• Register
• Login

SEO Theory - SEO Theory and Analysis Blog

Algorithm analysis, Web community relationship analysis, SEO practices and techniques, industry news, etc.

• Home
• About SEO Theory
• SEO Glossary

The mechanics of linking

Posted by Michael Martinez on April 25, 2008 in Link Theory

There are still people today who insist that search engine optimization is all about links and IP addresses. Although they are full of crap they can usually sway a few gullible people, which is a shame because that means the endless debates about links-versus-content will continue ad nauseum.

Despite what some people would have us believe, search engine optimization is not all about linking. Nor is linking even the most important aspect of search engine optimization. I don’t feel there IS a “most important aspect” to search engine optimiztaion, although I would certainly agree that while there are some things you can live without linking is something you need to live with. Still, if it really is just “all about links”, why is it that everyone who writes an SEO tutorial, guide, book, class, course, test, or whatever feels compelled to cover non-linking topics?

If I could solve the Web’s search ranking problems with links, this blog would be called “SEO Linking Theory” or something (and let’s see how long it takes some enterprising domain-keyword-spammer to gobble up those domains). SEO linking theory is actually a valid topic for discussion and that is why I’ve included a link theory category here on SEO Theory. SEO link theory should incorporate the link theory defined by mathematics but I’m not in a position to do that because I haven’t studied link theory as defined by mathematics (I have studied number theory, set theory, and several other topics, though).

It is appropriate to talk about “SEO link theory” with a very specific meaning and I think I’ll adopt that convention as of this posting to help distinguish what I write about from what the mathematicians write about.

Let’s start with some basic stuff so that we all understand what I’m focusing on. Search engines manage linking relationships in one of four ways:

1. They allow links to pass value from page A to page B
2. They allow SOME links on page A to pass value to other pages
3. They prevent page B from receiving value from other pages
4. They prevent page A from passing value to any other pages

Links come in many forms and serve many purposes. A few examples of link forms include:

1. Traditional text links (A HREF with anchor text)
2. Traditional image links (A HREF with an image)
3. Image maps
4. Scripted links
5. Page-internal links (also called “anchor points” or “page anchors”)
6. Form action links

We use links for internal navigation, internal reference, external navigation, and external reference. The difference between navigation and reference is structure. That is, navigational links have a consistent structure that you see across a multitude of pages. You literally assume all the pages sharing the same link structure are part of a group. Although most people are familiar with the concept of “internal navigation” most people don’t often think about “external navigation”.

External navigation is nonetheless one of the oldest forms of linking on the Web. We once relied extensively upon the external navigation resources provided by Yahoo!, DMOZ, Looksmart, NBCi, and other directories. Many people continue to create external navigation resources through social bookmarking sites, social media linking sites, etc. Sites like DIGG, SPHINN, StumbleUpon, and other referential resources are actually providing external navigation links.

A link is either navigational or referential, and if it’s not navigational that is only because it lacks structure. That is, the link does not fall into any sort of constructed page section that provides obvious, intuitive structure or order to information. Counter-intuitive as it may seem, a list of links may or may not be structured. I feel pretty confident that most search engineers would agree with that statement because they are constantly looking for structure in the information they index.

Structure is better for search engines, so it’s better for Webmasters, and therefore it’s better for search optimizers. If you can provide a logical consistency to your external links you create a navigational system.

Which is not to say there is anything wrong with referential links. In fact, many people in our community seem to prefer external referential links to external navigational links. But remember that links can be presented in a list format and still have little to no structure. A free-for-all links page may or may not provide structure, but even FFA pages that divide their links into categories could arguably be said to lack structure.

Linking structures depend upon something Matt Cutts often mentions: editorial review. I see no way to escape the inclusion of editorial review in a linking structure. Why? Because I can inject a link to my dog Web site into any unstructured list of links for cat Web sites. Structure is entirely founded upon editorial decision-making. You cannot have structure without purpose, without enforcement of a policy.

Some people realize that policies can be programmatically enforced. Spammers do it all the time with their Markov-chain generated, page-scraping crap. Their external links are nonetheless not structured because they don’t follow a coherent policy. That is, you can create an algorithm that is so loose it enforces an incoherent policy.

In linking structures, a policy is the set of rules or criteria that govern which links are to be included in the structure, how they are to be organized, and how they are to be presented. If you write a program that attempts to build link structures on the basis of RSS feeds or scraped content your policy doesn’t include any criteria to govern which links are to be included. That is, in order to create structure your policy MUST include a review process. Simply scraping RSS feeds, Google/Yahoo! Alerts, or Web pages provides no review process so the policy is incoherent.

To put it another way, links don’t express opinions but the policies behind linking structures do. That is the crux of the disagreement between the SEO community and the search engineering community over link citation. I don’t think most people would describe the debate that way, but at its most fundamental level the argument concerns where the search engines should be looking for the opinion that drives the links.

