Current thinking on Google

[Brian Turner]

This is from a post on another site, but I included a couple of my key ideas in it, so I've reposted it here in case anyone wishes to make comment.

Originally Posted by Mel

From where I sit it appears that Florida did go away, unless of course through some miracle a modified Florida algo produced exactly the same rankings as Pre Florida, which was my experience.

I would be interested in any evidence you have to support your contention that multiple links from the same site are filtered in some way, as I simply do not think this is happening.

For instance the Google link: search shows 23 back links from pages in my site, and those are all certainly in the same IP address block, and a search of the backlinks of this forum show lots of links from the same site.

Of course, the Florida update has been and gone and Google has changed since - but Florida, so far as I recall, was never about Google implementing a single change, but multiple changes. Some of those changes remain with us, such as word order being much more significant, and I'm seeing Google show favouritism for expert and authority rankings - - elements that really came into their own under Florida.

However, really, the theory of it all matters little. Whether anyone agrees with my observations or not, and I'm certainly not going to make an argument for them as it's really quite extraneous to practical SEO in general. The core concern is simply about link-building and I know that's what we can all easily agree on.

Since I started in commercial SEO I find that all the theory and latest SEO news is almost meaningless - my clients want ranks, I build links to give them ranks, and they get ranks and are happy. Although I try to engineer links with a concern for how the different algos operate - Hilltop, LocalRank, TSPR etc etc, so that I can try to pre-empt any new "Florida" (or, "Yorkshire ") it doesn't really matter what the nuances of SEO theory are at any particular time because the core principles still work, and nowadays that's all I'm bothered about - what works, rather than what people think.

As for links being devalued across IP - I mentioned it as a suggestion, from certain observations by myself and others of some form of devaluation - but I neither offer a "proof", nor care to waste my time trying to construct one. We know Google can devalue by IP, therefore I engineer links to take this into account. It makes sense to. If others don't want to then that's their decision. I may build a few thousand links on any particular domain within a client IBLN, but generally I'm looking for links across many IP blocks first, rather than lots of links across just one or two C classes.

As for using "link:" and the like for tracking down links - you'll find that this is actually a pretty meaningless tool. It may count links of a certain PR, but it certainly doesn't tell you what actual value Google assigns these links. From my own experience, the number of PR4+ links it itself is meaningless for targeted ranking purposes, and any ranking PR gives is easily drowned out by anchor text. And it is extraordinarily difficult to track low PR backlinks based only on generic anchor text. If you don't believe my on that point then try deconstructing the top commercial sites under "viagra" and "cheap viagra".