The Law of Preferential Attachment and the Future of SEO
January 9, 2009 • Glenn Murray
In 2005, I wrote that the Web is fundamentally inequitable. I said that those with money were more likely to dominate search engine rankings because they could invest more in quality content. Nothing profound in that, I know; not even back then.
What’s interesting is that I was really only seeing part of the picture. I’ve since learned that it’s not just money that sustains this dominance. It’s something much more powerful: nature.
Specifically, I’m talking about the nature of networks. Let me explain. Scientists have recently discovered that the World Wide Web evolves according to some pretty set-in-stone laws that are common to all networks. One is called the Matthew Effect or the Law of Preferential Attachment. It says that in any network, well connected nodes are “more likely to attract new links, while poorly connected nodes are disproportionately likely to remain poor”. (Quoted from Mike Grehan’s great article, Filthy Linking Rich.)
From a WWW perspective: “new links on the web are more likely to go to sites that already have many links” – i.e. the hubs. As a result, so too are the high rankings. In other words, Google’s dependence on link popularity is “inherently biased against new and unknown pages”, and that, as a result, “[q]uality and relevance are sometimes at odds with each other”. (Again, from Filthy Linking Rich.)
Of course, I knew this back in 2005, even if I attributed it to the power of money, rather than the power of nature. And it’s not this revelation about today’s SEO that prompted me to write this post. What really prompted me were my thoughts on what the Matthew Effect might mean to the FUTURE of search engine optimization (SEO)…
Most SEOs agree that the importance of link popularity is waning. I talk more about this in my SEO ebook, but the essence of the theory is that visitor factors will ultimately play a bigger part in the ranking mix. Things like how many people visit your site, how long they stay, how many pages they visit, how often they come back, and whether they buy (or whatever your conversion metric is).
It’s argued that this changing of the guard will ultimately result in search rankings that more accurately reflect both relevance and quality.
But in light of the Matthew Effect discussed above, I’m not so sure.
The important thing to realize is that even on the Web, the word “connection” doesn’t have to mean “hyperlink”. Loyal visitors and conversions are connections too (in the same way that friendship is a connection in a social network). So, just as a hub site is more likely to attract hyperlinks – thanks to the Matthew Effect – it’s also more likely to attract loyal visitors and conversions.
(Common sense tells us this would happen too, although we might normally be tempted to put it down to brand awareness, authority, credibility and trustworthiness.)
In other words, the changing of the guard from link popularity to visitor factors may be simply a case of ‘out of the frying pan and into the fire’. (How’s that for mixing your metaphors?!) While I agree that a search algo based on visitor factors would be harder to manipulate, and therefore more capable of assessing quality and relevance, it would still favor hub sites, so we’d still be in a place where relevance and quality are sometimes at odds.
We’re simply substituting one reflection of the Matthew Effect for another.