1. Who’s in the Crossword?

By Michelle McGhee
with Russell Samora and Jan Diehm

The New York Times crossword has been criticized for being too old, too white, and too male. They publish more puzzles by men than by women, their clues can be a bit outdated (and even offensive), and a few editors at the top can use their power to maintain the status quo.

As a puzzle lover, I wanted to better understand who is being referenced in crossword puzzles — an unexplored piece of this puzzle. I teamed up with The Pudding to see if the people in crossword clues and answers represent the people who could be solving them. To measure this, we looked specifically at clues and answers that include the names of real people.

A 1969 hit for NEIL DIAMOND
S
W
E
E
T
C
A
R
O
L
I
N
E
A person in the clue
Funny FEY
T
I
N
A
A person in the answer

But before we go too deep, let’s get a feel for an actual crossword puzzle. Here’s a mini, made up of clues and answers from real crosswords we analyzed. It is solvable with either set of clues; one exclusively made of non-Hispanic white people, the other of minoritized racial groups.

Play the puzzle or

Non-Hispanic white people

Minoritized racial groups

ACROSS

1
English author Lofts
5
I was this for Nero
6
Truman's Missouri birthplace
8
George Gershwin's brother/partner
9
Comedian Bruce
DOWN

2
Caesar, for one
3
Ivanovic who won the 2008 French Open
4
Ferlinghetti novel
6
Mae West character.
7
Second note of Guido's scale.
1 2 3 4 5 6 7 8 9
ACROSS

1
Jones with nine Grammys
5
'You da ___' (Rihanna song)
6
Pulitzer-winning rapper Kendrick
8
Stage actor Aldridge
9
NFL Hall of Famer Moore
DOWN

2
TV star Tami ___
3
'Devious Maids' star Ortiz
4
2020 Grammy nominee whose name looks like a pronoun
6
Rap's ___ Kim
7
'We ___ Never Meeting in Real Life' (Samantha Irby book)
English author Lofts
or
Jones with nine Grammys

(Non-Hispanic white)

(Minoritized racial groups)

Want some more puzzles?Check out the others we've generated with our data.

With the exact same answers, we made two very different puzzle experiences. Maybe you found one set of clues easier to solve, or more in line with the people you know about. Maybe you used some of each. Everyone has a different set of cultural touch points, and therefore a different solving experience. For too long, puzzles have catered to a narrow subset.

Crosswords tell us something about what we think is worth knowing. A puzzle that subtly promotes the idea that white men are the standard, the people everyone should know about, is a problem for all of us (yes, even the white men).

Shirley Ann Jackson's alma mater
M
I
T
Williams regarded by many as the greatest athlete of all time
S
E
R
E
N
A
Ali Krieger, to Ashlyn Harris
W
I
F
E

A less homogenous puzzle would be an opportunity for many solvers to expand their worldviews. But more importantly, if you’re a solver like me, it’s meaningful to see yourself and your experiences in the puzzle, especially if they are often unseen or underappreciated. When I see black women engineers, or powerful athletes, or queer couples centered in a puzzle, it makes me feel seen and signifcant. It’s a reminder that I can be the standard, not just the deviant.

How we did it

We sampled tens of thousands of clues across decades and publications from Saul Pwanson’s crossword corpus. Then, we manually labeled each person’s race and gender after researching them. For the purpose of this analysis, we classified people using US Census’s categories. We believe that the lines between races and genders are social constructs, and their precise delineations are moving targets without unanimous agreement. We recognize that this is an imperfect method, but it does not change our finding: crossword puzzles are dominated by men of European descent, reserving little space for everyone else. See a more detailed methodology.

Representation in Major Crosswords

Ratio of men vs. women and non-Hispanic whites vs. minoritized racial groups found in clues and answers of major crossword publications.

Note: The yellow line marks the split of the US population according to the Census (for 2020, we used 2019 estimates).

Gender

Race & Ethnicity


WSJ

2020
69 %
men
31 %
women
76 %
non-Hispanic white
24 %
minoritized racial groups

LA Times

2020
68 %
men
32 %
women
75 %
non-Hispanic white
25 %
minoritized racial groups

NY Times

See data from every decade since 1940.
64 %
men
36 %
women
72 %
non-Hispanic white
28 %
minoritized racial groups

Universal

2020
54 %
men
46 %
women
71 %
non-Hispanic white
29 %
minoritized racial groups

USA Today

2020
28 %
men
72 %
women
52 %
non-Hispanic white
48 %
minoritized racial groups

USA Today is the one publication that actually over-represents historically underrepresented groups. What’s going on there?

As of December 2019, The USA Today puzzle is edited by Erik Agard, a 27-year old crossword champ who told me, “bringing some balance on the representation front is something I actively try to do.” A prominent crossword blogger called USA Today’s puzzle “the most interesting, innovative, and provocative daily crossword” out right now. Let’s take a look at how USA Today, and other publications, are taking a puzzle that’s been called too old, too white, too male, and changing it up.

2. Modernizing old names

These are the people most commonly referenced in NYT crossword answers.

  • 2010s
  • 2000s
  • 1990s
  • 1980s
  • 1970s
  • 1960s
  • 1950s
  • 1940s

Rank

Name

Most common answer

Birth year

1
Josip Broz Tito
TITO
1892
2
Jawaharlal Nehru
NEHRU
1889
3
Erwin Rommel
ROMMEL
1891
4
John Wesley Snyder
SNYDER
1895
5
Vladimir Lenin
LENIN
1870

Many are short and easy to fit in the grid. These names aren’t going anywhere because of how crosswords are constructed. Short words with frequently used letters (known as crosswordese) are perfect for filling the nooks and crannies of a puzzle. But some publications are choosing to modernize these answers with new clues, changing who the answer refers to.

