I recently argued that the proliferation of self-links (like that one you just passed) on prominent sites and blogs is one of the factors that has made hyperlinks less useful in determining search results and, in general, in sorting, filtering, and evaluating the content of the web. A link is most valuable, in this regard, when it is an expression of an individual’s informed judgment — unbiased by self-interest — about the value and importance of content on the web. As many have pointed out, when links are used in this way they serve as “votes” about the quality or importance of web pages. But as the intent of links has shifted, particularly on popular commercial sites, away from expressing personal judgment and toward pursuing personal or institutional interest (boosting page views, ad impressions, or search rankings, for instance, or even simply using your own or your colleagues’ past work to provide context to readers), the utility of links as democratic votes on the value of web content has been reduced. That doesn’t mean that all self-links, or “internal links,” as they’re often called, are bad — they’re frequently appropriate and helpful — but when they are used routinely, as they tend to be today, they make links in general less valuable as a means of filtering and navigating the web.
A new paper by Mark Coddington, a journalism grad student at the University of Texas, shows just how dominant a practice internal linking is among top journalism sites. (I found the paper via a Poynter story, which I found via an Andrew Sullivan link.) Coddington examined the links in a sample of stories from three types of journalism sites: news sites run by big media companies, blogs written by journalists at big media companies (“j-blogs”), and popular independent news blogs. He found that 91% of the links in the stories on the news sites were internal links (pointing to other pages on the same site), and that 54% of the links in the j-blog stories were internally directed. In contrast, only 18% of the independent bloggers’ links were internal links. I’ve termed the practice of inward linking “nepotistic linking”; Sullivan uses the more memorable term “linksturbation.”
Coddington also measured the percentage of links in the stories that pointed to “mainstream media” sites. On the news sites, fully 93% of links go to mainstream media sites. For j-blogs, the figure is 77%, and for indies, the figure is 33%. As Coddington observes, the overwhelming tendency of mainstream media sites to link to mainstream media sites tends to concentrate authority on the web. Whereas links once served to broaden the conversation, they increasingly serve to narrow it. Coddington, who supplemented his analysis with interviews with journalists, notes that
those within news organizations overwhelmingly expressed philosophies of openness regarding the sources of their links. They placed very few restrictions on what types of sources they would link to, and they were emphatic about their willingness to link both outside of their news organizations, and outside of traditional media sources. As we have seen and will examine further, however, these linking philosophies have yet to be borne out in the actual linking practices of mainstream news organizations, particularly outside of their blogging content.
The sites are open in theory, but largely closed in practice. Blogs once provided a counterforce to such homogenization, but as personal blogs have been displaced by commercial ones (in terms of traffic) over the last seven years, the blogosphere has come to amplify the insularity effect. As Coddington’s study suggests, institutional bloggers are far more likely to link inwardly and to link to mainstream sites than are independent bloggers.
Although Coddington notes that “links can wield immense power to define the parameters of an online text and the Web itself,” the focus of his research is on journalism, in particular how links serve to shape news reporting and hence “frame” public perceptions. He sums up his findings this way:
Inside news organizations, a link is predominantly a tool for providing context, a largely internally directed reference for curious readers hoping to delve deeper into an issue in the news. It points primarily to undated sources and general pages, reaching outside of the day-to-day developments of a news story toward a general, static body of knowledge from which to draw a fuller sense of the environment in which the story is occurring. […] But the logic of this linking practice also circumscribes the frame of the news story, just as it contextualizes it. The body of knowledge to which a news organization’s links point is, by and large, accumulated by that news organization itself and others like it. This is consistent with previous findings that news organizations primarily link internally, and this practice also locates the nexus of online authority largely within the same institutions that constitute it offline.
Links carry not only meaning and context but also ideology. Where links were once celebrated for their ability to undermine old, centralized power structures and information sources, the way they are used today increasing seems to be reinforcing those structures and sources. As they become more exclusive and less inclusive, links themselves turn into a mainstream medium.
Any search engine worth its salt will filter out self-links. Before a search engine can crawl a site, it must know its URI, and so can use this to filter out self-links from its evaluation mechanism.
That’s true, but self-linking (and other such linking practices) has two sides: commission and omission. Search engines can filter out (or give lower weight to) a self-link, but they can’t take into account the links that weren’t written because of the bias for self-linking. So we’re talking about not only “misinformation” but “missing information.” (The search-engine issue is separate from Coddington’s framing discussion, of course.)
UPDATE: Also, I don’t think it’s true that, say, Google ignores self-links entirely, at least judging from the stress on the practice that comes from SEO consultants. For instance: “it is easy to overlook how internal linking is important. If you’ve heard that inbound links are like other sites voting for your content and telling search engines what your content is about, internal links are like voting for yourself and also letting the search engine know about your vote.”
As a secondary effect, as links within articles on big websites increasingly become self-links of dubious quality, readers will prefer search engines to find more relevant articles. And “search engines” means Google. So the uselessness of author links indirectly empowers Google…
None of this is surprising though. It’s unrealistic to expect a majority of authors and publishers to provide value for people other than themselves. Keeping search engine algorithms fair is our best option.
Interesting paper with some pretty telling numbers. To reply to the conversation above me, it’s true that search engines don’t weigh self-links as highly as those from another site, but linking to your own pages is definitely a major part of SEO, as noted in the Hubspot post.
I think another part of the puzzle, though, is that many times sites add internal links solely to get all their pages crawled better by Google; otherwise there wouldn’t be a link at all. Although the implications of linking for algorithms instead of humans is troubling, I don’t think major sites are always omitting smaller ones by consciously linking to themselves in a biased manner, they are simply adding self-references after writing an article because their SEO told them to. (I know there even are some blog plugins that automatically insert internal links.)
Not that this doesn’t still reinforce centralized power structures, though.