New York Times, Thursday, August 29, 2013

The original (rejected) version had THE PERFECT STORM at 55a, crossing OF NEWT. Having found a way to thread two EYE extensions (of which there were precious few) through two theme answers apiece, I was sure no other suitable 15-letter entries existed and so stopped looking. I felt fortunate to have those entries and was convinced the puzzle's constraints all but precluded alternatives. There must have been 30 ways to fill the lower middle, none of which was more than marginally acceptable. Every attempt included at least two of the following: TMAC, NOTCHY, OEUF, EFOR, NTS, PUNTO. Discouraged, I began researching again.

As soon as BROOKLYN CYCLONE materialized I thought I might have hit the jackpot. The new entry filled the un-fillable section cleanly and improved half the grid. The rush of such an aha moment during construction, when a seemingly intractable grid or fill problem resolves itself, dwarfs all the happy epiphanies I've experienced while solving.

From a practical standpoint the hardest parts to fill were where entries ran in only one wrong direction and crossed three themers (the ROLE PLAYS/LEERING AT sections), and the transition areas where abutting entries ran in opposite directions. These sections required spacial gymnastics that are difficult to describe but involved rotations about axes and bizarre, walled-in grid patterns. (I'm hopeless at filling by hand). Strangely enough, finishing the NE corner was a simple matter of filling a modified SW corner and reverse-transposing the fill back to the NE.

Thank you for reading and solving!

A longtime solver told me recently that his all-time most-hated New York Times crossword was the one by Elizabeth Long on April 1, 2011, in which all the vertical answers on the right side of the grid ran upward. Of course, at the time he solved it, he didn't realize it was an April Fool's puzzle. I'm thinking he's going to hate this puzzle by Timothy Polin even more.

A construction feat! These days many (most?) constructors use software, either Crossword Compiler or Crossfire, which helps automate the process and eliminate symmetry errors. I personally use a word list I've built up, and the "auto-fill" feature helps me determine if the black square pattern I've arranged will be fillable or not (I don't use auto-fills though, since I think step-by-step filling gives me better results). But a construction like today's requires graph paper and pencil, specialty software written to handle the crazy constraints, or ways of tricking software to do what you want. Impressive what Tim's pulled off!

I enjoyed this puzzle, but even knowing many of the answers in advance (I had to fix up about half the grid answers for the xwordinfo database) it was still a bear to solve. Note that the puzzle is not totally symmetrical, in that the answers are reversed only above row 7 and right of column 9. It's a WITT of a puzzle (Wish I'd Thought of That), but that asymmetry felt inelegant. I likely would have given it the POW! (Puzzle of the Week) if exactly half the answers had been reversed.

I don't envy Will for the mail he's going to get on this one. Devious. =]