It was all about the great clues, showing off such wickedly brilliant wordplay. My favorite was "reproductive system" repurposed for MIMEOGRAPH.

Themelesses built around grid-spanning entries typically don't have much real estate left over for much else. And it's usually difficult to convert the limited number of longish slots into great entries, due to rigid grid inflexibility. TAKE STEPS isn't going to win any awards, for example. It's so critical that your grid-spanners seriously sparkle, since they're likely among the few that will.

I've heard some constructors muse about creating themelesses with zero feature entries that shine on their own, but which allow for fantastically clever clues. I wouldn't go that far — I wouldn't want a themeless loaded with TAKE STEPS or even MIMEOGRAPHs, for example — but I can understand the notion.