A Conversation Between Cruciverbalists: Michael Griffith Interviews Natan Last About Across the Universe

Fiction Editor Michael Griffith: Natan Last’s Across the Universe: The Past, Present, and Future of the Crossword Puzzle, just out from Pantheon, is a brilliant, exuberant, witty book that hopscotches from the initial crossword craze of the 1920s to the present. Along the way Last profiles innovators, traces the puzzle’s complicated evolution, explores the ways technology has altered and is altering crosswords for constructors and for solvers, celebrates ingenious puzzle-makers and the way they’ve broken the bounds of the grid and thus created new possibilities. Last sees the crossword as a kind of Schrödinger’s Cat: simultaneously a morning’s brief, light diversion and a site where complex cultural politics are contested; simultaneously a closed system with a single solution and a multiverse of possibility and ambiguity. In the grid’s intersections, “low” and “high” cross and mingle; the math of the grid and the poetry of the fill tweak and jostle each other. The crossword is lively play in deadly earnest.              

Natan Last is a policy analyst, writer, researcher, and a prolific and gifted cruciverbalist whose puzzles have often appeared in The New York Times, The New Yorker, and many other venues. As I was reading the book, I realized that he is the author, too, of some of my favorite clues (“Collapsed red giant?” for USSR? Chef’s kiss) and some of my favorite juxtapositions (Tone Loc and Gertrude Stein!). I was grateful for the chance to sit down and talk to him about crosswords and about his terrific book.

Michael Griffith: I started doing crosswords in the mid-’80s, and at that time there was an atmosphere of gloom and doom around them, part of which was generational: I was a young person playing what some friends considered a stuffy old person’s game. At the time, one heard a lot of skepticism about the future of the form. One of the many things that struck me about your book is that in telling the history of the last forty or so years of the puzzle, you make a compelling case for optimism. I wondered whether you could expand on that, talk a bit about why you feel hopeful about the long-term endurance of the crossword.

Natan Last: I was in the same boat as you a decade or so later. I went to high school in New York City, where we got the Times delivered to AP US History class, and in our free period we’d cluster around it, friends and I, and be forced to collectivize our knowledge just to get past a Tuesday. I think subconsciously, we knew we were participating in a ritual of passing, trying to seem erudite and cosmopolitan and New Yorkish. I feel the same way every time I play bridge or any old trick-taking card game. These are rituals from the past that have a sophistication to them.

I immediately wanted to throw my voice and try my hand at making puzzles. Not everyone did; I was, for a long time, one of the few young people in the game. That changes when the Times takes its app in-house in 2014: now there’s an army of iPhone-armed young people who did not grow up, as I did, solving on paper, and who start to develop really strong opinions about the puzzle.

Simultaneously, there’s a nascent blog scene, and there are people frustrated with the long lag time from submission to acceptance to publication. I remember the morning after Sonia Sotomayor’s Senate confirmation in 2009, her name appeared in a Brendan Emmett Quigley blog puzzle—illuminating the way these blogs would not only respond to current events but, because putting a name in a crossword canonizes it in some way, also have a hand in shaping common knowledge.

And then COVID is this enormous accelerant: It’s a moment in which that swath of digital-native solvers is now at home, solving way more, and finally thinking, Well, hey, maybe I could make this, right? You get this proliferation of more blogs, you get online discussion spaces like the Crossword Discord, you get much more communal puzzle-making–group chats in which theme ideas are workshopped, the rise of the multiple byline. Alongside that, you get this explosion in revenue generation from the Times app, and that means that post-COVID, there’s an expansion of puzzle offerings at places like Vox, The Atlantic, Boston Globe, Bloomberg. All have added or expanded digital games offerings since the pandemic. So those younger people find themselves entering a totally renovated economic landscape.

For those reasons, I’m optimistic about the future of puzzles . . .  with a sort of hyper-capitalist asterisk: all the incentives of the habituating, casino-like app, I think, are here to stay, for better and in many obvious ways for worse. That’s part of the reason the Times can be so successful, an interest in more subscribers and an economic mandate encouraging accessibility.

MG: Another kind of accessibility you discuss in the book is the way that puzzle fill has changed over time. I came from the opposite direction: You were a New Yorker trying to own your New Yorkness; I was a hick from South Carolina, trying to get access to urban and urbane sophistication. The Times puzzle in particular was an entrée into a world of sophistication and, as you said, of erudition. I didn’t go there to hone my sense of slang or cultural currency or pop culture. I knew I was going to break my teeth on ballets and arias and Broadway shows, was going to get my share of Shakespeare. But my understanding of hip-hop was not going to be advanced by the Times puzzle, not then. The range of reference was going to be fairly narrow. I wonder whether you could talk about the ways that collaboration and diversity and cultural openness have changed the fill of puzzles and resulted in a different kind of learning for those who solve them.

