Abridgment As Added Value
From Peter Suber’s September 2009 issue of the SPARC Open Access Newsletter.
Imagine that an open access (OA) journal could generate revenue by selling abridgments of its full-text OA articles. Imagine that the revenue even made it unnecessary to charge author-side publication fees. That would be a supremely elegant business model, if only it could be made to work.
BMJ has made it work for more than 10 years, and next year will take the idea even further.
All BMJ research articles are full-text OA in the digital edition of the journal. The print edition, which is toll access (TA), contains 3-5 page abridgments of each research article. BMJ calls this system ELPS (for "electronic long, paper short"). The OA edition of the journal charges no publication fees, and the full-text research articles have no word limit.
Nine months ago, BMJ introduced even shorter, one-page abridgments called picos, and published selected articles in the pico format rather than the longer 3-5 page format. The experiment has been so successful that BMJ announced last month that it will phase out ELPS and go all-pico. Starting in January 2010, all BMJ research articles will have pico abridgments in the TA print edition, and the full texts will still be no-fee OA in the digital edition.
When BMJ announced its new pico policy, it also released the results of its survey of BMJ authors. (All the surveyed authors were also BMJ and pico readers; 67.6% were frequent or very frequent readers.) It turns out that the 3-5 page ELPS abridgments were "the least read content" of the print edition of the journal. But 71.1% of the survey respondents said that pico summaries will be read more often than ELPS summaries. About two-thirds (65.9%) said that readers of pico summaries will want to read the full-text editions: 43.9% because picos whet their appetite for more and 17.3% because picos omit important information. Perhaps because picos are more inviting to readers, the surveyed authors found them more inviting to authors. 39.3% said that picos were better for authors than ELPS summaries, and only 13.3% said they were worse. 11.6% predicted that the shift from ELPS summaries to pico summaries would increase submissions, and only 7.5% predicted it would decrease them.
The survey gives BMJ good evidence that the shift will appeal to both authors and readers, which has to be good for the bottom line. But there's another economic benefit. The ELPS summaries were prepared by BMJ editors, at BMJ expense, and approved by authors. Pico summaries are written by authors using BMJ templates. Dr. Trish Groves, the Deputy Editor of BMJ who will run the pico program, told me that the transition should save the journal £50-100,000/year (about $82-164,000/year).
If readers really do prefer picos to ELPS summaries, that preference may show up in increased subscriptions and revenue, just as the author preference for picos may show up in increased submissions and selectivity. However, BMJ is careful not to predict a revenue bump, not to estimate one, and not to bank on one. But when I look at the survey data, I see the distinct possibility and I'd be surprised if BMJ didn't see it as well.
Groves told me that the primary goals of the pico program are "(1) to encourage print readers (mostly 'ordinary' UK practising doctors) to actually read, understand, and even use some BMJ research..., (2) to innovate: we wanted to develop summaries/abstracts that were more evidence based, more transparent (incl funding and competing interests) and more readable than standard structured abstracts, (3) to reduce delays between online and print publication of research (now that we're getting and accepting more high quality papers) by fitting more studies into each print issue (6 or 7 a wk instead of 3-5), (4) to cut some costs: the abridging for ELPS was done by the BMJ at our expense..., and (5) to give authors control over the abridging and save them from having to approve two long proofs (with ELPS even the abridged one is quite long)."
Picos will occupy about 8 of the 60 pages, or 13%, of each print issue of BMJ. The rest will be devoted --in Groves' words-- to "education, clinical reviews, analysis, news, other medical journalism, letters, and views and reviews."
That's the model. Here are a few thoughts.
* The BMJ survey results don't surprise me at all. I'd love to see structured one-page summaries of new articles in my field. I know I'd prefer one-page summaries to 3-5 page summaries. They'd better serve my need for a rapid overview of large swaths of literature. Whenever I wanted more than one page's worth, I could go to the OA full text.
* This model works better for some kinds of summaries than others. As BMJ found, pico-length summaries have more marketing power than shorter abstracts and longer ELPS summaries. The sweet spot will probably differ from field to field.
Moreover, if abridgments are done badly, they'll provide no added value at all, or the subtracted value will exceed the added value. BMJ has refined its pico template over the past nine months in order to keep quality high and editorial oversight low. But the optimal template or recipe will also probably differ from field to field.
* If a journal's summaries hit my sweet spot for length and quality, would I be willing to pay a reasonable price for them? Absolutely. In my own case, I can truthfully say that I've been looking for a good source of article summaries since I started grad school 34 years ago.
I'd find it even easier to pay for TA summaries if the full texts were OA. There are two reasons. First, I could move costlessly to the full text whenever a summary piqued my curiosity. I'd still want to read summaries when the full texts were unavailable to me, but I'd find it much harder to justify paying for them. Second, I'd know that my money was supporting OA. Other things being equal, I'd rather spend money in support of OA than, say, shareholder dividends, and I'd rather help a good cause and get something like picos for my money than just help a good cause.
