Numlock Sunday: Michelle McGhee on the crossword puzzle

By Walt Hickey

Welcome to the Numlock Sunday edition. Each week, I'll sit down with an author or a writer behind one of the stories covered in a previous weekday edition for a casual conversation about what they wrote.

This week, I spoke to Michelle McGhee who wrote “Who’s In The Crossword?” for The Pudding. Here's what I wrote about it:

An analysis of a sample of tens of thousands of crossword clues, assigning the answers to the actual human people who were the solutions, found that in 2020, 64 percent of the proper noun clues in the New York Times crossword were linked to men, and just 28 percent to a person in a minoritized racial group. Similar disparities were seen in the L.A. Times crossword (68 percent male, 25 percent non-white), Wall Street Journal crossword (69 percent male, 24 percent non-white) and Universal syndicate (54 percent men, 29 percent non-white). Interestingly, in 2020, the USA Today crossword flips that, with 72 percent of the proper noun answers being women and 48 percent of them being from a minoritized racial group, which is actually a serious effort enacted by Erik Agard, the new 27-year-old editor in charge of the USA Today puzzle since the end of 2019. The analysis also turned up some cool evolutions of classic clues, like how following Ava DuVernay’s direction of the 2014 film Selma she surpassed Ava Gardner, classic Hollywood actress, as the default “AVA” clue.

I loved this story because it’s a seriously clever approach that puts hard data to an otherwise difficult-to-quantify problem. The findings are not only unambiguous, but also turned up someone doing some really, really cool work to address the problem, demonstrating directly how clear the route is to a solution.

Michelle can be found on Twitter and on her website.

This interview has been condensed and edited.

Michelle, you wrote a story for The Pudding all about who’s in the Crossword? How did you approach it?

I really love puzzles. I hadn't done a lot of crossword puzzles before though because I felt like I was really bad at them. But I had started getting into them during quarantine and also reading a lot about people like this guy, Erik Agard, who edits USA Today Puzzle, talking a lot about making puzzles more inclusive. I thought this was interesting. It was exciting to envision like, "Oh, what if I could be doing crosswords that highlighted people and things that are actually more relevant to me, rather than just sort of the stuff you memorize to be good at crossroads?" I thought that was really interesting and wanted to see if I could take a data-driven approach to it and quantify how inclusive these different puzzles were and see what the results were.

In this piece, I worked with The Pudding and we looked at tens of thousands of crossword clues and answers, across five different publications, five of the big name ones. And we basically filtered down to all the clues that involve people, and then we looked at the gender and racial identities of those people to sort of see how many women are being included in the puzzles or referenced in clues. And we found, sort of unsurprisingly, that most of these big publications under-represent women and minoritized racial groups when comparing to the census US population. But actually, one stuck out a lot: USA Today. They actually sort of over-represent those groups. Their clues of people are 72 percent women, for example, and I thought that was kind of cool. So, the rest of the piece does a little deep dive into what USA Today is doing differently.

You had a line in here that was just really, really striking, which is "crosswords tell us something about what we think is worth knowing." Can you expand on that idea?

I mean, I think this is why I was really interested, I think I got this sentiment from what I've read a lot about Eric Agard's work and saying like, "Hey, if all our crosswords are referencing knowledge, that is sort of catering to a specific set of people, that is maybe something we should look at." Crossroads are sort of saying, "Oh, this is our common knowledge. This is the stuff we all should know and care about to be like informed, person of the world." So, it kind of said something, if all of those people about who we say, "Oh, these are the people you should know about" are all white men. It feels similar to the books you're asked to read in high school conveying, this is what we think is important, this is what we think is the baseline of what you should be exposed to.

And I think crosswords are kind of like that too because they're puzzles anyone should be able to solve, but really it's sort of catering to sort of a narrow subset of the types of knowledge that it values. I mean, puzzles are fun. They're kind of frivolous, but I think there is sort of something deeper there, if we think about crosswords conveying what we think is common knowledge, what we think is important.

Do you want to talk a little bit about what you found with the name Ava?

