Alex Wright

Palimpsests and Page Rank

June 9, 2006

A few weeks ago I posted an entry about Eugene Garfield's notion of the "obliteration phenomenon," the bibliographical paradox in which important original works often get eclipsed by subsequent derivative works. The canonical example is Einstein's Theory of General Relativity, which almost never gets cited in the footnotes of scientific research papers*. Another example is Archimedes' discovery of the number now known as pi (a term not coined until 1706); you will almost never find a reference to Archimedes in a mathematical research paper, even though his contributions are fundamental to all of mathematics.

In my original post, I suggested the possibility of a corrollary phenomenon on the Web, where our reliance on explicit influence-weighting mechanisms like Google Pagerank and other populist measures seems to create a misleading picture of how cultural influence really works.

A couple of days ago, I received a note from Dr. Garfield himself, who had somehow stumbled across my earlier post, pointing me to a more in-depth column he wrote on the subject back in 1975, "The 'Obliteration Phenomenon' in Science - and the Advantage of Being Obliterated!" Here Garfield gives several more examples (like Archimedes) and explores the mechanism in more depth, invoking his mentor Robert K. Merton's earlier work on the notion of "palimpsests."

The term palimpsest refers to a piece of parchment used for manuscript copying more than once - earlier texts having been erased to make room for some newer work. The erased texts can sometimes be deciphered by close examination and special techniques. Thus, like a trace of writing which is rubbed out to make a place for a newer message, certain scientific documents are obliterated in order to make way for citations to more pertinent, less well-known, or more modern papers.

It is perhaps a little surprising to find Garfield - the father of citation analysis, whose work directly paved the way for Google - arguing for a more balanced and nuanced view of citations. "Neither too much nor too little attention to citations will do," he writes, "what is needed is just enough."

Unfortunately, such moderation is in short supply on the Web, whose great strength is the sheer explicitness of citations - that is, links - between documents. This strength is also its fundamental weakness. The easy visibility and measurability of links creates a false picture of influence, privileging the explicit over the implicit, and constantly threatening to mask the murkier machinations of human culture.

By way of example, let's consider a few recent memes du jour on the Web: Web 2.0, AJAX and the Long Tail. In each case, these terms have become closely associated with particular authors. Web 2.0 "belongs" to Tim O'Reilly (at least according to the recent lawsuit); the term AJAX was coined by Jesse James Garrett in his widely cited essay, and "Long Tail" seems to be squarely associated with Chris Anderson, whose blog and forthcoming book have linked the term inextricably with his name. Despite the overwhelming Web consensus that associates these authors with these inarguably important developments, no one would argue that each of the phenomena they described actually have much deeper roots. Of course O'Reilly did not really invent Web 2.0; Garrett coined the term AJAX to describe a set of technologies that had already been around for years; and the use of the term "Long Tail" as a way of thinking about Web markets can actually be traced to an earlier essay by Clay Shirky. Yet in each case, the Web's "paper" trail would suggest a clear-cut correlation between each meme and its ostensible inventor. This pattern seems to bear out Merton's hypothesis:

Naturally enough, most of us tend to attribute a striking idea or formulation to the author who first introduced us to it. But often, that author has simply adopted or revived a formulation which he... knows to have been created by another. The transmitters may be so familiar with its origins that they mistakenly assume these to be well-known... And so it turns out that the altogether innocent transmitter becomes identified as the originator of the idea.

Well, exactly. None of this is to suggest that the above mentioned people don't deserve credit for their work; but surely none of them would claim quite the level of credit that Web search algorithms seem to award them. Here, I think, is the point: to the extent that we rely on too literal a reading of link influence - that is, if we subscribe blindly to the Google world view - we end up with a skewed understanding of how the world really works. And that, in a thousand small ways, is what is happening to our culture every day on the Web.


* UPDATE 6/15/06:
Dr. Garfield sent me a note pointing out that citations to Einstein have actually been increasing in recent years.

File under: Informatics

« Brooklyning | Little Green »


Glut: Mastering Information Through the Ages

Mastering Information Through the Ages

New Paperback Edition

“A penetrating and highly entertaining meditation on the information age and its historical roots.”
—Los Angeles Times     

Buy from