* The term "added value" is often used in two ways: value above and beyond what we started with, and value worth paying for. We've all seen examples of the former that fall short of the latter (weak articles slightly improved by peer review), and examples of the latter that fall short of the former (strong articles not improved by peer review). Intelligent summaries or abridgments could be added value in both senses.
Boiling down a long article to a one-page summary could give us value we didn't have in the original. Authors know this if they've ever written an abstract and found that the exercise gave them new clarity about what they were saying. Intelligent abridgment could also give us value worth paying for. Readers know this if they've ever consulted an encyclopedia when they already had easy access to primary sources.
* I love the idea of selling access to abridgments or summaries in order support full-text OA, especially no-fee OA. I'd love to see the idea spread. But there are reasons why it may not spread very far very fast. At least two aspects of BMJ's situation, which boost its benefits from the model, are seldom found in other OA journals.
First, BMJ already has a print or TA edition and uses the revenue to support the OA edition. It only has to tweak the TA edition, not launch it from scratch.
Most OA journals don't have print or TA editions, and most that have tried both have not been able to sustain both. On the other side, however, the exceptions like BMJ are a fascinating and instructive bunch. One of the major exceptions is Medknow, the first for-profit OA publisher to show actual profits.
Second, BMJ already publishes summaries of its research articles and prepares them at its own expense. Even if picos won't increase circulation of the print edition at all, dropping ELPS summaries will save BMJ £50-100,000/year. By contrast, picos at most other OA journals would be a labor addition, not a labor subtraction. That doesn't foreclose the possibility that they could trigger a net financial gain, but it does reduce the odds.
* While few OA journals have TA editions, many TA journals have partial OA editions. Those TA journals might consider picos as part of a larger business model allowing them to provide OA to their full-text research articles. If a TA journal already carries a significant amount of content other than research articles (to quote Jones again, "education, clinical reviews, analysis, news, other...journalism, letters, and views and reviews"), then it might be able to justify a partial OA edition consisting of just its full-text research articles. This is even more likely if its own polling showed, like BMJ's, that the pico-fortified TA edition would be more appealing to readers and authors than the standard TA edition.
* OA journals without TA editions could consider partial TA editions. This wouldn't detract from OA if all their research articles were still full-text OA. These OA journals could launch small TA editions limited to picos, or larger TA editions of picos+ (e.g. picos plus "education, clinical reviews, analysis, news, other...journalism, letters, and views and reviews"). These TA editions would be less expensive, hence more viable, if they were digital TA rather than print TA.
TA picos don't have to generate much revenue or much additional revenue. They only have to generate enough extra to enable an OA journal to justify a print or TA edition for the first time. It might already be close and TA picos could put it over the top. The result could be a new revenue stream without new access barriers for the research articles.
* As far as I can see, the BMJ model would work just as well if its TA edition were digital and online, rather than print. BMJ is not currently planning to move the TA edition online, but that's an option it may consider in the future.
* Suppose that both the picos and the full-text research articles were OA. Clearly the summaries would attract more readers than the full-text articles. Articles invite readers working on the article's narrow topic, or on closely related or overlapping topics. But summaries invite readers scanning a whole field or subfield for new developments. Readers who have confirmed the relevance of a given article will always be fewer in number than those who are scanning to find relevant work or trying to decide an article's degree of relevance. Article readers have a particular interest while summary readers have a general interest. While we all have particular interests, more of us have a general interest than any *given* particular interest.
Bottom line: Summaries appeal to a much larger audience than full-text articles. The market is larger. The demand is greater. That's a fact worth knowing for any journal wishing to generate revenue by selling some content and providing OA to the rest.
* Picos don't merely appeal to a larger market than full-text articles. They better fit the economy of attention. When we need to read full-text articles, we need to find the time to do so, and we often fail. But whether we need to read full texts or merely skim, summaries are friendlier to our crowded, demanding lives.
We're ready to pay for services that save us time. From one point of view, this is the chief value of peer review itself. Picos don't do what peer review does (or vice versa), but both picos and peer review prove their worth by helping us save time, set priorities, husband our attention, and cope with information overload.
* No one should think that by selling access to summaries and providing OA to full texts, BMJ is saying, let alone proving, that summaries are "more valuable" than full texts. Both are valuable: full texts for depth, summaries for breadth. Summaries may appeal to a wider market, but only because more people scan for breadth than dig for depth in any particular spot. To advance research on any particular topic, summaries are inadequate and full texts are indispensable. If anything, full texts are more valuable for the later stages of research than summaries are for the earlier stages.
