Saturday, July 1, 2006

Digital Downside


The Digital Downside

The Obsession with Open Access[1]

J. Ewing

Scholars and librarians usually focus on the advantages of electronic journals—faster processing, reduced costs, and new features (such as searching and linking)—and there are indeed many such advantages. But like all technology, electronic journals have a downside as well. Most people have ignored this downside because these problems are presently little more than annoyances. But we should understand the downside in order to prevent these small annoyances from turning into big crises in the future. In one or two cases, it may already be too late.

Here are five problems with electronic journals—problems that arise when the miraculous new technology is combined with the human fraili­ties of carelessness, greed, myopia, dogmatism, and infatuation.

1 Careless Scholars

In the digital age, we can do things we could never do before. Here are some examples from the recent literature of things we can do:

  • A journal posts an article in January; in April, without any notice, the editors replace the article with a “corrected” version.
  • A journal posts an article in July; in November, the publisher sim­ply removes the article (without notifying the authors!).
  • A journal posts an article in April, but the author posts a corrected and substantially changed version on a well-known server in Oc­tober, with an indication that the article appeared in April, but no indication that it was different.
Mathematics relies on its literature for its underpinnings; its liter­ature is interconnected. Imagine a world in which one percent of the mathematical literature is affected in the above way, in which one per­cent of the articles one finds are not the “authentic” versions. Over time, as work based on faulty references spreads, the fraction of unre­liable literature will increase. Experts may be able to overcome this, but nonexperts will be overwhelmed. We ignore this potential crisis at the peril of future generations of mathematicians.

These sloppy practices occur because new technology allows us to do things never before possible. That doesn’t mean we should do them! To prevent this from destroying the scholarly literature, we need to insist on high standards. There are two ways to handle this—back­-linking or forward-linking—and both require discipline. Every author regrets publishing mistakes (as do publishers!), but we have to resist the temptation to hide them.

2 Big Deals

The electronic age has made it possible for big publishers to offer big deals. Here’s the way they work. Rather than subscribing to journals one by one, an institution is offered electronic access to a huge package of journals across many fields (many of which were previously unavail­able to the institution). Initially, the cost of this package is comparable to the cost of the publisher’s journals to which the institution previously subscribed, and (the publisher points out) it’s always far less than the total cost of the individual subscriptions. This seems to be a wonderful opportunity for the institution, and the vice-president for information (the one who negotiated the deal) crows about the fiscal prowess that brought about this arrangement.

Big deals are now offered by Elsevier (an innovator in this area), Springer, Wiley, Blackwell, and Taylor & Francis. In a recent survey of US research libraries, 93% indicated that they held bundles with at least one of these publishers. Just about half (49%) held bundles with at least four. Wiley, Elsevier, and Springer have achieved 70% market penetration.[2]

Figure 1. Percentage of commercial journals among all mathematics journals


When asked about their motivation, most said that it was a “good return on investment” and that the “alternatives . . . were prohibitively expensive.”

But these institutions have paid a heavy price for that “good in­vestment.” Such big deals are almost always multi-year contracts that do not allow cancellations or changes. The extra titles are often of marginal value to scholars. Most importantly, decisions about what is purchased are made at a high level, far removed from scholars them­selves. In the end, big deals make it more difficult for scholars to make sensible decisions about journals based on price and need. Of course, big deals give the big publishers a substantial advantage over smaller publishers, which is their real purpose.

Big deals are hard to resist, although a few prominent libraries have done so. But the commercial publishers are winning. In 1985, roughly a third of mathematics articles were in commercial journals; by 2004, over half the articles were commercial.[3]  We need to fight back.

3 Walt Disney

The Walt Disney story is a metaphor. The first Mickey Mouse cartoon was produced and copyrighted in 1928. By the late 1990s, the term of copyright for Mickey Mouse (then 75 years for corporate works) was about to expire in 2003. The Disney corporation was making lots of money on the character, partly because new technology allowed new uses. What did Disney do? It persuaded the US Congress to change the law, extending copyright by 20 years. It is now 95 years for corpo­rations; for authored works, copyright extends for the life of the author plus 70 years.

Copyright—the ability to own intellectual property—was never meant to be forever (the late Jack Valenti, as head of the movie producers’ as­sociation, was overheard to say the term should be “forever minus a day”). But as technology progressed over the past 400 years, copy­right became longer and longer . . . barely indistinguishable from for­ever. Until the digital age, this did not affect scholarly publishing much. Now, however, there are new reasons to worry about copyright’s reach.
Scholarly publishers have never made much money by selling jour­nal back volumes. In the age of print, publishers kept a few copies of old journals to sell to libraries when they wanted to replace missing volumes or (rarely) to start a new collection. Typically, such sales of back volumes amounted to one to three percent of journal revenues. Publishers expected to recover their initial costs (and make a profit) by selling current subscriptions, not by selling back volumes.