NL: The Times editor before Will Shortz was Eugene Maleska, a Latin teacher and superintendent, who came at puzzle-editing with a kind of knuckle-rapping rigor and decided that if you didn’t know what an ambo was, or what a Latin cloak ought to be called, then you’d get detention. And so when Will Shortz takes over in 1993, at the time he’s a bit of a middle-America avant-gardist. He’s from Indiana, he’s the editor of Games magazine; he brings this  Technicolor, guffawing, dad-joke sensibility to what had been a buttoned-up affair.

The Games magazine puzzles of the ‘80s generated messages from members of the National Puzzlers’ League asking who Brian Eno is or why “boogie” is the answer for the clue “Get down.” Whether the puzzle’s language should or should not reflect current usage is a time-honored debate.

Then you have that new guard, in the late ‘10s and early ‘20s, who want to install their sensibility into the grid. That’s able to be done in part because of technology. The fustiness of the late-‘80s, early-‘90s grids has a lot to do with how paper-forward those processes were. It used to be, if you wanted to make a puzzle, you’d get out graph paper and pencil and this door-stopping dictionary facsimile called a crossword puzzle answer book, where instead of headings like A to B or B to C, as in a dictionary, the headings would be ??N?D?, and give you all the words that fit that string. The next one would be ??N??D, and so on.

Construction was much more analog. It is now an extremely computer-assisted process. And it’s drawn to the practice young, tech-savvy, script-slinging Millennials and Zoomers who are intent on improving on what’s called a word list file. Puzzle-making software has this big dictionary file with two columns: On the left are your hundreds of thousands of words, and on the right are numerical scores representing how much that particular person would like to see each word in the grid. They’re going through these lists and deleting the bad stuff, but also adding the Janelle Monáe song “Pynk” or Star Wars character Jyn Erso, scraping current events and culture and knowledge and putting it all into the grid.

MG: You reminded me: Years ago, a friend and I got to talking about our frustration that when you put somebody on a prize committee, they stop thinking in the quirky, delight-directed ways that they think as readers or viewers—instead they abstract themselves and read half for some imagined posterity. As I recall, my chief outrage was that The Big Lebowski got zero Oscar nominations in a year when the Best Picture prize went to some bloated, hokey epic about not sharing a floating door. Which of those films, I asked VERY rhetorically, might someone want to rewatch in fifty years?

The analogy I’d make to crossword puzzles is that in a way, the switch to much more collaboration, the switch to encouraging diversity, to allowing more slang, and as you said, scraping pop culture . . . all that means that puzzle-makers aren’t aiming at abstract posterity. Forty years ago, crosswords were expected to try for timelessness. That was maybe inevitable for technological reasons, and because there was lag time and so on, but it’s a different animal when you’re thinking, This is like any other piece of pop-culture detritus, this is ephemeral, and it will go away. Twenty years from now, people could look back at my puzzle and have no idea. And that’s OK. There will be more puzzles . . . puzzles that represent that moment. Back then, the idea was “I’m creating this so it can appear in the Times in a few months, and then in a crossword puzzle book in a few years. And ten years after that, when somebody pulls the paperback out of a drawer, they’ll still be able to work it.”

NL: That’s right. I’m always interested in how economic infrastructure changes forms, and it’s the case that part of the evergreen-ness demanded of puzzles was so they could eventually be collected into best-of books. Those best-of shelves are now flooded with hyper-specific competitors: Here’s a bunch of crosswords that are made by young people, here’s a bunch of crosswords that cater to the Black diaspora, here’s a bunch of crosswords about baseball. That kind of evergreen-ness has in many ways been dispatched with.

Of course, the demand for evergreen-ness has always been guided by bias and skew. There are stories of Will Shortz, this arbiter of staying power, getting it wrong: not knowing what an everything bagel is despite being a long-time New York (in his case Westchester) resident, or making the judgment that “Oops, I Did It Again,” the Britney Spears song that happens to be fifteen letters, isn’t going to stand the test of time, but then accepting it six months later when perhaps her stardom was inevitable. There are many stories of constructors who think they’re introducing something culturally iron-clad but instead receive editorial pushback. And there are fights over how we assess cultural staying power. Google hits? The Billboard 100? Is Urban Dictionary sufficient? These are ongoing debates; editors are going to make calls.

