Dr. Garcia (Orion): “Link-based models have failed miserably”

Rand has posted an interview with Dr. Garcia (Orion) at SEOmoz. Dr. Garcia is ‘one of the SEO/M world’s foremost authorities on the subjects of the IR (Information Retrieval) field & search engine technology’. Definitely a good read… I found two parts to be especially interesting:

Rand: Are there any standout shifts you see towards greater or lesser use of things like PageRank, Trustrank, VIPS, Local Link Popularity, etc.?

Dr. Garcia: Link-based models have failed miserably. Some few âverticalsâ are using feedbacks and reviews to improve results using Slope One methods and item-to-item collaborative filtering (personalization based on recommendation). Iâm not sure if this technology is ready for general searches. A system based on item-to-item collaborative filtering and that considers usersâ actions as âreview votesâ can be gamed, without not even considering the standard issues (scalability, security, spam etc).

I do agree that link-based algo’s are not perfect, but I have to disagree that they have "failed miserably". For instance, take Google vs. MSN on relevancy. Why is there so much more spam is MSN? Because MSN’s link pop algo’s are so much easier to game. Google’s, on the other hand, are relatively elegant–this is a large part of what drives their relevancy.

Dr. Garcia’s comments on a "system based on item-to-item collaborative filtering" are pretty timely considering Yahoo’s recent release of MyWeb 2.0.

On the future of links in SEO:

Links will always be part of the âranking mixâ but not as much as many think. Contextual associations and semantics (meaning), on the other
hand, is a must for search engines that want to assess the sense of
documents.

I’ll have to take issue yet again. Certainly contextual associations (a la the Semantic Web) will improve search relevancy as they are better integrated. The truth however is that search engines are still extremely weak with this stuff, and, IMHO, the Semantic Web (or anything like it) is still decades away.

Meanwhile search engines will need to use off-site metadata as a heavy factor in their algo’s to stay relevant. And the truth is, there aren’t very many types of off-site metadata that are "universal" and also useful in a WWW search engine algo… besides links. After all, the hyperlink is the basis of the Web… both its format (HyperText Markup Language), and its protocol (HyperText Transfer Protocol).

Until we have a new "type" of World Wide Web, link building is here to stay as the main focus in SEO.