Five "W's" and "H" of the Half-Life Phenomenon
Where We Began
Michael Bugeja was fact-checking
the final manuscript of his
Divide: The Search for Community in a Technological Age
(Oxford University Press, 2005), when he found that 30 percent of his
Web-based footnotes no longer functioned on the Internet.
One of the first footnotes to lapse, appropriately, was Microsoft’s mission statement, which he had retrieved in late 2001 from the Web. At the time it stated: “Microsoft and its employees recognize that we have the responsibility, and opportunity, to contribute to the communities in which we live, in ways that make a meaningful difference to people’s lives.”
Two years later, Microsoft’s values apparently had changed. Now it vowed to show “leadership in supporting the communities in which we work and live.”
The change in wording may be subtle, from Microsoft’s perspective, but it corrupted Dr. Bugeja's citation. Any scholar checking his references would wonder whether he fabricated the footnote.
Alas, there was little he could do as an author writing about the Internet. Simply, he had to cite Web pages. He also had to rewrite his book manuscript for Oxford, but this time on the advice of Assistant Professor Daniela Dimitrova, he printed out copies of Web sites used in footnotes and sent two thick binders to his editor to validate sources.
Web pages lapse on the Internet, a phenomenon known as "linkrot." Footnotes are prone to linkrot. A common reason for a dead hyperlink is a message that the page cannot be found—sometimes explicitly saying “404 Error” or just “Page Not Found.” Some sites expand the message to “The requested document was not found” or “The page you requested could not be found.” Some attempts to get to Internet citations inform the researcher that they required a user name and password. Subscription requirements are not too common, but happen often enough to make them noteworthy. Here is an example of a recurring message: “Since you are not a member of the CIOS and your institution does not contribute support for the CIOS through the CIOS/Comserve institutional affiliates program, you are not authorized to obtain files from this server.”
As you can see, the reasons for these errors are many. Some newspapers and magazines, for instance, archive their stories after a brief time on Internet, changing URLs so that users who cite articles have to pay. Sometimes technicians reformat a journal's servers. Sometimes Web pages come down. Sometimes different content is added to the same URL and so on.
What is the "half-life"?
The “half-life” description is based on a metaphor, namely, the length of time required for half of the atoms of a radioactive substance to decay. Michael Bugeja and Daniela Dimitrova employ that term to ascertain the length of time for half of a given number of Internet footnotes to decay in an article. By analyzing each article in a journal, the “half-life” of Internet footnotes in a specific publication also can be ascertained—a measurement distinct from other journal benchmarks, including "linkrot" (a lapsed URL, not necessarily a footnote) and "cited half-life" / "citing half-life" (terms used by Journal Citation Reports), noting the age of received and referenced articles, respectively. It is important not to confuse our term—"the half-life of Internet footnotes"—with these very different terms. See Journal Citation Reports fact sheet. Retrieved July 2, 2004 from http://www.isinet.com/media/presentrep/facts/0108jcrfs.pdf.
When is the "half-life"?
The half-life varies from article to article, from journal to journal, and from book to book. In our first pilot study, we analyzed 2003 paper acceptances in the Communication Technology Policy Division of the Association for Education in Journalism and Mass Communication. To our surprise more than 40% of online footnotes in that division at lapsed within one year and 57% did not operate properly. The half-life of paper presentations was about 1.4 years. We did that study to document the phenomenon as a warning to our Internet-savvy division. Subsequently we analyzed online foornotes in top communication journals, including Journal of Communication, Journalism and Mass Communication Quarterly, and Human Communication Research; in these, the half-life was a little more than 3 years. Other studies cited in our literature reviews show longer and shorter half-lives, depending on the journals and the databases.
Why does it matter?
While the phenomenon of linkrot has been known for years, and a relative few studies have addressed the erosion of footnotes, the research of Drs. Bugeja and Dimitrova add a unique perspective and contribution that set their work apart from others:
Drs. Bugeja and Dimitrova believe that Internet research is vital to scholarship because the medium serves as a convenient electronic warehouse of data accessible at all hours and in great quantities, thereby increasing the scope and breadth of scholarship. However, for scholars to base research on online data, the Internet must be stabilized or a new methodology must be invented. The current scientific method—consisting in part of peer review, source verification and replication—is based on the printing press, which is Internet's opposite. As Dr. Bugeja has written, "The book is 'the ultimate fire-walled medium.' Manipulate a book in the library and you are committing a crime—literally. Internet is all about manipulation, especially in the library; and while that can be convenient, it comes with its own conundrums, with respect to research. The half-life research of Drs. Bugeja and Dimitrova not only attempts to document the phenomenon but also help resolve it.
How to correct
Drs. Bugeja and Dimitrova have a theory: The closer to the book, the more reliable the online footnote. How does that translate into longer online life for footnotes? For starters, if you use a library database to access a journal, use the pdf. version rather than the text/html version. Half-life research shows that the latter corrupts footnotes because datafeeds sometimes insert a space when the printed footnote extends from the left to the right side of the column or page. A pdf., essentially, is a snapshot of the journal page, much like a facsimile or mircofiche used to be. As such, the pdf is closest to the book. Also, be wary of using material from Web sites using less reliable extensions such as “.net,” “.int,” or “.biz.” Relatively stable extensions are “.org,” “.edu,” and “.gov.” The extension "com." can be reliable or unreliable, depending on the domain owner.
Finally, print two hard copies of Web citations (one for the author’s files and one for the editor’s files, to be sent upon request). Authors also might want to save Web documents on disk simply by using the “File>Save As” options in a Web browser or “Print Screen” options on their desktop. These methods not only make available citations upon request by editors and researchers; they also safeguard authors against accusations of invention or inferior scholarship when URLs lapse.