Let’s look at a popular in-answer name; AVA. At three letters, it is the shortest allowed length, and has two highly usable vowels. Since the 1940s, it has almost always been clued to refer to Ava Gardner, a movie star whose career spanned from 1941 to 1986. We’re starting to see some publications, especially USA Today and Universal, modernize this name, often cluing it as Ava DuVernay, an award-winning film director (the first black woman to direct a live-action film earning $100 million).

Use of the answer AVA across all publications

2000 2020* 'Selma'directsDuVernay 0
Gardner
DuVernay
Other Avas
*Note: 2020 is projected for the full year using the first 10 months.

fake

Universal
USA Today
NY Times
LA Times
WSJ

AVA DuVernay in 2020

fake

Universal
USA Today
NY Times
LA Times
WSJ

AVA Gardner in 2020

3. Redefining “common knowledge”

These are the people most commonly referenced in NYT crossword clues.

  • 2010s
  • 2000s
  • 1990s
  • 1980s
  • 1970s
  • 1960s
  • 1950s
  • 1940s

Rank

Name

Sample clue

Birth year

1
William Shakespeare
Scram: Shakespearean.
1564
2
Walter Scott
"Fortunes of ___" by Scott.
1771
3
Edmund Spenser
"To ___ thy soul with crosses"—Spenser.
1552
4
Alfred, Lord Tennyson
"The old ___ changeth."—Tennyson.
1809
5
John Milton
"The rule of ___ too much."—Milton.
1608

Answering the clue relies on knowledge about that person or their work. You likely know many of these dead white people (William Shakespeare, George Washington, Julius Caesar) because they're considered “common knowledge.” Choosing to include people like this is practical and easy because it capitalizes on their widespread recognition.

Instead of taking the worn out path, leaning on and reinforcing this common knowledge, USA Today — with Agard at the helm — often takes a different approach. For example, in September of 2020, USA Today clued ONT, which has been clued hundreds of times as something like “Canadian prov.”, like this:

Taking testosterone, for short
O
N
T

Agard instead wanted to bring in the experiences of transgender people (if a transgender person seeks testosterone hormone therapy, they might describe that regimen as being “on T.”) USA Today’s puzzle is full of clues like this, that subvert the norm and bring new people and experiences to the center. Agard credits this to his diverse team of puzzle constructors, who bring a variety of perspectives and ideas.

There is plenty of opportunity to make both our common knowledge and puzzles less homogenous, even with simple, everyday answers.

211 clues for the answer POEM


All clues

All clues

129
clue s do not mention a person
75
clue s mention non-Hispanic white people
7
clue s mention minoritized racial groups (MRG)

Clues that mention a person

200020102020 White MRG

185 clues for the answer MIT


All clues

All clues

169
clue s do not mention a person
15
clue s mention men
1
clue mention s women

Clues that mention a person

200020102020 Men Wom.

Crosswords don’t have to be stodgy and out of touch. They can reflect who we are and what we care about. They can teach us about people and things we don’t know about, but maybe should.

One way to get there is for more people to make and solve crosswords.

  • If you’re interested in making crosswords, Erik Agard (editor of USA Today’s puzzle) helps run a Facebook group for underrepresented crossword constructors to get tips and mentorship.
  • If you’re a new solver (or want to get more confident), here’s a guide from the New York Times on how to get started.
  • Check out publications like Queer Qrosswords, Inkubator, and Women of Letters that intentionally center minoritized groups.

Want some more puzzles?Check out the others we've generated with our data.

Methods

We used Saul Pwanson’s crossword corpus, which contains puzzles from over 30 different publications. We focused on 5 of the most well-known publications: New York Times, LA Times, Wall Street Journal, Universal, and USA Today. We randomly sampled puzzles from each publication and decade, and then narrowed them down to the clues and answers that referenced people. The answer needed to be the first name, last name, full name, nickname, acronym, or abbreviated name of a real, famous, individual (not a family or group of people), or the clue needed to contain the name of a real, famous, individual. For the New York Times, we sampled at least 670 of these clues per decade per publication, exceeding the requirements for a 95% confidence interval and 5% margin of error. For the 2020 publication sampling, we only looked at the first 6 months of the year, and sampled 630 clues, for the same confidence and margin.
We manually labeled each person’s race and gender after researching them. For the purpose of this analysis, we classified race and ethnicity using the US Census’s categories. We removed any people who we couldn’t classify confidently based on their self-identification, ancestry, appearance, and context. We recognize that classifying people this way (rather than asking how they personally identify) is imperfect, but at this scale, and since many of the people are dead, that was not feasible.
For the graphics on race, we split people into two groups — non-Hispanic whites and minoritized racial groups (everyone else). We made this choice to highlight what we see as a fundamental hierarchical divide. As Isabelle Wilkerson writes in her work on racial caste systems, “The American system was founded as a primarily two-tiered hierarchy with its contours defined by the uppermost group, those identified as white, and by the subordinated group, those identified as Black, with immigrants from outside Europe forming blurred middle castes that sought to adjust themselves within a bipolar structure, and Native Americans largely exiled outside it.”
We did find nonbinary people (0.6% of USA Today’s person clues in 2020 were nonbinary, so were 0.3% of Universal’s), but since the numbers never exceeded 1%, they do not appear in the graphics.
Thank you to our data collectors (Sonia Blinderman, Leslie Diehm, Paul Han, Miranda Jen, Iris Leung, Raymond James McGhee, and Jackelyne Nguyen), to Megan and David, for inspiring and cheering me on, and to Mr. Stein, who taught me to love puzzles (and statistics).