fbpx

Rewards of Diversity – Tracking the Representation of Diversity in MTG

This image has an empty alt attribute; its file name is image-2.png
Void Mirror

As a cultural phenomenon, the history of diversity and inclusion in Magic: The Gathering is somewhat fraught. Fortunately, as a commercial entity, Wizards of the Coast recognized this some years ago, and has made an effort to promote diversity within its own rankswithin the game itself (D&D as well) and in the player base.

Mark Rosewater, defender of the color pie and the face of Magic for many players, has written extensively on the virtues of diversity and answered many (many many) questions about the state of D&I in the game, as well as the ways in which WotC hopes to improve on that front. Indeed, things seem to be improving, but issues still remain.

It is often argued that one way to make the game more welcoming to a wider variety of gamers is to represent a wider variety of characters in the game with which players can identify. To that end, I thought it would be interesting to track WotC’s efforts to promote diversity through the game itself by increasing the number of non-white, non-male characters represented in the game’s illustrations and by employing a larger proportion of non-male illustrators to produce them.

Before we proceed, I should note that the data is complicated and incomplete, although I still believe the story it tells is reflective of reality. Specifically, several of the questions in which I’m interested revolve around gender, and the best available data tends to discretize gender, rather than represent it as a more fully continuous spectrum. Further, much of this article rests on the amazing Scryfall Tagger project, which is a great resource, but not without its potential issues. If you want to know more about how I put the data together and the places it falls short, I’ve written some notes here (you should read them!).

 

 

Gender diversity in Magic illustrations

Every year, WotC commissions a lot of art to illustrate Magic cards. Since the beginning of 2018, there have been over 9,200 distinct new illustrations printed on cards, which works out to about 172 per month! Some of these feature beasts, others humanoids and still others no one at all. But the public good that is the Scryfall Tagger project has identified a large number of illustrations with prominently featured femalemale and non-binary characters.

We can count the number of cards tagged with each of those labels and see how many of each are printed each year. We can also look at the ratio of female characters to male characters printed each year, and see how that has varied over time.

Click here for a still image of the full time series.

For much of early Magic history, men appeared more frequently in illustrations than did women. Over time, however, the proportion of newly-commissioned art depicting female-expressing characters has gradually eclipsed 50 percent, such that the total number of cards tagged with male and female is now very close to equal.

You can also see that Wizards has been more consistently portraying non-binary characters over the last decade-or-so.

Ethnic diversity in Magic illustrations

The Scryfall tags also indicate cards that depict persons of color, though I’m not sure these are as consistently labeled as are the gender tags. With that caveat, here is the yearly time series of new illustrations with at least one of the gender or ethnicity tags (orange) along with the count of illustrations with one of the ethnicity tags alone (purple).

You can see that, as the number of cards printed has ballooned in recent years, there has also been a clear step up in the number of POCs in card art. But the ratio of POC:non-POC remains fairly low. Another way of looking at the relative frequencies of POC characters is to literally list out the names of all of the cards represented by the orange line (left side below) and the purple line (right side below). Go ahead and scroll down that timeline 👇.

 

 

That is a lot of cards scrolling by on the left side; not so many on the right. I don’t think there is a specific ratio that WotC should be aiming for when trying to increase the representation of non-white characters in card art – or at least, I don’t know what that ratio should be – but it appears to be pretty low, if the Tagger data can be relied upon.

Gender diversity in illustrators

I had one more idea about a way Wizards might be working to diversify the game – by commissioning more illustrations from female (or, non-male) artists. You should read the methodology details below, but I went through every single illustration in Magic’s history and made a guess at the gender (either male or female only, due to the limitations of the available data) of the artist.

One of the things that stood out to me when making this animated history of Magic artists over time is that some of the most prolific illustrators early in the game were women, but at this point in time, there are only a small number of women among the 43 most-frequently-appearing artists.

Indeed, using data on artists’ inferred gender, we see that women (or rather, people with typically female-identified names) were commissioned for 1/6 to 1/4 of all illustrations in the first few years of the game, but those numbers dwindled substantially through the late 90s and even for most of the early 21st century. You can see an apparent concentrated effort to become more inclusive starting in 2014 and possibly ramping up in 2019.

Indeed, 2021-22 appear to be the most gender-diverse years ever for Magic illustrators, and perhaps the trend will continue upward.

