03 May 2010

CoverFail Revisited

I'm one of those silly buggers who does not like to assume conclusions without some supporting data. A little less than a year ago, Justine Larbalestier's frustration boiled over, resulting in her blog post entitled "Ain't That a Shame." In that post, she lamented her lack of control, as an author, over her own covers... including an outright misrepresentation of the protagonist's race. Ms Larbalestier got this response from the publisher when she questioned the proposed cover:

The US Liar cover went through many different versions. An early one, which I loved, had the word Liar written in human hair. Sales & Marketing [at the publisher] did not think it would sell. Bloomsbury has had a lot of success with photos of girls on their covers and that’s what they wanted. Although not all of the early girl face covers were white, none showed girls who looked remotely like Micah.

I strongly objected to all of them. I lost.

Then, a little farther down:

Since I’ve told publishing friends how upset I am with my Liar cover, I have been hearing anecdotes from every single house about how hard it is to push through covers with people of colour on them. Editors have told me that their sales departments say black covers don’t sell. Sales reps have told me that many of their accounts won’t take books with black covers. Booksellers have told me that they can’t give away YAs with black covers. Authors have told me that their books with black covers are frequently not shelved in the same part of the library as other YA — they’re exiled to the Urban Fiction section — and many bookshops simply don’t stock them at all. How welcome is a black teen going to feel in the YA section when all the covers are white? Why would she pick up Liar when it has a cover that so explicitly excludes her?

The notion that "black books" don’t sell is pervasive at every level of publishing. Yet I have found few examples of books with a person of colour on the cover that have had the full weight of a publishing house behind them. Until that happens more often we can’t know if it’s true that white people won’t buy books about people of colour. All we can say is that poorly publicised books with "black covers" don’t sell. The same is usually true of poorly publicised books with "white covers."

(some typography corrected to US standard, footnote omitted, bold emphasis added) Go read the whole entry.

Being an inquisitive SOB myself, I decided to look into this problem to see if there's a discernable source. And there is, and it's fixable. Of course, how I get to that conclusion is, itself, an interesting journey. I'll start with data that has been pulled from several major retailers. Datagathering involved going to the "feature table" in the YA/teen section of the store, visually inspecting what was on offer, and doing a little bit of math. Here are the results for three stores (the dataset is considerably larger):

  A B C
Titles 50 17 62
Titles depicting humans 38 16 51
Persons depicted 51 20 68
Noncaucasians1 depicted 2 0 1
Noncaucasians/all titles 4% 0% 2%
Noncaucasians/human title 5% 0% 2%
Noncaucasians/persons 4% 0% 1%

Keep in mind two other data points: The overall US population was approximately 20% noncaucasian (allowing for half of multiracial identifiers as being not visibly distinct for this purpose) in July 2008, with the proportion of teenagers who are noncaucasian almost certainly higher; and that the data summarized above is both statistically consistent with the larger (multicity) dataset2 and statistically distinct from the overall US population data.

Enforcing antitrust law would probably (not certainly) have prevented this after the demise of Jim Crow. But what does this have to do with antitrust? First, you must understand that antitrust concerns not overall firm size — or, at least, it doesn't under modern US antitrust law; your results will vary in Europe! — but concentration in a market. Those who were paying attention in either a basic logical reasoning course in college, or during the part of high school chemistry in which you learned about the ideal gas law, should immediately spot the problem here:

A measure of market concentration, and therefore a judgment made about when that market concentration is "too high," is only as valid as the definition of the market.

And here's where things get really interesting... in the Chinese-curse sense of "interesting." In each of the last seven major "publishing industry" transactions that were subjected to initial antitrust scrutiny by the FTC and DOJ, the definition of market that was adopted was "book publishing," with no attempt to discern if there was more than one coherent market within that market that required closer examination. That is, YA fiction was lumped in with self-help and "inspirational" books, and cookbooks, and murder mysteries, and GED preparation guides, and celebrity pet memoirs, and everything else that one would expect to find in a "general bookstore." This is invalid, both historically and economically, whether comparing numbers of titles or numbers of volumes sold. There is no monolithic publishing industry — there is the bastard offspring of a three-century-long orgy among thirteen3 incompatible business lines that have only resource-based fungibility (that is, substitution of supposedly like products is encouraged primarily by the resources available for production and consumption — not by consumer preference).

It gets better, too; further inquiry has satisfied me that the cause of Bloomsbury's position, as expressed to Ms Larbalestier, was an almost-offhand remark by one buyer for a Major Book Retailer4 combined with unfocused efforts at sales by both publishers and book distributors and sellers. I suspect that this would qualify more as "disparate impact" than it would as "intentional discrimination," but the result is the same. Of course, if the FTC and DOJ had not had their heads up and locked for the last two decades on scrutiny of mergers, perhaps this wouldn't be such an issue: No one buyer could have had that pervasive an effect, absent actual conspiracy; a greater number of decisionmakers of equal standing makes consensus (helpful and harmful) much more difficult. It doesn't help that the publishing industry as a whole doesn't look like America, either; but that's for another time.

The irony, as Ms Larbalestier notes in closing, is that this nonsensical meme rests on mere assumptions, and not on controlled data (or even equivalency of effort, let alone of product!). Too often, the publishing industry puts its marginal marketing dollar behind books that don't need all that much support, and lets the hard ones languish. The opposite extreme would be no better in the end; but surely there's some space for some risk-taking combined with actual support. Ms Larbalestier's post indicates that she's been happy with Bloomsbury overall, and that she's gotten promotional support that she hadn't elsewhere. That's great; it's not, however, good enough.

  1. Allowing for artistic style and giving as much benefit of the doubt as possible. It's not time to even try to unravel the Hispanic subset, or try to ethnographically narrow things down to Northwest European — leaving aside that about half of the samples were artist's interpretive renderings (and not photographs or photograph-like paintings).
  2. Sure, the individual stores bear some blame here, but — particularly as members of chains — they can only put out on the table what the buyers in New York buy for the stores.
  3. Or perhaps as few as eleven. There's room to quibble over whether, for example, nonprofessional educational materials (like those GED guides mentioned above, or books on writing for publication) belong with "trade nonfiction," "educational," or in a category by themselves. There's no room for argument, though, over the core categories of nonadult nonfiction; trade nonfiction; academic and serious nonfiction; professional; educational; noneducational instructional; adult fiction; children's fiction; ancillary materials; serials and periodicals; and reference works — each of which has distinct economic, and indeed legal, characteristics that require consideration as entirely separate markets.
  4. Without ascribing intent, this Major Bookstore Retailer is well beyond the "excessively concentrated" level for at least five of the publishing markets described in note 3.