PageRank and Links as Votes:

Larry Page and Sergey Brin publicly pioneered the use of link measurement as an indicator of search relevance when they created the Google search engine. The initial idea behind their legendary PageRank system was that a link to a particular page is equivalent to a vote by the linking page for the linked-to page.

As the theory goes, by measuring not only the keyword relevance on a page, but also the number of “votes” it had, you could accurately determine which pages were considered by the web’s users to be most valuable for the given search terms. Furthermore, links from web pages with high PageRank would be more valuable (considered more authoritative and reliable) than links from low PageRank pages.

This initial model, while revolutionary, didn’t consider the quality of on-page content, trust metrics or semantic relationships and was thus extremely vulnerable to manipulation.

SEOmoz homepage with Google Toolbar showing Page Rank 6:

Improvements in Link Quality Scoring:

Over the years all of the major search engines adopted the link-based ranking model with some stylistic variation from engine to engine. The technology, quality and “intelligence” of the various search algorithms continually evolve in an effort to improve the quality of returned search results.

Anchor text is now considered when evaluating the relevance of a given link to the given keywords. If a site about Seattle Boat Tours has lots of links pointing to it with anchor text written as ‘Seattle boat tours’, those links will provide greater value to the link recipient than links with anchor text such as ‘click here’ (at least for searches on “Seattle Boat Tours”). We will discuss anchor text in greater detail in the next section.

Semantic attributes the of on-page text surrounding links is also analyzed. For instance, if a link to Seattle Boat Tours is in the middle of a page about theoretical physics as it relates to the study of Scientology, that link won’t be considered as valuable as the same link with adjacent content about Seattle, tourism, boating, etc.

Seattle.gov tourism page, this would be highly relevant for ‘Seattle Boat Tours’

Search algorithms also consider relationships between linked sites. By analyzing things like IP addresses, reciprocal links and domain registration information, the engines try to determine if the links are valuable organic links, or if they are manipulative, artificial links created solely for ranking purposes.

Links as Quality Control:

By using the methods discussed above to measure link quantity and quality, the search engines create a ranking system that is much harder to manipulate than one based solely upon on-page factors. That is not to say, however, that page structure and actual content are not evaluated. The link-based model simply places more importance on links based upon the theory that only well-designed, content-rich pages will get high-value links from reputable sources.

Trusted Domains:

Link factors such as anchor text, semantic relevance and page relationship certainly matter, but perhaps no factor matters as much as the trustworthiness of the domain providing you with your link. A single link from CNN or The New York Times is worth more “link-juice” than dozens of similar links from no-name blogs and MySpace pages.

Trusted domains have proven over time (ironically, through the acquisition of thousands of trusted links) to be worthwhile and reliable sources of quality information about their given subject matter. As such, when the search engines see that these sites link to you when discussing your area of focus, they pay attention. This tells the engines that a reliable and trusted source thinks you are an expert and you offer content that’s extremely relevant to the topic.

Think about it this way: If the Weekly World News runs the headline ‘Two-headed Dragon Boy Born in New Jersey,’ would you take it seriously? No? What if the same headline was on the cover of Time magazine? In nearly everyone’s mind, the Time link to ‘Two-headed Dragon Boy Born in New Jersey’ is far more valuable and credible than the same story from the Weekly World News.

This is why we don’t go into a state of shock when we see bizarre WWN headlines every week in the grocery checkout line: they’re simply not reliable for anything other than a laugh. In the online world, the search engines use their artificial intelligence algorithms to make similar determinations. Thus, 50 links from Moe-does-Mortgage.com is not nearly as valuable as one link from Bankrate.com or CNN’s money.com.

This is a self-perpetuating process on the web. The more trusted, valuable links your site receives, the more trusted and valuable your site (and the links you give) becomes.