Now, however, like the Disney corporation, publishers see an op­portunity to make money on their old material using new technology. They want to sell their journals twice, once as a current subscription and a second time as a collection of backfiles. There are many varia­tions on this scheme, some merely sell the backfiles with the current subscription, pointing out that it makes the subscription more valuable (and hence more expensive). But the central point is that publishers carefully control access to the backfiles.

Of course, publishers who digitize their back volumes want to re­cover their costs. But rather than find a way to pay the one-time cost of digitization, and then make the material freely available, they want to continue making money next year, and the year after that, and after that, and on and on into the future. This includes publishers of every kind, not just the commercial ones.

This doesn’t make sense for Mickey Mouse, but it makes even less sense for scholarship. By “owning” scholarship and restricting access indefinitely, publishers make it impossible to realize the dream of con­necting the large body of the past scholarly literature to the present. This makes our present journals less valuable, not more, and ultimately hurts the publishers themselves.

Copyright was invented to make publishing profitable. But no one— no publisher, no author, no one—needs to hold copyright for more than two or three decades. We ought to make a pact that everything goes into the public domain after 28 years.[4]

4 Mindless Accountants

Using flawed statistics to make (equally flawed) decisions is not new. But it is so, so much easier to do in the digital age. Many people have written about the “impact factor,” which is more and more frequently misused, but I want to talk about another troublesome statistic—journal usage, which is even more dangerous than the impact factor.

Librarians all over the world are insisting on usage statistics for journals, which roughly translates into the number of downloads of var­ious articles over a given period of time. They claim this is necessary in order to measure the value of journals and to make decisions about sub­scriptions. Publishers in general seem happy to oblige—almost eager in many cases.

This is a hopelessly naive and dangerous game. What is the meaning of such usage statistics? What does it mean that some article has been downloaded 100 times? Have people read it 100 times? Surely not— you don’t read every item on which you click while browsing the web, so why would scholars read every article they download? And which is more valuable, an article downloaded once a week for ten years, or one downloaded 520 times in its first month? What about caching and the many flaws in browser software that give rise to faulty counts? There are many questions, but almost no answers—just the demand for usage statistics in order to measure value.

Making decisions using ignorant accounting has always been a bad idea, but in this case it may have some disastrous consequences. If li­brarians are really going to measure value by clicks, surely publishers will force users to click more often. It would be foolish to give away abstracts or references or even bibliographic data, for example, if this leads to fewer clicks and hence less value. And if value equals clicks, then why on earth would publishers let authors post copies of papers anywhere except on the publisher’s website. The more liberal a pub­lisher’s policy in this regard, the more the publisher risks losing value. Making decisions by using flawed usage statistics will inevitably shift publishers’ practices, all in the wrong direction. We need to explain this to those who dogmatically claim that value can be measured by a few flawed numbers.

5 Fads and Fashions

A corrupted literature, a literature controlled by a handful of giant publishers, a literature hidden away forever, a literature shaped by nutty accountants using flawed statistics—all these are potential crises caused by the advent of electronic publishing. None is insurmountable, but each is worrisome. And so with these worries pressing upon us, what do we advance as our most pressing issue in the new electronic age? Access—open access to the literature.

On the face of it, this is bizarre. In the digital age, scholars have more access to the literature than ever before. When an institution subscribes to a journal, the articles are now delivered straight to a user’s desktop. Finding articles is far easier than ever before using any number of search tools. Articles can be downloaded, printed, and (dare we admit?) sent to others via email. Even when a user’s institution does not subscribe to a journal, the user can see the abstract and (often) the list of references. This allows scholars to decide whether the article is useful and then to send email to the author to ask for a copy (or simply to find it elsewhere on the web—recent changes in publisher policies make this easy). And many publishers provide access to older articles without any subscription at all. None of this was possible in the print-only world—none of it! In the print world, good access to the scholarly literature was restricted to a few institutions, and near-universal access was unheard of. This has all changed.

Nonetheless, instead of focusing on the four potential crises men­tioned above, the scholarly community has decided to focus on access, the one aspect of the scholarly literature that has already been greatly improved in the digital age.

I have spent a lot of time trying to understand why people seem obsessed with access, and I have come to realize that there is no simple answer. In part, it is human nature: when things improve, we want to improve them still more. Having a taste of increased access, people want completely unfettered access. In part, this is because the call to “open access” is simple to understand. Scholars are not willing to invest time in wrestling with the tougher issues of scholarly publishing mentioned above, which are messy and sometimes hard to unravel. But there is a more subtle, and more insidious, reason for the obsession with access: whenever technology changes the world around us, we become more susceptible to fads and fashions. New technology opens up new opportunities, and new opportunities bring forth opportunists— zealous people who promote their own special causes.
On the face of it, increased access is surely not a bad thing: it is hard to argue against having more access to scholarship. On the other hand, it can be bad if it causes us to ignore the real problems we face, and it can be tragic if new enticing technology combines with an irresistible fad to mislead us into acting against our own interests.