Taken together, what does all of this say about Wizards’ recent efforts to make the game more reflective of the diversity they would like to see in their player base? To me, the record is decidedly mixed. I do think we can see a real effort to increase gender diversity in the game, starting around 10 years ago. I think the evidence of a similar increase in racial/ethnic diversity is less compelling, although I do think our measures for that are probably worse.

That being said, R&D begins work on sets several years in advance of their release. The Black Lives Matter movement, and the protests against police brutality and systemic racism, really came to the fore in the spring and summer of 2020. Wizards, at that time, expressed solidarity with the movement for racial justice, banned cards with racist or culturally insensitive art and came out with a “Black Is Magic” Secret Lair in celebration of the subsequent Black History Month (and have recently followed that up with a well-received “Pride Across the Multiverse” Secret Lair Drop).

I do think WotC’s collective heart is in the right place. I suspect that the game, and those who play the game, will grow increasingly inclusive and diverse over the next few years, and I look forward to it.


 

Thoughts? Questions? Critiques? Suggestions? Leave a comment below or talk to me on Twitter @MtG_DS. I’d love to hear what you think about my attempts to measure the diversity of the game, and whether you have other/better ideas! What do you think about the current state of diversity and inclusion in Magic, and what do you think should change?

I’d also like to express my excitement about the opportunity to write for ChannelFireball, historically home to some of the greatest minds and players in Magic, and now, somehow… me!


 

Notes on the data

Scryfall, along with being the incredible search engine and card database we all know and love, also hosts a Tagger project, in which each card is assigned a set of descriptive crowdsourced tags. Some of these tags denote gameplay mechanics (cantrip, pinger), others relationships between cards (references, strictly worse than), and still others detail the content of the cards’ illustrations (hats, scale ships, orbs).

I use these crowdsourced tags to identify illustrations depicting male, female and non-binary characters, as well as those that portray persons of color. Each tag is associated with an “illustration id,” which I then tie to the release date of the first card printed with that specific illustration.

It’s important to note that while impressive in its breadth, the Tagger tags do not appear to be exhaustive. Consider, for example, the tags for Throne of Eldraine. Here are all of the cards that do carry tags indicating one or more of “male”, “female” and/or “non-binary.”

And here are cards printed in ELD that do not carry one of those tags. Some of these are inanimate objects, others are creatures or constructs for which, even if we could identify a gender, couldn’t be said to be increasing representative diversity in the game. However, others are human or humanoid (1, 2, 3, among others) characters with potentially identifiable gender expression left untagged. It’s possible that the taggers are careful to only label known genders of identifiable characters, but I don’t think that explains why some Faeries do have gender tags, while these do not.

I certainly don’t fault the taggers for this! It’s a huge undertaking, a labor of love, and they don’t owe anyone perfect accuracy or exhaustive coverage. I only mention it to note that while I suspect the precision of the tags is very good, the recall may be lacking, and possibly in a non-random way. I have no reason to suspect any particular bias in the tagged/non-tagged status of any given card, whether by gender, ethnicity, or release date, but that possibility does exist. A worthwhile follow-up project might be to do some hand-labeling to compare against the existing set of tags, but I am reasonably satisfied that the trends detailed herein are reflective of the actual truth behind the data.

I took a different, and less direct, approach to evaluating the male/female balance of the artists behind the illustrations. Each card credits the artist for its illustration, and though there are exceptions, most of the artists are listed with first names in the Latin alphabet.

Using those first names, I use a method called “Algorithmic Gender Prediction“, which is just a fancy way of saying “if you look at historical records, what is the probability that [this first name] was assigned male/female at birth?” This method has many shortcomings, and I would encourage you to read the original researchers’ caveats (“in particular its dependency on a state-defined gender binary”). For my purposes, which are mainly to evaluate the extent to which men have received art commissions from WotC, I believe this method will lead us to an approximately correct understanding of the truth.

Essentially, for any given set of cards – I focus on the year-by-year print run – we can sum first name gender probabilities across all illustrators to come up with an estimate of the true proportion of male illustrators. Even though any individual name’s gender frequency might inaccurately predict a specific artist, the aggregate and probabilistic nature of this tool ought to bring us to a reasonable approximation of reality.

I would welcome any suggestions for other ways to approach this task, especially methods that take into account the variety of gender identities we observe in the population of Magic artists.

Click here to resume reading where you left off.

 

Leave a Reply

Scroll to Top