But crosswords have always been ephemeral: You solve it, and then you throw it away. One of my book’s focuses is that contradiction: It’s a game, but it’s asking you for erudite, canonized language. It’s supposed to be this hand in the wet cement of important names, but then you discard it. That long-standing nature and that fleeting nature are built into the form, and some of the best visual art that uses crosswords—Laurie Anderson’s cross-stitched weavings, which were a reaction to the big-dick sculptural monuments of her male peers; or Cameron Granger’s scratched-out, personalized crossword islands—is about that dialectic between disposable daily ritual and immovable canonical name, about solvability and impenetrability.

To your point about literature, there are analogies to the canon wars. The proliferation of outlets, the way puzzles have become even diaristic, mean the crossword can function more like a mixtape or a recommendation engine than this stalwart, stoic sense of what’s important.

MG: That’s great. And leads to my next question, which has to do with technological changes. It’s a new world now for solvers and for constructors—gone the ink-or-pencil debate, gone the notebook-eating trial-and-error ways of building letter stacks. Could you talk a bit about how what things technology has changed, what things technology hasn’t, and what the role of AI potentially is for crosswords?

NL: Construction requires three pretty dissimilar tasks. First, you have to dream up a theme. That could be everything from “Here are three Dr. Seuss books that all happen to be fifteen letters long,” to adding a letter to phrases and reinterpreting them in kooky ways, and so on. Then there’s building the grid, which feels like constructing, using letters as raw material. And finally there’s writing clues. The middle part, building a grid, has always already been amenable to computer assistance.

In the book I talk about one of the first people ever to use a word list file, Eric Albert–who like a kind of jolly Edmund Burke went word by word, ranking the goodness of every one, thus installing in his word list, and so his puzzles, these thousands of aesthetic micro-decisions, and therefore his sensibility, maybe much more than people who were doing it by hand.

Where you place black squares is the most undersung part of crosswords. Georges Perec–one of the pleasures of researching this book was learning that two people in my writing pantheon, Nabokov and Perec, both were not just crossword lovers but made puzzles, sometimes as a living. Perec calls grid-making, that second task, a sort of maniacal letter-based arithmetic. Where all that matters is that words have this or that length, and that their digraphs go well together. In English, that arithmetic is learnable. Around 80% of words, because of the language’s Germanic roots, begin with consonants, and they like to blend, which is why you have strap or block in the first row of a crossword, and the release valve of aioli or Oreo or area underneath.

Because I started on graph paper and pencil, I have this twinge of “eat-your-vegetables” that I don’t think many younger constructors have: I think it’s quite useful to learn the phonotactic skeleton of words by being forced to use your mind, not the software.

Writing clues will always feel like writing. You can sense when someone’s clues are not their own, the same way you can sense when a grid is what’s called auto-filled, where someone didn’t curate every word but just let the computer fly.

The first task, coming up with a theme, has been computer-aided in ways that can make it more interesting. In the book I talk about how the phrase “sore loser” repeats every letter except for the l. It’s almost impossible for the human brain to perform that kind of search. But it’s easy to write a three- or four-line Python script, or an AI prompt if you’re feeling evil, to unearth other possibilities like “Hippocratic Oath,” which repeats every single letter except for the r. Using a computer will give you a wider range of possible answers, enabling human ingenuity and, in this case, you could have the extra letters spell out, in order, “leftovers”: the l from “sore loser,” the r from “Hippocratic Oath,” and so on.

The AI question is always a question about labor. I’m a, you know, moralizing leftist, and whether or not crossword constructors have control over the things we have produced is a matter of our collective bargaining, not the technology’s capability. A tool that helps publishers embed puzzles into websites is piloting an AI feature, and there’s a huge push to not have those tools built on the backs of our puzzles.

People have this sense, and I often do too, that language is special, that punning is special, and that semantic play and lateral thinking are special. AI isn’t great at coming up with dad jokes now. I’m not a thousand percent confident it won’t eventually, but like many in the crossword world, and the writing world for that matter, I do like that there are forms of intellectual and linguistic play that the AI is just shitty at for the moment.

MG: What you’re describing exemplifies the thoughtful procedure one hopes for: Identify what AI is good at, what it’s not good at, and use it for the things it’s helpful with, period. Make sure it’s our tool and not vice versa. What I worry about is that if you outsource your ability to engage in the lateral play you describe, you’re going to lose that ability. I’m also struck here, as I was frequently in the book, by the ways in which the relationship between constructor and solver is analogous to the relationship between writer and reader.