This is something interesting. I thought to myself the question, "Okay, how could puzzles be better?" We see that most of these puzzles are mostly white men. Okay, great. That wasn't that surprising. But what could constructors do to make this better or different and make these clues that are more inclusive or relevant? And one big takeaway I feel like from in my analysis is that, even if the answers don't change that much, you can do a lot by changing the clues. So, the answers probably won't change that much because there are letters that are just frequently appearing and they have to intersect with other letters. So, something like "AVA", A-V-A, it's short and it's also, it has some very highly used vowels in there so it's in the crossword a lot.

And it's pretty much all those clues are Ava Gardner. We looked at New York Times clues going back to like the 1940s when the puzzle started. And it's clue to Ava Gardner for a really long time. And then in this chart, I showed the rise of people instead cluing it as Ava DuVernay, the film director, who's also a black woman. And yeah, so I just thought that was sort of an interesting example of you can modernize clues without actually changing the answer. You can just change the clue and highlight a new person who's maybe more relevant like to current day, or representing a different group or a different experience. And you can kind of do that with any answer. Most answers probably.

You looked at five very popular and well-known crossword puzzle distribution networks. Tell me about USA Today, it seems like there's something really sensational going on in that puzzle compared to the other majors.

So, like I said, of the people clues — clues where either the clue or the answer has a person in it — and USA Today its people clues are 72 percent women and 48 percent minoritized racial groups, both of those over-represent those groups in the subject, Which is completely different than all the other publications that are a lot of white men. I kind of looked at it from two angles when it comes to what USA Today is doing. One was the sort of modernizing names. They're using Ava DuVernay a lot more than using Ava Gardner while other publications are sort of still relying on Ava Gardner. So, modernizing of the names. And then the other thing that I noticed that they're doing, that I just thought was really cool, was just taking answers that they have to include and not doing them the same way over and over.

They're just simple clues, I kind of thought of them as like blank slate answers. Like I have a graphic of the answer POEM, and the answer ONT, or ONT. These are just like short, simple words, but they have these sort of new, creative, fresh ways of cluing them. They bring in different groups and different experiences. So, for example, with the word poem, you can see this more easily on this graphic, but it's really often when clean poem, a lot of the clues are like "a Walt Whitman piece" or just when they reference a person, they reference a white man. And USA Today, in 2020, with this new editor, the times that they clue "POEM" they have included relating to a woman of color being like, "A Sonia Sanchez creation" or "Rupi Kaur creation."

And similarly with ONT, they clued that in, "taking testosterone for short," which is usually clued to something like "a Canadian province" for ONT. But, yeah, I think that's just like a really clever, fresh way to clue it, but also brings in the experience of transgender people that I don't know if I've seen those experiences highlighted in the puzzle in that way before. It's just creative and fun and it makes it new and not boring. Like you always see that clues that you know exactly how it's going to go.

I think a lot of times these stories will just be like, "Oh, there's a problem." And then this one, I really love the piece just because it went into, yeah, there's an issue, but it's actually fairly clear the path with which one needs to go down to solve it. And it's also not as hard as it as people tend to make it seem. And I felt that it was very illuminating in your reporting here.

Thank you. That was the goal. Because I think, my initial question was, "Oh, how inclusive are these publications?" And it became really clear really quickly that most of them are really not representing women or people of color very much, but that's not really that interesting. I wanted to get more into, okay, now that we know that, what can we do differently? How hard is it to actually to change? And it turns out it's really not that hard if, at least from my perspective from seeing what USA Today is doing, hiring more diverse instructors and getting creative with how you clue these same answers over and over.

You have been published on The Pudding, where else can folks find you? Anything you want to get out before we wrap up?

This is one of my first published pieces, which is really exciting. I've been wanting to make something like this for a long time and I admired The Pudding for awhile.

Michelle can be found on Twitter and on her website.

If you have anything you’d like to see in this Sunday special, shoot me an email. Comment below! Thanks for reading, and thanks so much for supporting Numlock.

Thank you so much for becoming a paid subscriber!

Send links to me on Twitter at @WaltHickey or email me with numbers, tips, or feedback at