That's exactly why BMJ wants to make the full texts OA and subsidize them with revenue from another source. It's also what's wrong with the publishing lobby's suggestion that funding agencies should follow the US National Science Foundation (NSF), and provide OA only to summaries, rather than the US National Institutes of Health (NIH), and provide OA to peer-reviewed full-texts.
No doubt, selling access to abridgments and providing OA to full texts is an odd inversion. But it's only odd relative to what we're used to --treating the full-text article as the end of the value chain, and treating the summary as the free advertisement for the article. After we get our sea legs in the new model, there's nothing odd about selling content which is somewhat valuable to a large group and giving away content which is very valuable to a smaller group.
While BMJ is not proving that summaries are more valuable than full texts, it is proving that when they are done well, they sell. They sell because they appeal to authors and readers. That's the discovery I hope will sink in.
* Some people call the journal literature the "minutes" of science as if it were just a summary. But it's more than that. If the minutes of a meeting summarize a discussion, the journal literature is a large part of the discussion itself. Importantly, however, in an age of conferences, preprint servers, blogs, wikis, listservs, and email, it's not the whole discussion.
Wikipedia aspires to provide OA to a summary of knowledge, and doesn't even accept original research. But the larger OA movement wants OA to knowledge itself. Or at least it wants OA to the full-text articles and other primary sources where knowledge is taking shape: through proposal and report, confirmation and disconfirmation, challenge and response, and yes, even summary and review. It's a messy process which, in its totality, is neither consistent (as it works through the clash of conflicting hypotheses) nor stable (as it discards weak claims and considers new ones that appear stronger). The messiness and instability are properties of a discussion, not the minutes of a discussion. The journal literature isn't just a report on the process but a major channel of the process itself. And not incidentally, OA is valuable not just for making the process public but for facilitating the very workings of the process.
This is just to say that summaries will never play the role of full texts, and that we can appreciate the BMJ model without mistaking it for the subordination of full texts to summaries. Indeed, we can appreciate it precisely because it keeps its eye on the prize and provides OA to the full texts.
* BMJ isn't denying that if priced summaries are desirable to readers, then free summaries would be even more desirable to readers. In fact, readers may value OA summaries at least as much as publishers value TA summaries. If the pico model locked up all summaries or abridgments, that would be a high price to pay, even if it made full-text OA possible. But the pico model doesn't do that. BMJ still provides OA to the abstracts accompanying its full-text articles, and plans to continue.
* Like BMJ, RNA Biology also asks its authors to write summaries, though the RNAB summaries are posted to Wikipedia. RNAB doesn't have a print or TA edition and chooses to benefit readers by sharing the summaries openly. But by the same token, because it doesn't sell access to the summaries, it doesn't get the revenue from doing so either. Naturally I prefer the RNAB model to BMJ's, for journals which can manage without the extra revenue; but the BMJ model could save OA at journals which do need the extra revenue.
* In 2007, Emerald's TA journals in engineering asked authors to write summaries of their research articles, and rewarded those who did by making their full-text articles no-fee OA. This was not only one of the rare no-fee hybrid OA journal models; it was also one of the even rarer models to value author summaries as highly as some other journals value cash. (I can't tell whether Emerald is still using the model.)
* Would some pico variation work for monographs? Because monographs are longer than research articles, does it follow that the demand for monograph abridgments is higher than the demand for article abridgments? Or is the demand greater for article abridgments on the ground that most working researchers (in the sciences) consume many times more articles than monographs? Regardless of where the demand is greater, is there *enough* demand for monograph abridgments to enable a creative press to offer full-text OA for its monographs while selling summaries and POD editions of full texts?
What if a university press launched a few TA journals containing abridgments of its OA books? I say "a few" so that the abridgments could be clustered roughly by field. I wouldn't pay for a TA journal with book-abridgments in all fields, but I might pay for one with book-abridgments in my field, philosophy, or in the humanities generally.
What if a publisher of many journals in the same field --say, medicine-- made all the journals OA and published a TA journal of abridgments? The journal of abridgments might sell well, but it might not sell well enough to pay for the conversion of many separate TA journals. One advantage of the BMJ model is that one TA journal (print edition of BMJ) only pays for one OA journal (OA edition of BMJ).
* I'm not taking a position on whether summaries ought to be OA or TA. I'm saying that selling access to summaries in order to support full-text OA is elegant and promising, even if counter-intuitive, and should be considered in every niche where it might work. If intelligent summaries really benefit readers, authors, journals, and OA, even when the summaries are TA, then they are an experiment worth testing and replicating elsewhere.
BMJ announcement that (starting in January 2010) all its research would appear in pico, August 2009
BMJ pico-summary of the user surveys about pico
BMJ editorial introducing BMJ pico in December 2008
RNA Biology asks authors to write summaries of their articles for Wikipedia and conducts peer review on the summaries.
Emerald's TA engineering journals formerly asked authors to write summaries of their research articles and rewarded those who did by making their full-text articles no-fee OA.