NL: When I first started making puzzles, I was aware of a kind of nascent reviewing apparatus. I mean, there was Rex Parker, who angrily and lovingly reviews the New York Times puzzle.

MG: Until your book, I had no idea that he was an English professor, but I should have known.

NL: There’s a lot of overlap there! The more academic criticism I read and English professors I met, the more I understood that they prized the same high/low blurriness that I did, that good crosswords tend to have. Because Rex Parker liked Nabokov and The Simpsons, and because I liked Nabokov and The Simpsons, I wanted to put them into my puzzles. And when I put Ned Flanders in a puzzle, and he thrilled to it, that mattered to me.

I don’t read the blogs much anymore. As I’ve become more of a writer, I’ve learned that my own creative writing, whether it’s fiction, nonfiction, or poetry, does best when I’m writing for me. It’s the old Ashbery chestnut: When I started to explain my art, everyone got bored, but when I started talking to myself out loud, everyone wanted to listen in.” Weirdly, making puzzles like that has been fruitful, and it’s also exposed me to the very reasonable and common experience that not everyone’s going to like what I want to put in a puzzle, the same way not everyone’s going to like my poetic sensibility. That’s been freeing.

It’s also co-occurred with the proliferation of outlets, so if, when I was in my late teens, I was aiming in that boyish way to make a New York Times puzzle everyone would care about, especially Rex Parker, I now much more make puzzles for me, and maybe the other nerds thrilled to find poetry, African heads of state, Spanish, a lot of food, in their grid.

In the context of literature, that sensibility makes sense, but a puzzle is meant to be solved. A good book is often defended on subjective terms. When I first started out, a good puzzle was defined on pseudo-objective terms: It didn’t have many abbreviations, the fill was “clean.” The website I mention in the book, X-Word Info, talked about puzzles quantitatively: How many high-value Scrabble letters did you manage to cram in, and so on. That’s good when the words in the grid are Jacuzzi, but it doesn’t make a difference if you’re crossing QED with QID.

Byline-forward outlets like The New Yorker have, I think, educated even more solvers that the puzzle is made by a person, and that it might be a diaristic expression for them. That mixtape idea: here are some words and sounds that mean something to me, do they mean something to you?

Constructing makes you much better at solving, and so you can end up disconnected from the “average” solving experience, such as it is. You still have heuristics, like don’t cross two proper nouns at a seemingly unintuitive letter. But watching a solver attempt my puzzles, seeing that agency, that path through the grid, is always a re-education in those assumptions.

MG: This is analogous to the way we teach creative writing. For a lot of people, when you start out there’s a perceived pressure to universalize, but the fact is that anything you write will be an invitation into your own idiosyncrasy, into your own curiosity, your own pleasure, your own playfulness. When you start to feel that as a possibility, it’s liberating. Obviously you can veer into indulgence, you’ve got to watch that, especially as you forget what it’s like to make your way through your own puzzle. To me, the case you’re making for what you call the “diaristic” is compelling, is exciting. There’s that old Toni Morrison injunction: Write the thing you want to read. But there’s no guarantee it will go over.

I mainly read Rex Parker if there’s a puzzle I especially admire (or don’t), and usually I find his snark and finickiness amusing, likable. Often he loves the same puzzles I do, but occasionally there’s one I thrill to that he’s extremely cranky about. On those days he’s as smart as ever, but the acerbity annoys me. What I mean is that there’s room for disagreement, for quirks on both ends, constructor and solver.

You can’t delight everyone, but you want to create the possibility of delighting someone. You’re aiming at a solver who’s willing to be invited into the zone of playfulness you want to inhabit. It does hold your feet to the fire that—here puzzling diverges from fiction—there’s a solution. It’s helpful that there’s no tolerance for obscurity, that you’ve got to be cognizant of that.

But I’m meandering now. Another phenomenon you talk about in the book that I’m fascinated by is what you call the neurological back burner.

NL: My childhood and teenage years are defined by the crossword, in a way, and once I grew up, there were many other things I was interested in. I studied cognitive science in college, I studied economics, I studied literature, I write nonfiction and poetry and criticism, I work in policy advocacy and politics–in a way, the book is this frantic, Sisyphean attempt to use the crossword to stop talking about the crossword. I’m all these other things; I’ll tell you via the crossword. But it’s true that I got interested in cognitive science while trying to solve a Saturday, not getting very far, going for a swim and eating some barbecue, then returning to it and being amazed that my brain had produced the answers that hours ago eluded me.