I think Matt Cutts and his fellow search engineers — if they were to phrase a discussion in the terms I have laid out — would put forth the position that their algorithms and filters do seek out the policies behind linking structures. While I am not in a position to put words into their mouths, in my opinion the search engineering community has been only partially successful in building technologies to ascertain probable linking policies. Still, if you look at what they are doing from this perspective, it’s more impressive than how Larry Page and Sergey Brin put it in their “Anatomy of a Large-scale Hypertextual Search Engine” paper.

Linking policy is the funnel for editorial review. Now, many paid link advocates argue that editorial review does go into their linking programs, and while that is certainly true that editorial review was never really part of a linking policy. Does that mean SearchEngineGuide can or should set up a paid links directory like Yahoo!? I don’t see why not but then Google is under no obligation to allow such links to pass value because, frankly, a linking policy also includes intent and I think the search engines are all pretty good at figuring out intent in many cases.

Why does the Web need another paid submission directory? What is the purpose of the directory itself? If it’s just a generic directory then clearly someone is only trying to make money or otherwise capitalize on a well-established practice. If, however, the intent is to provide a new perspective, to innovate, then I think there is sufficient additional value for the average Web user to say that the linking policy is justified as well as coherent.

Linking policies work best when they are designed to enhance the user experience. You can articulate your linking policy but I would be surprised if a search engine decided to allow links to pass value simply because you create a “Linking Policy” page (and several thousand Web sites do have linking policy pages, believe it or not). A good search engine is not interested in what you say about your linking policy but rather in how you execute it.

We all have some sort of linking policy. If you analyze links from SEO Theory you’ll find a fairly consistent set of patterns. There are only a small number of sites I’ll link to from these blog posts and a large number of sites that I won’t link to. Most of the sites I link to provide additional information about the topics I discuss. Some of the sites I link to illustrate points I make (most of them my own), and some of the sites provide query results or other incidental information. Once in a blue moon I’ll link to another site as a means of providing recognition but SEO Theory — unlike most other SEO blogs — does not link out liberally to other SEO sites. That is, this site is not a linking resource.

Xenite.Org, on the other hand, IS a linking resource but its links are chosen carefully, organized carefully, and presented for specific reasons. Most if not all the feature articles on Xenite include links to truly relevant, related sites (although my definition of relevance may be either “other (important) pages in the Xenite network” or “other pages with similar information”). Xenite uses a sophisticated navigational structure that varies by topic, section, or timeframe (older content may use different navigational structures). But Xenite also links out to many other Web sites to help visitors find more information.

Linking policies can be simple or complex, but complex policies still need to be coherent in order to be most effective. Many people in the SEO industry have noted that you can often slip in a few questionable links if you have a resource that provides a lot of good links to trusted sites, but I think there is more to it than that. I think you still need your own trusted links in order to be trusted enough to get away with that.

Link building is also driven by policy, believe it or not. That is, when you build links (as opposed to earning links naturally) you go out and create links on other sites according to some set of criteria you have given yourself. This is where all the cheap “Top Ten Linking Resources” articles SEOs like to write are put to use. You can easily create a very visible, canned link profile for a site by following linking advice from SEO blogs and forums.

Your link building policy can crown your hard work with success or bring it down in complete humiliating failure. Your building policy determines where you deliberately obtain links. It cuts a swathe through the Web like a lawn mower plowing a trail through 2-foot-high grass. You may feel you’re getting somewhere but from a higher level you create a very distinct pattern. Some patterns are acceptable and other patterns are less acceptable.

One of the things that makes a link building policy’s footprint acceptable to a search engine is the level of editorial review incorporated into the linking policies of the sites where you obtain links. The less discriminating you are about where you accept links from, the less discriminating your linking partners tend to be. And the less discriminating your linking partners tend to be, the less reason search engines have to trust those links.

It seems a bit circular, the way I’ve presented it but I think this description of the process sounds more familiar to people. We’re trapped in the conundrum of having to be as nit-picky and choosy about our links as the people we would like to have link to us. The thing is, most query results are not decided by links, so why do people obsess over links? You need just enough links to let your content be fairly scored on the basis of its relevance to the user’s query. How many is enough? That depends, we don’t know, but less is better in all cases because the last thing you want is to participate in queries that are so hyperoptimized you need thousands of value-passing links.

Your link building policy should include criteria such as: visibility (sometimes called “reach”), traffic (potential referrals), credibility (often associated with relevance but not quite the same thing), authority (is the source actually knowledgable enough to make a good recommendation?), and trust. Your link building policy should NOT be concerned with what SEOs tend to call “link authority” or “authority sites”. That is, if you’re seeking links from sites that have lots of links pointing to them because they have lots of links pointing to them, you have a weak — possibly incoherent — link building policy.

A link building policy is incoherent if it doesn’t provide structure in a link profile. Think of a link profile’s structure as whatever criteria are necessary to define a network, community, or neighborhood. Your house is always on the corner connecting several neighborhoods, and so it defines its own neighborhood. It’s important to understand that if you don’t control where you seek links by being selective that the links you obtain will generally be poor in quality. Just because some SEO toolbar tells you a page has lots of value doesn’t mean squat.

We’ll come back to the mechanics of linking another time. There is still a great deal to say on the topic.