The New York Times crossword editors reveal their process for evaluating and editing a puzzle submission.
In an effort to shed more light on how we work, The Times is running a series of short posts explaining some of our journalistic practices. Read more from this series here.
Ever wondered how your daily crossword puzzle gets to the pages of The New York Times or to your puzzle app? How do the editors Will Shortz and Joel Fagliano evaluate constructors’ submissions and decide which day of the week it should run? What changes, if any, do they make to the author’s creation?
Wordplay, the crossword column, pulled back the curtain recently in a series of articles in which four successive pairs of Times constructors collaborated on a sample crossword puzzle — focusing on theme, grid design, fill and clues. Below, in a condensed version of the final installment of the series, Mr. Shortz and Mr. Fagliano review the sample puzzle, divulging their editorial process along the way.
To make it easier for you to follow along, the theme here is song synonyms consistently in the second position of two-word phrases: i.e., PUZZLE PIECE, TENURE TRACK, CALL NUMBER and MACBOOK AIR.
The Envelope, Please
WILL SHORTZ: Typically, when we accept a puzzle, we file it for the day of the week for which we think it is most naturally suited. Monday has an easy theme with easy vocabulary. Thursday tends to have the trickiest themes. Tuesday and Wednesday are in the middle. Friday and Saturday are the hardest, and usually themeless.
Then when we come to editing, we select a week’s worth of puzzles at a time, trying to vary the themes and the puzzle makers.
JOEL FAGLIANO: So if Tuesday’s theme has circles, we might avoid a puzzle with circles on Monday and Wednesday. If the theme involves puns, we’d be unlikely to run more than one other puzzle during the same week with puns.
SHORTZ: I think of the Times crossword as being like a three-ring circus. What brings joy and awe is being surprised. I like every day’s puzzle to have a little surprise.
SHORTZ: When I first saw the theme of this puzzle, honestly, I wasn’t sure it was my cup of tea. But I’m liking it better now that I’ve seen the clues that go with it.
FAGLIANO: Yeah, what I think is done well in the theme is the parallelism. The synonyms are all placed as the last words, which allows the solver to have some expectation — O.K., the next theme answer is going to involve some synonym of “song” at the end and another word at the beginning.
Another asset of the theme is that the dictionary meaning of each phrase is pretty far from how it’s clued. For example, TENURE TRACK has nothing to do with a song, which is good.
SHORTZ: In an ideal puzzle, all the key words in the answer are not referred to in the clue. Here, PUZZLE PIECE — clued as “crossword enthusiast’s favorite song” — is still about a PUZZLE, although it’s changed a bit from a jigsaw puzzle, so there’s a little twist there. CALL NUMBER is maybe my favorite theme example, because both key words in the answer are used playfully in the clue: “telemarketer’s favorite song.”
The Grid Design
SHORTZ: So, this grid (designed around the puzzle’s themed answers) has 76 words. The maximum we allow typically in weekday crosswords is 78. It’s nice that the word count here is two under our maximum, which means the answers average a little longer than usual. Something else I like about the grid: It has great flow. No corner or section is isolated from the rest, hanging on by a single square, say. Once you start solving, you can keep moving around the grid. If you get stuck, there are several ways to get unstuck.
SHORTZ: If we like a theme and grid well enough, then we look at the puzzle’s fill (the words, peripheral to the theme, with which the rest of the boxes in the grid have been filled).
We ask for submissions on paper rather than by email because it’s easier for us to examine the whole grid at once, and to mark up the manuscript with pluses, minuses and other comments.
FAGLIANO: We look at all the Across answers first, and then all the Downs, making minus marks for answers we think are subpar, check marks for answers we like, exclamation points for “Wow!,” question marks for things to be looked up and sometimes written comments. When we’re done, this helps us to visualize potential issues: “O.K., there are a lot of minus marks in this one corner — this is an area that needs to be revised.”
SHORTZ: Looking at the Downs … ANTE UP … wasn’t there another answer with UP in the puzzle? Yes, MEET UP. That doesn’t bother me, though. UP is an inconspicuous word.
And OLIVIA MUNN … Well, I’m going to expose my ignorance, but I don’t know who she is.
FAGLIANO: She’s an actress. On “The Newsroom” and other things.
SHORTZ: Hmm, and next to that answer is DANA SCULLY. I do know her, of course, but that could be troublesome for some solvers.
