It's my first (completed) 21x puzzle! The idea for the puzzle sprouted from the revealer (67A), which was a convenient and enticing length of 21. After dozens of hours of laying out the (absurdly constraining) theme, ~20 versions of fill, and a couple rejection letters from newspapers, the puzzle is complete. In total, this puzzle has been cooking for over a year now, but I'm really proud of the final result! The full story of my process creating this puzzle is beneath the applet, just beware of spoilers.
Also! I printed out 300 color copies of this puzzle for the Puzzle Table at ACPT this year, as a little sneak peek before it got officially posted here. It was pretty awesome watching so many people take a copy, so if you were there and grabbed one, thank you! I hope you enjoyed it. :) I'm pretty sure all the copies were taken by the end of the weekend, so whatever I make for next year, I guess I'll have to bring more...
I enlisted the help of several people during the creation of this puzzle, so thank you to everyone who helped! Especially Will and Spencer for workshopping theme arrangement, and Frisco for test-solving on short notice.
-SPOILERS-
The inspiration of this puzzle came from interpreting ONE IF BY LAND TWO IF BY SEA literally, specifically having one IF "by" something representing land, and two IFs "by" something representing sea. The question,though, was what would represent land and sea. My first idea was to have names of actual lands and seas in the grid, but while the seas are easy (YELLOW, AEGEAN, CASPIAN, etc.), the lands are far harder. What constitutes a "land"? Just names of countries? Continents? Or prefixes for countries that end with -LAND? (GREEN? IRE?? SWITZER???) Besides the issue of a "land" being an ill-defined entity, this execution of the theme would also result in lots of names of seas in the grid, which would just have to be clued as, well, seas. You'd just end up with several clues like [Body of water that...] and stuff. We can do better.
The other obvious presentation of the theme would be to have the letter strings LAND and SEA hidden in longer phrases, which definitely seemed more possible. Initially, I played around with having the IFs as rebuses, but this presented some major challenges.
The main issue is that rebus IFs double the amount of words that need the IF letter string in them. Although it may seem like there would be no shortage of options, using phrases like IFFY, IF SO, EVEN IF, etc. is off-limits, and options for patterns like ??IF and IF?? are practically non-existent. I'm pretty sure this is where the idea lay dormant for a while, until I decided to try the non-rebus execution as a hail mary, which I thought would be harder. Turns out it actually is easier, but only barely.
Very early on, I stumbled across the only layout that seemed to have any synergy. I spent countless hours trying to explore other options, but I found myself coming back to this again and again, hoping that maybe I missed something that would open up a bit more flexibility.
Although this showed signs of life, it proved to be a huge challenge squeezing in four more theme answers (and their IFs!) into the grid. I started by trying to space out the SEA answers as much as possible, since obviously the two neighboring IFs are more constraining than just one. (Weirdly, the layout that eventually ended up working has all three LAND entries in the top half, and all three SEA entries in the bottom half. Funny how that worked out.) I even experimented with vertical theme entries, as seen here:
After many more hours of poking at it, I was finally able to fit all 6 theme answers, plus their IFs, plus the revealer, all horizontally. It's a tight fit, but it's just barely possible. The layout seen in the final grid is really just a fine-tuned version of this, many more iterations down the road.
The plan all along was to submit this puzzle to the New York Times, so my goal was always to get something fillable in 140 words. That turned out to be the one challenge I could not conquer. The final version of this puzzle is actually 144 words, and it's definitely the cleanest version of the grid I was able to produce. The version I submitted to the Times is 142 words, which is as low as I was able to get it. As you can see, the fill is mostly the same, except in the regions around the removed black squares (at the S of ARPS and V of VOLE).
You'll notice no circled cells or colored shading in the Times submission, by the way. That's because my plan was for the theme to involve as little copy-paste by the solver as possible. My thinking was that once you discover what goes in the circles and shaded cells, you can just write in the rest of them, which ruins the fun, especially of the "twist ending" of SCI-FI FILMS. But after some test solves, it was clear that the theme was too easy to miss. Besides, the NYT feedback email said that the theme was "a cute joke, but it felt a little bit too repetitive once solvers understand the gimmick" even without the circles and shading, so I figured it was the right call to make the theme understandable at the expense of difficulty.
The only thing left to do after receiving rejection emails from both the New York Times and the LA Times was post the puzzle here on my blog. But since I wasn't held to any word count limits, I knew that the version I actually liked the most was the 144-word one, which was still missing clues for much of the fill that changed between the versions. That was where this puzzle got hung up the most. Writing clues is easily the part of crossword creation that I find the hardest and least inspiring, so it was really the approach of ACPT that finally motivated me to finish cluing this puzzle.