Open access has had both effects on mathematics. When plan­ning for our digital future, we spend most of our time talking about access (already greatly improved), and almost no time talking about the integrity of scholarship, copyright issues, foolish bureaucrats who use faulty statistics, or (worst of all!) avaricious publishers who have created a crisis in scholarly publishing. Instead, we talk about ac­cess. And, of course, those avaricious publishers are delighted by the distraction.

We also formulate ideas that are clearly bad for mathematics. The author-pay model for journals simply does not work for mathemati­cians. Our funding levels do not support it and they never will, at least at the levels of medical sciences. We will always be at a competitive disadvantage to other scientists in an author-pay world.

The self-archiving model of open access is clearly bad for mathe­matics, which more than any other discipline depends on the long-term survival of a reliable web of scholarship—into the distant future many decades from now. And preprint servers, without any obvious source of long-term funding, are not much more attractive.

The government-funded model of open access is also clearly bad for mathematics, which has never competed well with the other sci­ences for government largess. Besides, the budget for the government-funded model seems to assume a stable publishing environment for the long term, without the need to invest in ever changing technol­ogy. Surely this is short-sighted. We will have to invest even more in the coming decade than we have in the past in order to keep up with changing technology. Will the government really do that investing?

Indeed, every model of open access that has been proposed is clearly detrimental to the broad interests of mathematicians. But open access has become an obsession for many that has blinded otherwise thought­ful people into acting against their own interests. That is a real down­side to new technology that may do the most damage in the long term. This is a downside that already affects us.
Don’t believe me? Just check out those large commercial publishers who are all beginning to embrace open access. They are creating sepa­rate corporate units, just to promote and implement open access. They can do that—they have the resources. They know that no matter how the business model changes, they will be able to take advantage of it to make money—lots of it. Surely publishers who have raised subscrip­tion prices for years will feel equally free to raise author charges in the future. Change is good for large corporations, who have the resources to invest in change; it is a lot harder for the little guys who operate on a tight budget and make a tiny profit. This is especially bad for math­ematics, which has traditionally relied on more small and independent journals than any other discipline. If open access is so great for us—if open access is about to solve all our problems—why are the commercial publishers jumping on the bandwagon?

6 Summing Up

Are there disadvantages to electronic journals? Of course there are. Should we advocate abandoning the digital age because of those dis­advantages? Of course not. And in any case, the digital age is here to stay, no matter what we advocate. But as we move forward into the new age, we need to make clear the principles that should underlie our scholarly literature.

I believe in formulating clear policies about the integrity of scholar­ship. I believe in fighting to keep decision making in the hands of schol­ars. I believe that no one should own ideas for too long—scholarship belongs to all of us and moving walls should be universal. And I be­lieve that scholars, and not accountants, should measure the value of scholarship.

But I believe most of all that we must stay focused on these real problems as the world changes around us, and not be distracted by fads—especially fads that will hurt mathematics (and scholarship more generally) in the long run.

The digital age will dramatically change the way in which we dis­seminate scholarship in the future . . . but we have to guide that change so that it makes things better and not worse.




[1] Based on a talk given at a conference in Aveiro, Portugal, 2006.
[2] The survey collected data from 89 of the 123 member libraries of the Association of Research Libraries during November and December of 2005.
[3] These data were derived from Mathematical Reviews.
[4] The term of 28 years was standard for more than two centuries after copyright was invented, and therefore has historical meaning. See Notices of the AMS 51(3), March 2004, p. 309.

Thursday, June 1, 2006

Measuring Journals

By John Ewing

Not everything that can be counted counts, and not everything that counts can be counted.
Albert Einstein (1879 - 1955), attributed
The impact factor was created in the late 1950s as a way to measure a journal's value by calculating the average number of citations per article over a specific period of time. Since citations generally reflect the interest of scholars in an article, the impact factor ought to reflect the average interest in articles appearing in a journal. This seems to be a sensible use of citations.

But the impact factor has far outgrown its original purpose. A recent article [4] on the impact factor asserts that "impact factors have assumed so much power ... that they are starting to control the scientific enterprise" and that they "play a crucial role in hiring, tenure decisions, and the awarding of grants." The same article quotes Eugene Garfield, the creator of the impact factor, who laments: "We never predicted that people would turn this into an evaluation tool for giving out grants and funding."

People misuse the impact factor because there are no explicit principles governing its interpretation. The impact factor is used to measure the value of things for which it was never intended (articles and authors, for example), and it is used to make faulty comparisons between unlike objects, including journals themselves.

This is not a new problem. For decades, scholars have complained about...... MORE

(Also published in the Notices.)