Fond of rumpots, crackpots, and how are you, Mr. Wilson?

I mentioned in yesterday's post that I construct crosswords, which led sandrylene to ask a bunch of really good questions, whose really good adequate answers I'm turning into a post because the difference between conversation and content creation is a "send" button:

At what point do you impose the whole symmetrical structure onto the puzzle? Is the shape something that happens at the beginning? Toward the middle? I guess in general - what is the order of operations for figuring out clues, answers, shape of puzzle, etc?

So the order of operations goes like this:

1) Come up with a theme. Since the theme will mean usually three or more non-optional long entries, everything else will have to be packed around them like putty. Even if you're making a themeless puzzle, you'll want to start with a seed entry, i.e. something that makes you go, "Hey, that'd be really cool to put in a crossword."

2) Make the grid. The grid size will almost always be square or nearly so, standard 15x15 for dailies and either 19x19, 21x21, or 23x23 for Sundays, depending on the market. This is when the symmetry happens: theme answers get added (probably) symmetrically, and then black squares added to make it feasible without lowering the word count any more than necessary. This arrangement might get tweaked or even completely rearranged once you find out that stupid frickin' corner can't be frickin' filled in without frickin' two boring acronyms and an obscure Mesopotamian demigod, but this is where you'll start.

3) Find words to fill the rest of the grid with. You can simulate this process by placing a sheet of graph paper on your desk and then whacking your forehead against repeatedly it for an hour or two. This is officially the Hard Part, and the part that you cannot afford to screw up. The frustration comes from not knowing if you even can fill the grid you've created, and then once you've shown you can, you don't know if you can maybe do better. A lot of yak shaving, is what I'm saying.

4) Find the bit that you goddamn screwed up and spend a half-hour fixing it. While cussing. Cussing helps.

5) Write the clues. Unless you're self-publishing, there'll be one or more editors at the place you're sending it to touch up this bit to match the tone of their other puzzles, but the better clues you write, the more likely they are to buy the thing.

Then also wondering if there are elements that separate difficulty aside from obscureness of clues and size of puzzle? Like, do things with two word answers generally then cause you to aim for a harder overall puzzle? (From solver POV, those are always harder.)

The difficulty level generally comes down to the clues. For the most part you want to use entries that people have heard of, and then write clues to make them hard to answer—having a lot of technical jargon or little-known historical names is the place where fun goes to die. If you have a grid that's filled with things that most people are familiar with, it can become easy, tricky, challenging, or good-lord-kill-me-now just by rewriting the clues.

The question about multi-word phrases in a puzzle is an interesting one. Once upon a time, the idea was to avoid them if possible—only entries that the solver could look up in a reference book if they needed to. Great, except puzzles made under those guidelines are usually boring as fuck. Multi-word entries may be trickier, but a lot of the time they're way more fascinating (compare entries like ELIMINATIONS or CARDIOVASCULAR vs. TWITTERHANDLE or DONTMAKEITWEIRD). I don't do a lot of considering about whether a phrase is harder than a single word, but I do spend time thinking about whether said phrase is interesting enough to make it worth the extra effort.

What are the bits that are hardest about making the puzzles?

See above about the grid fill. There are two considerations here: what words are allowed, and what words are desirable. The former is pretty straightforward, but there's a lot of push and pull on the latter, trying to decide if a really good entry is worth a really bad entry to make it possible. There is software to help, and as a tool it's hugely important, but it's limited by both lack of nuance (for instance, it can't tell if two especially obscure entries are crossing, which is a bad thing) and by whatever dictionary file you're using (it can't use words it doesn't know, so I spend quite a bit of time adding shit like NERDCHIC and BIGMOOD and WHATDIDIJUSTSAY to mine), so in the end you're the one doing the work.

Do you get editorial comments back on submissions, generally, or do things get a sort of binary acceptance/rejection?

It depends on the market, the editor, and how good the submission is. I've often said that the greatest lesson in crossword construction I've ever gotten is in Will Shortz's rejection letters, because he and his staff are very good about at least mentioning both good parts and bad. I usually only hear from Games Magazine when something is accepted, but that's because I used to send them boatloads of stuff and we had an ongoing editorial relationship, so I knew I wasn't getting lost in the shuffle.

Thanks you so much, sandrylene, for the fantastic questions! I'd be glad to answer any other ones that anyone has—for such a public art form, we crossword dorks don't get many opportunities to talk about what we do, especially in comparison to the amount of time we spend thinking about it. (Seriously, I wish there were a damn override button in my brain sometimes. I mainly stopped writing all this up because it was my only hope of getting any meaningful sleep tonight.) Keep 'em coming!

Just a note to let the universe know that I have just put the finishing touches on the grid for the best crossword I have ever constructed. Sunday-sized, good theme, funky and modern grid fill, pangrammatic, could be a little more open and minus a cheater or two, but goddamn am I doing victory laps on this one, especially given what a rat bastard it was to finish. (If you've overheard me cussing under my breath about how much I loathe crossword puzzles any time in the last several weeks, this %@$# was the reason why.) If this thing doesn't sell, I see kerosene and a blowtorch in its future.

Also, a brief addendum to my top 15 songs of 2012 post: I somehow completely forgot about "Salvador Divinorum" by Beaten by Them, off their album Kinder Machines. Damn. I also left off Richard Thompson's "Good Things Happen to Bad People," but that was because I'm saving it up for when the accompanying album, Electric, comes out in 2013 (which was probably a foolish decision, as the song's not only out now but jaw-droppingly good). Give it a listen now. Seriously. Right now. The Internet will wait for you. You'll thank me.

This weekend's Sunday puzzle in the New York Times is by me! *unseemly happysqueebounce activity* It's gotten damned good reviews in the crossword blogs (well, okay, the one), and Will Shortz kept a good number of my clues and only somewhat modified several others, which is always heartening. If you can't get today's paper you can see the thing here; be warned, however, that the page includes the answer. If you're a Times subscriber and use the Across Lite software, you can also download the puzzle for solving here, or pick up the PDF here.

Did I mention the squeeing? Yeah, I don't anticipate this feeling ever going away.

So I admit that I turned in last night perhaps a bit later than I should have, albeit not nearly as late as I've sometimes been known to. Mea maxima culpa. It's not my conscious decisions that I question, though, so much as my subconscious workings. You know, the ones that decide at two in the expletive-deleted-ing morning that it has to figure out how to fix those problematic corners on that crossword that Will Shortz rejected back in February right damn now. Suffice it to say that I was looking at the dead wrong side of 5:30 a.m. before sleeping, which is why today has been another exciting adventure of Narcolepto Boy (and his trusty sidekick Vat o' Caffeine!). At least I got Abbey's birthday present picked up before Skype-time with belgatherial.

Tomorrow night it's dinner with fiddle_dragon and zeyr, both of whom I've not spent time with in eons, and hopefully finally blog-stropping the events of Thanksgiving weekend, but all that will have to wait until I Windex the inside of my brain with a little coma time, commencing about 15 seconds after I finish writing this. Pardon the rush for the door, friends; doesn't mean I've stopped loving you. Good night!