There is an article in a recent issue of the Guardian reporting on statements by the editor-in-chief of Nature, Phil Campbell. It says two amazing things. The less amazing thing is that he says open access is inevitable. The more amazing thing is that he says that the cost of to publish in Nature or Science is “in excess of $10,000” per article.

Let me just write that one more time for people who don’t speak arabic numerals: in excess of Ten Thousand Dollars for a Single Article in Nature.

Now Nature is a pretty phenomenal publication, and I don’t doubt that it is expensive to produce, but I think most of us cannot fathom what it is that these publishers are doing that might result in such costs. $10,000 is, from the perspective of most anthropologists, is what it costs to do the research not what it costs to type it up. How can this be?

I can answer this. At least, I can tell you what the Good and the Bad reasons are for these costs, and why we should consider paying for the good reasons and stop paying for the bad ones.

The reason I can answer this is because I now publish an magazine– on the web and in print, called Limn. It’s free (as in speech) but it is also for sale (it’s beautiful, btw, and you should buy one, since I’m here explaining to you why you should). The three issues (comprising two print issues and about 30 articles) and website combined have cost about–wait for it–$10,000 to produce. Most of that has gone to the labor of a developer and designer (and anthropologist) who has done an absolutely amazing job. 30 short articles (about 2000 words on average), works out to less than $250 per article; probably less, since that price includes the start-up costs of creating and designing a going concern from scratch. The money came from a combination of research accounts, in-kind support from departments, and my pocket.

There is a huge difference between $250 and $10,000. And really the comparison between our little experiment and Nature is not fair to either side. However, I do think it’s fair to make a comparison with the AAA because the AAA is insisting that publishing is so expensive that switching from subscription to open access will ruin them. But from what I can tell, publishing in a AAA journal is significantly cheaper than $10,000, so it should be within reach of an alternative model. In January of this year, Oona Schmid, director the AAA publications program at the AAA gave a talk at Columbia (which also included our own Alex Golub as commentator; Oona starts at 13.10; Alex at 58.50). Based on her presentation and the AAA annual report from 2010, I have the following facts:

• 2010 AAA members = 10,294
• 2010 Meeting attendants = 5987
• Number of Articles published in 2011 = 600
• Number of Reviews published in 2011 = 482
• Cost of Publication program 2010 = $965K
• Revenue from Publication program = $970K
• Cost recovery for Publications program from membership = 36%
• from Library subscriptions = 63%
• Revenue from membership in 2010 = 1.994M (34%)
• Revenue from Annual Meeting 1.18M (20%)
• Revenue from publications 970K (17%)
• Revenue from investment 785K (13%)
• Revenue from other stuff 16%
• Cost of Publication program 965K
• Cost of Public education 514K
• Cost of Annual Meeting 391K
• Cost of Management and governance of AAA 1.51M (p.s. someone might want to write a separate post about this number)

In her presentation, Oona showed a slide suggesting that in order to cover the costs of open access by membership fees, the average fee would have to increase from $81.63 to $175.49. Multiplying that difference by the number of members would imply that the AAA is earning about $965,195 from libraries who buy subscriptions, which accords with the claim about the publications revenue. But the costs listed in 2010 are less than the revenue, so why she says that membership dues contribute 36% to the cost recovery is a mystery to me. (And by the way, this is further confusing because if revenue from membership was 1.994M in 2011, that works out to $193 per member on average, which is $20 more than it would cost to have open access according to Oona). (And also by the way, none of these numbers allow me to know how much money is coming from, or going to, the AAA for their contract with Wiley-Blackewell, whose profit margin in 2010 was an equally jaw dropping 42%–nearly twice that of Apple’s).

But what’s interesting is that if this $965K number is correct for the cost of the publishing program, then using the 2011 numbers (~600 articles) that works out to ~$1600 per article. If you include reviews, the average drops to $890 per article. That’s still nearly 4 times as much as I am spending to publish open access articles, but it is not a jaw-dropping $10K. Nor is it what AAA folks have occasionally asserted is the cost of publishing an article.

Most publishers now offer an “open access” option by which you can pay an “article processing fee” that ranges from $1000 to $5000 (see Springer’s list of APCs). Nobody in anthropology wants to pay to publish their work, we being a poor people, but given what passes for market rates, the AAA’s cost would be a bargain if they chose to do that. But they don’t need to choose to do that, because they either already get enough from membership dues to publish open access, or they could do it by increasing those dues. The political problem of doing that is not insignificant, but if people understood what they were getting for their money (all the publications of all AAA publications available free to everyone in the world everywhere for as long as we pay our dues), they might just do it. Problem Solved.