We all have experiences like this; mine were just refracted through the prism of the puzzle, and it got me interested in what’s going on in the brain. I’ve guest-lectured in classes on the cognitive science of word games, and researchers will tell you crosswords link many different kinds of cognitive effort: whether it’s looking at a string and piecing together where breaks are, understanding polysemy and looking at C-L-O-S-E and not knowing if it’s the verb or the adjective, and so on.

Or even just the silly, punny part of your brain that is willing to suspend disbelief for a joke. I’ve had all these experiences through the linguistic cacophony of a puzzle . . . and one thing I love about a good crossword is how it draws on these varied neurological apparatuses, the same way my favorite books do. You could read Lolita just for the anagrams, the wordplay; you could read it as a moral tale; you could read it as an exploration of frame structures; you could read it just to delight in the music of the diction.

MG: That makes perfect sense. I think, too, for the solver, part of the joy is being brought to conscious knowledge of something that’s usually subconscious. One pleasure for me in puzzles is discovering I know things I didn’t realize I knew. Take a recent Times puzzle: If you’d asked me in advance, have you ever seen the name Charli XCX, I would have said I had no idea who or what that is. But spot me a few letters, and . . . it turns out I do know. That becomes an invitation to look up Charli XCX. Which is part of the pleasure of reading Nabokov, too: looking up neologisms, looking up the recherché words he chooses. It’s an invitation to learn.

My last question has to do with crosswordese. I remember being told years ago that when you face the blank page—the writer’s perennial terror–one thing you’re afraid of is just the infinitude of possibilities. Every time you sit down, especially when you’re starting out as a writer, there’s the faint, unspoken hope that this book, this story will contain everything I know, everything I have to say. With that mindset, every sentence you write is a betrayal, because what you’re doing is limiting and limiting, impoverishing and impoverishing. Every sentence you write forecloses possibilities, and you feel the loss of potential. But pleasure eventually comes from the fact that you’re working within a tight space, a cage made of the bars you’ve chosen, and you have to learn how to make things that are unavoidable feel like choices. It’s the kind of compensation we do in our lives all the time, right? We’re trapped in circumstances, and we want to own those circumstances, want to feel like they emerged in some way from our intent.

That’s something you get a lot of when you’re devising crosswords, and the form it often takes is crosswordese. What is it, and how you feel about it?

NL: Crosswordese refers to the words that seem to appear more in crosswords than in real life. They are strings that, because of their letter patterns, bail constructors out of a corner. Not just words like area and era, whose vowel-heaviness and consonantal bends make them very useful, but words like ursa and Smee that you don’t see anywhere except for a puzzle. You might need to learn them as a kind of secret dialect to solve.

For a long time, crosswordese was seen as a necessary evil. Puzzle-making was so technical that these shorter, obscurer three- and four-letter words felt inevitable. And a couple things changed. One, software use has made the average grid cleaner, which is to say, freer of crosswordese. Another is that we’ve been able to reinterpret crosswordese. If you love good poetry, you might clue aria as the poet Aria Aber. Of course, one person’s arcana is another person’s lived experience; you don’t tell an opera singer that aria is crosswordese.

MG: You mention in the book a constructor who said, hey, I use an etui every day.

NL: Exactly. I was going to bring up Liz Gorski, who’s a friend and a wonderful constructor, and uses an etui every day. I didn’t know what an etui was until I started solving puzzles. The puzzle gave me this misshapen vocabulary, drawn from the darkest corners of the adult world.

But part of the skill of a seasoned constructor is reanimating old bits of crosswordese, if you can. And as grid-making software got better, you could feel a kind of Apple Store lack of edge when you’d solve a largely computer-generated grid. It felt a little lifeless. It had the same high-ranked words as every other grid, but was missing the bitter flavor of crosswordese. With Liz’s puzzles, there’s this wabi-sabi quality to an old-school three-letter word in a grid; it confirms the handmadeness.

MG: I really like that idea that, to use a word that’s useful to constructors, these markers of handmadeness provide umami, right? This is another link with fiction writing: You want evidence that there’s a human being behind there. I want to see somebody sweating behind the grid every now and again, and when you see somebody re-clue some hackneyed bit of crosswordese in an ingenious, defamiliarizing way, hey, here’s somebody who is simultaneously calling my attention to the fact that this was a blind alley, and making me adore the way they got out of it. That’s what I’m looking for, in some sense.

NL: Yeah, exactly. You can sense someone else having solved the problem.

MG: Thanks so much for talking with me. I enjoyed it, and I learned a lot.

NL: Thank you.

Read more from Issue 22.2.