PageRank, worth it or not?
UDPATE: Andy Beard posts about more PR craziness. While I think it’s unfortunate and makes PR more unreliable, I still don’t believe that it is worthless - comparing the number of heavily affected sites vs. the total number of sites on the internet, I think only a very tiny portion are really affected. What is interesting, though, is the noose Google makes you hang yourself with trying to get your site reconsidered if you’ve been punished. It’s like if you were arrested, in order to get to a hearing you had to sign a confession that you committed the crime and are asking for an early parole. That’s definitely wrong and needs fixing. Crazy.
With the recent Google PageRank craziness the internet is all twisted panties about what to do. Personally, I’m on board with this guy who believes that all this call for getting rid of PR is a little extreme… On the other side you get the folks at EatonWeb who are much less impressed with PR (via).
They’re a blog directory who ranks blogs according to a variety of metrics which they mix up into their own ranking system. Now they’ve said they’re phasing out PageRank from this metric over the next year - so it will gradually feature less and less in their system till it’s gone. They criticize PR thusly:
Google has been systematically introducing the equivalent of theoretical epicycles to its display of PageRank to the public, and we think it’s about time to face the facts. You can’t manually penalize hundreds of influential sites and expect to be used as a reliable source of information any longer.
Here’s a long post on what all this epicycle business is. Basically the complaint is that within the natural to and fro of G’s PR algorithm changes, they are manually busting some sites down a notch - penalizing them for perceived breaches of Google policy.
This notion in itself is a whole different topic that I am not immersed in SEO enough to have a reasoned opinion. But the thing that caught my eye was the “expect to be used as a reliable source of information” part. That is, if you believe, as I do that PR is still relevant (that is, if Google itself pays attention to it in some fashion, if not as doggedly as before) then it provides some information if not as reliably as before.
And if it doesn’t measure up to some internal bar for reliability, I was really curious to find out what other reliable metrics they were using to develop their own ranking system, as far as I know, there aren’t really any widely available ones that had good information. Quantcast would be one, but very few sites actually use it. So I hunted around EatonWeb’s site and found this page which doesn’t specify their algorithm but provides historical graphs for a variety of metrics which I assume are the ones they are blending up into their own.
There’s nothing surprising in there. The first one is one “site visibiity” I don’t know what that is and couldn’t find any description of it. Then it goes on to all the usual suspects technorati, alexa, page rank, delicious, and links. These are the stats everyone uses to figure out how good things are. BlogJuice does the same. They’re all terrible metrics.
Alexa? I mean, that sets the bar pretty darn low. Is PR really more meaningless than Alexa? Alexa has all kinds of bias problems and panel self-selection problems and general problems with providing meaningful stats. It’s the same with all the metrics, it’s really easy to game any of them. Each provides a very broad view of a site which becomes arguably more accurate the higher the ranking.
So just like you could probably (although arguably not) infer something from a site with an Alexa rank of 1000 versus one of 100,000 - you can probably infer something from a site with PR8 vs PR4. And just like you probably can’t infer something from Alexa rank 1000 and Alexa rank 2000, you can’t infer something from PR5 and PR4. The difference, though, is that if Google is telling you something is PR4 that used to be PR6, at some point Google might decide to actually decrease it’s prominence in the SERPs and thus reduce traffic. It may have (or have had) some actual relevance to real life, where Alexa is completely divorced from any sort of reality.
Just seems to me that busting a metric for reliability when all you have are unreliable metrics is an odd thing to do. When that metric is the only one that comes from a company that so many sites live and die by, I don’t know, it seems crazy to me.
