We're Doing Anchor Text all Wrong - A Search Theory

I think that nobody would argue that the major search engines are getting much more intelligent. The sophistication of the kind of data and understanding they have of the web is getting to HAL levels (from 2001: A Space Odyssey). Based on some reading of new white papers and technology theories being spun around the web, I have come up with my own theory about the new interpretation of anchor text and the use of semantics and keyword stemming. For the example I'm going to use Google, because I believe they are the only ones who might currently be implementing this idea, although in the future it could very well become more widespread.

What if Google uses anchor text to theme sites into a hierarchical category structure?

Say you were looking for a Motorola V66 Cellphone - Rather than rank sites that focus exclusively on Motorola Cellphones or Motorola V66 Cellphones, Google shows you sites that are focused on Motorola products or on Cellphones in general. They pick thee most important sites that deals with cellphones as a whole, rather than a site that is a clear leader on Motorola V66 Cellphones.

Why?
Because they figure they're providing more relevant data by showing you a big, authority site that has tons of information on cellphones rather than a site that's uniquely focused on the V66.

Anchor Text
Google uses the anchor text that points to the site (as well as the content and the content of the pointing pages) to determine what your site is an 'authority' on. So, by using the anchor text "Motorola V66 Cellphones" you're actually hurting yourself. Sure, one or two instances of that particular anchor text might help, but it would be far more beneficial to get links that say "Motorola Cellphones" or "Cellphones" with other related keywords.

You MUST be an authority
Niche players will eventually fall by the wayside if this method of search rankings prevails. It's inevitable that the web will grow to 10 times its current size in the next few years. In order to be competitive, a site will HAVE to be an authority on a semi-broad subject base in order to rank well for subtopics.

Google could do that?
They could. Google (or another engine) could eventually develop a complete hierarchy for every type of subject matter known and add to it as new ones become known. They may even have a system in place like this right now that can recognize that a search for the Motorola V66 Cellphone is just a subcategory of Motorola Picture-Taking Cellphones, which is a sub-category of Picture-Taking Cellphones in general. This kind of categorical structure could give Google the power to decide that the most relevant site for a query is some arbitrary number of levels deep.

Agree, Disagree, think I'm Crazy?

I think you have good points. Evolving with the SEO whiz's is a must, otherwise they would eventually be worthless. With the "google bomb" being plastered all over the net and bush becoming the "miserable failure" of the internet, I think Google had to re-think their anchor text theory. I still think anchor text will always be a heavy weighted technique, but they will ensure the site being linked to has something to do with what the text states.

Is this a theory of what you think may happen or what you beleive is happening?

If the latter, im sorry but i'll have to party poop This kind of speculation just doesnt have legs without some demonstrable basis in fact.

Having said that, it was an interesting read with my sat. morning coffee, thanks!

Nick

This is pretty close.

After that - you are observing incorrectly, or possibly misintrepreting messages I send to you and/or results I am helping you produce (not sure which).

If the latter two - this is merely fresh crawling and has absolutely nothing to do with normal algorithmic processes and update.

It far better to: observe, conclude and then attempt to prove yourself wrong "at all costs" than guess "this must be what's happening" and claim it so.

3 problems that occur - more often than not you are wrong...

but believe you and correct...

and others start believng it "incorrectly as well".

You MUST use scientific methods to prove or disprove all theories and disproving is far easier than proving.

BTW - Nick_W is, thus far, by far the best CSS guru I've come across... don't let him escape without getting some tidbits! ;)

Nick - I don't see it being used right now, or if it is, it's a minor player in the big picture.

I don't have the resources or time neccessary to do real experimenting that could stand up to scrutiny, but like all of us, I watch the SERPs and my brain picks up on trends and features.

I think all of us have noticed that in many segments (themes/flavors/etc), certain sites dominate whenever they have the keyword phrase in the title. The concept of 'authority' sites is certainly well-documented, if not fully proven.

Fathom -

"It far better to: observe, conclude and then attempt to prove yourself wrong "at all costs" than guess "this must be what's happening" and claim it so." - This I agree with, although SEO boards are all about 'speculation', right?

"more often than not you are wrong..." - Ouch... Well, at least I guessed the right day for the PR update - not that it helps me, or anyone else for that matter.

and others start believng it "incorrectly as well" - I hope not!

People here and everywhere should take what they read, especially when it's presented as a 'theory' with a grain of salt. They should keep it in the back of their heads, and if it turns out to be true, and they start finding facts to back it up - great, if not, at least we're all on our toes. I apologize if you think I'm spreading bad info with my theories - If you think it's a good idea, I can try to be more reserved with what pops into my head in the future and not spread it around...

I assume you're rebuking me for my 'March Filter' theory as well. I can't back it up with proof, but I still think it has some validity.

it wouldn't surprise me if the SEs move in this direction, but I don't seem any evidence of it yet.

randfish - I don't mind reading your theories at all, even if they are sometimes a bit premature. After all, it gives us (dare I say) SEOers a chance to comment, experiment, add or change. Or dismiss. Just because there is no evidence (yet), doesn't make the theory false. So please keep on posting them here, it sure saves me the effort to conjure them up myself ;) ...and to write them down so that everyone understands.

Hope people will join in every time to try and prove, or disprove, them. Preferably with lots of examples as back-up...

P.S.: As long as you mark your theories with the lightbulb/crystal ball, then people will know what to expect

I think your theory is not taking the step it needs to. Break it open and think bigger.

Your theory is interesting. Just a thought: doesn't your theory fly in the face of the fact that search queries have gone from single word searches to search phrases?

Most searchers are moving towards specificity driven by longer search strings primarily because of the increasing sophistication of the engines. I doubt if anyone would be happy to request a specific cellphone model (via a long search string) only to be served up megasites that covers all cellphones.

We love G because it serves up somewhat relevant site results for a multi-word search; those searches controlled by how specific the user is being through the use of longer and longer search statements. The SE dream would be an intelligent search engine that knows through plain language descriptions (maybe even spoken descriptions) exactly what a user is looking for and presents it (not Ask). As I said, it is just a thought. :

seo-am,

Sorry, maybe I didn't explain it well - what I'm saying is that GG picks out a page about V66 phones from a big site about cellphones in general, rather than a site about only V66 phones.

randfish,

Got it, but not overly excited.