New York Times, Thursday, February 2, 2017

I submitted this puzzle on September 3rd, 2016 and it was accepted on October 20th (a pretty quick turnaround in my experience). It's my second accepted puzzle at the Times, but the first one was pegged for a day that has a significantly larger backlog (per the most recent inventory I've seen).

My constructing program of choice is Crossword Compiler v9. I use grep/sed/awk (via Cygwin) for word list manipulation and searching. In addition to the word lists I acquired with Crossword Compiler, I also use Matt Ginsberg's list (www.otsys.com/clue) and the one curated by Jeff Chen (the latter also includes the excellent XWI word list). Thanks guys!

As you might guess from the nature of the gimmick, constructing this puzzle required extra front-end work. After laying out the themers and making a tentative grid skeleton, I created a mirror image of the grid with the themers entered backwards to facilitate construction of the portions that use reversed words. As I worked back-and-forth between the two versions of the puzzle, I used temporary black squares to create smaller regions for filling. I also had to do some manual trial-and-error since many of the vertical entries intersect both types of across answers. Likewise, differences in typical vowel and consonant layouts in each direction (forward and backward) posed an additional challenge.

In retrospect, perhaps I should have created a word list that excluded palindromes and words that make other words when reversed. Since I didn't, I had to manually avoid such words when filling the reversed areas (having reversed words that spelled something else in the forward direction felt inelegant). On the other hand, leaving those extra words in gave more options for the rest of the fill, which didn't need to be constrained in that manner.

For further thoughts on this crossword, please check out my constructor notes on the Wordplay blog.

Fantastic debut. Check that — fantastic puzzle, period. I quickly realized something backwardy was going on halfway through, and I enjoy a good backwards puzzle. But I got a little annoyed that random words were reversed. YOU HAVE TO BE CONSISTENT, YOU STUPID CONSTRUCTORS, MYSELF INCLUDED!

Boy, did that slight(-to-gigantic) irritation flip to delight in a big way when the a-ha hit. Alex runs an orderly ship, the first several across entries going left to right as usual. But as soon as you hit BACK TO FRONT, you need to start entering the acrosses ... in BACK TO FRONT order! And then when you get to FORWARD HO, you go ... FORWARD again. Things switch again, appropriately at IN REVERSE, and then flip one last time at LEFT TO RIGHT.

This in itself was a fresh and amusing take on a backwards-type puzzle. But the friendly-sounding cluing made it stand out even further. I didn't get what [watch out now!] and [you can relax ...] meant during my solve, but afterward, it all came across so fun and amusing, like a square dance caller yelling out instructions or something. Colorful, really playful, right on my wavelength.

And the gridwork! Usually backwards puzzles have enough glitches in the matrix that my entertainment level diminishes. It's especially tough to construct something like this, where only certain entries are flipped. I was utterly amazed at how well Alex did with his fill. (I've used the dual-grid trick he described once before with a backwards puzzle — I originally learned it from Patrick Merrell some time ago. It's quite handy!)

There were a couple of toughies in TAMA, COATI, and GERI, but they are all legit(ish). And a minor ERN, that's it? Whoa. Such care to give us a smooth solve is very, very much appreciated.

There weren't a huge number of long bonuses, but RECORD DEAL and DAVINCI were both great, and EPONYM helped out too. (PINKEYE, though … eew.)

Loved this puzzle. It's so rare for me to see something as innovatively fun as this, while still working within all the general rule of one-letter-per-square. Can't wait to see what Alex has planned next.