FAGLIANO: Yeah, and there’s a third name to the right of it: SPACEK. In this case, we’d really need to check the crossings to make sure we’re not setting up solvers to get stuck.
SHORTZ: Over all, the fill looks good to me. But I would want to clue the upper right of the grid on the easy side, for solvers who don’t know all those names.
SHORTZ: Often we’ll edit the theme clues first, because those are the most important ones. Then we’ll return to the top.
We ask constructors to send us manuscripts with the clues typed on the left, double spaced, with the answer words on the far right. This is for our convenience when editing. We’ll go through the clues one by one.
The most important thing is accuracy. It doesn’t matter how interesting or clever a clue is if it’s wrong. So anything we aren’t 100 percent certain of, we will verify.
Besides accuracy, we edit clues for the level of difficulty appropriate to the day of the week on which the puzzle will appear; then for colorfulness, freshness, sense of fun.
On average, about half the clues are changed in the editing process. The number can be as low as 5 percent for someone who writes terrific clues, and as high as 95 percent for someone who has a great theme and grid but isn’t necessarily an experienced clue writer.
Brevity is important, too. Partly it’s for reasons of space on the printed page, which is limited. But even online, where space is not a real consideration, it’s nicer to have generally shorter clues. It’s like that old saying from “Hamlet”: “Brevity is the soul of wit.”
On using databases
SHORTZ: There have been some constructors we’ve noticed who have taken most or all of their clues from a database, such as XWord Info. That turns me off. Part of the process of making a great crossword is writing original clues, and it makes for more interesting solving. Some of the clues, of course, may repeat old ones. There are only so many ways to clue certain words. But I’d like constructors to make an honest effort.
On repeating words that are in the grid
FAGLIANO: Our basic rule is that no answer in its entirety should be repeated as a clue, and no clue in its entirety should be repeated as an answer. But if just part of a clue appears as part of an answer, we usually don’t mind.
For example, we would avoid “ice cream” as a clue if ICE was already an answer on its own in the grid. But if a clue said “Eat some ice cream” and there was also ICE SKATING in the grid, that would be fine.
On “say” clues and question mark clues
SHORTZ: Clues shouldn’t have too many waffle words. It’s O.K. to use qualifiers like “perhaps,” “maybe” and “say” once in a while, but if they’re used too much a solver may get frustrated: “Just tell me what the damn thing means!”
FAGLIANO: I would extend that same sentiment to question mark clues. When done well, a clue with a question mark or a joke can brighten a whole corner of a puzzle. But if every single clue is trying to mislead you — even on a Friday or Saturday — that can become annoying. So even in our hardest puzzles, we try to provide plenty of straight definitions for the solver to work with.
On fill-in-the-blank clues
SHORTZ: Fill-in-the-blank clues should be used sparingly, and they should be interesting. “___ circus” (with FLEA as the answer), for instance, really is not an interesting clue.
FAGLIANO: Exactly. FLEA is a perfectly interesting word on its own, so we’d probably want to clue it straight.
On the consistency between clues and answers
SHORTZ: For the answer C’MON (47D), the clue was “You’re pulling my leg, right?” The problem here is that the clue is in the form of a question, while the answer is a statement. I’d rephrase the clue to be in the form of a statement as well.
On brand names
SHORTZ: At 1D, the clue is “Sleeveless Victoria’s Secret purchase, informally.” I have no problem with commercial names in puzzles, but I don’t include them gratuitously. In this case, there’s no strong reason to mention Victoria’s Secret. I’d probably change the clue to “Sleeveless women’s undergarment, informally.”
FAGLIANO: If there was an answer of MURDER, we might clue it as “Group of crows” rather than choosing any of the numerous more grisly ways to clue it.
SHORTZ: Or we could say “Topic for Agatha Christie.” Often putting something in a fictional context can lighten it.
My feeling is that the crossword should reflect everything in life, positive and negative. Not everything in life is peaches and cream. While I wouldn’t want a whole puzzle theme that’s depressing, a single downbeat word here and there is fine.
FAGLIANO: In the case of the word SERF, for instance, there aren’t likely to be any serfs doing our crossword, so it’s hard to imagine someone taking personal offense. That said, crosswords should entertain and uplift — and generally, a light tone is desirable.
If you’re thinking about submitting a puzzle, check out our crossword puzzle submission guidelines.