But does this number–$1600–actually represent the cost of publishing an article? This is where the good and bad reasons need to be considered. The reality of scholarly publishing is that scholars do nearly all the work and don’t expect to be paid (beyond a salary) for doing so. We conceptualize, we write, we edit, we review, we peer-review, we seek our own permissions, design our own figures, tables and graphs, more often than not we copyedit as well. And we want to do all this. It means we are maintaining control over what we produce. That doesn’t leave much for a publisher to do, but it does leave some things: copyediting, design, printing services, legal and accounting help, web-site archiving, findability and indexing, and so on. Arguably as a small business I should be paying more for these services than a large organization that can negotiate better prices. Somehow I managed to do all this for $250/article or less. So claiming that this part is expensive is a Bad Answer.

The other reality is that publishers run real organizations, as opposed to a garage operations with no official employees. I love our new magazine, and I hope it grows… maybe even becomes part of a real organization, but it is true that it is limited in its flexibility, scope and growth by the fact that it is a labor of love produced on the side of what already feels like three different jobs. Running a real organization requires paying people who are good at it to be professional and to take responsibility for the realities of the market. That’s not free, that’s very expensive. This is a Good Answer, but only if the size of the organization is matched to those realities of the market–why have a huge organization for a product that has a very small market? That makes no sense at all.

But it may also be a Bad Answer because the costs of running a “real” organization have dropped dramatically in the last decade. There are now scores of service companies that parcel up and provide core business functions as simple fee-based services. Salesforce.com is the darling of this transformation, but there are thousands of companies that will take care of your accounts and your infrastructure, manage your brand, do your email marketing, find you employees or do “customer relationship management” for a fee. You can rent office space by the hour, and “cloud-based” webhosting by the minute. So “running a business” if that’s what I am doing, no longer requires any major investments. In the publishing industry, a lot of these sunk costs have gone into setting up supply chains and investing in publishing infrastructure. But somewhere around 2005, people figured out how to package and sell this too as a fee-based service. Enter companies like Qoop, Lulu and CreateSpace (the service we used, now owned by Amazon); because the whole infrastructure has been effectively standardized, it’s now possible to sell a publication almost as easily as having a yard sale. I say almost, because a yard sale can be physically and emotionally grueling, whereas publishing on demand is neither. Of course, if you don’t want your publication to look like crap, you need a good designer–and good designers cost money.

But maybe those prices listed above don’t represent costs–maybe they represent market value? Maybe the price that we and our libraries are willing to pay is in fact somewhere between $1K and $5K per article? This would be a Really Bad Answer. There are a number of absurdities here, but the most obvious is that the prices libraries and universities pay for access to big content are secret. They are negotiated confidentially and publishers work hard to keep the prices secret. They all have “published” prices for access, but no one really pays those prices except individuals and very small organizations. As a result large or rich institutions (like UC or Harvard) can command very different prices for access. So what publishers charge does not reflect what articles cost to produce, nor what people are willing to pay, it reflects how good publishers or their customers are at negotiating. And I’m not an economist, but because they must negotiate with every customer separately, and for every kind of access, it must cost an enormous amount (i.e. several full-time salaries for good negotiators) just to sell the damn content to the handful of people willing to buy it.

For a large organization like Wiley Blackwell (2000+ journals)–you can see why publishers force their customers to buy “bundles” of journals– they simply can’t afford to negotiate a price for every single journal with every single publisher. It’s freaking crazee this system. It’s like Wiley Blackwell is treating the global market for scholarly publications like an enormous Souk–haggling endlessly over every last ounce of content. Wouldn’t it be easier just to put a price on them and be done with it? Hint: the answer is yes, but it would cause their profits to fall, so instead they charge outrageous prices to cover the costs of charging outrageous prices.

The worst reason of course is that publishers think they are selling content, to which they expect to have all the rights. In reality what they sell are services–and they are services that are very valuable and for which we scholars should be prepared to pay if we don’t want to give them all our rights to the content. So unless we can move things towards a system where scholars pay publishers for a suite of services, rather than publishers charging us for access to content they own, then we cannot ever hope to have open access at any significant scale. Whether that price is $250, $1600 or $10,000 is what remains to be worked out.

So scholarly publishing is expensive for a bunch of bad reasons and a few good reasons. The amount we as scholars (and our libraries or institutions) should pay should correspond to the cost of the good reasons. There are two lessons to learn from this: 1) TINSTAAFL. There is no such thing as a free lunch. Somebody somewhere has to bear these costs, whether it is the current publishers, the new breed of OA publishers, universities and libraries or we scholars ourselves… it’s not going away; but more importantly 2) We are currently paying the costs that correspond to the much more expensive Bad Reasons, and if we just stopped doing that, we’d have money left over for beer.

UPDATE: Apparently some people would like to know where I get the 42% profit number for Wiley. It’s from Heather Morrison’s dissertation: ” In the first quarter of 2012, John Wiley & Sons (2011) reported profit of $106 million for their scientific, medical, technical and scholarly division on revenue of $253 million, a profit rate of 42%.” I haven’t verified this claim.