What Price PageRank? Part 2

 

When I wrote on PageRank (PR) a couple of weeks back, I had no idea I’d be revisiting it so soon.

Because I’ve been on the road most of the time since then, I’ve not had much opportunity to go through all the various forums, blogs, and newsletters that passed comment. However, a number of friends and colleagues forwarded snippets and links, which helped me get a general feel for it.

Amused? I certainly have been by some of the comments I’ve seen.

It appears to me some folks in this industry are so emotionally attached to that little green gauge on the Google toolbar, the mere thought of dismissing it as having nothing more than novelty value (which I do) is deemed almost sacrilegious.

Ladies and gentlemen, get a grip on yourselves! Don’t be afraid. Let go of your little, green security blanket. It’ll be all right. Trust me — I’m a search engine marketer.

There is life and search marketing without the PR meter. Honest. Truth be told, that’s the way it’s always been.

I don’t want to get too in-depth regarding PR computing. But I have to tell you, I’m looking at a document written by the senior director of data mining at Yahoo It must be the most comprehensive, authoritative survey ever carried out on hyperlink-based algorithms and connectivity data.

But I want to get a touch philosophical first. Bear with me.

What and Who Is PR Really For?

One of the biggest threats to SERP (define) relevancy is spam. Search engines are plagued by it. Ask yourself, “Why would Google, being under permanent spam attack, want to give the enemy such a huge visual, vital clue to the authority of every single page in its index?”

Isn’t this like leaving the front door open and providing a breadcrumb trail the wolf can follow right to your sleeping baby? It doesn’t have to give you this clue, you know. The toolbar and probably Google toolbar work just as well without the PR meter. Other good search engines don’t feel it’s necessary to give you a little mark out of 10 for your Web effort.

Another thing: who’s the PR meter actually intended for, anyway?

Can you imagine some surfer finding the digital camera of his dreams at a knock-down bargain price but refusing to buy it because the page it’s on only has a PR of one? I don’t think so. If you popped into the street now and quickly surveyed casual surfers about PR, you’d probably discover they wouldn’t know what PR was if it bit them on the butt.

That puts the Google PR meter squarely back in the realm of search engine optimizers. How kind of Google to be so thoughtful on our behalf. Isn’t that just fine, dandy, and convenient — it gave us our very own spamometer!

OK, tongue out of cheek. I did an interview in 2003 with Google engineer, Daniel Dulitz. I asked him about data accuracy in the toolbar meter. He told me there are two elements: accuracy and precision. The toolbar is accurate but not very precise. He added, “We have a lot more precision available to us than we represent in a 10-step scale.”

I also asked if he thought search engine optimizers are too engrossed in this PR thing. His reply:

For search engine marketing, search engine optimization purposes, yeah, I’d say that there’s too much emphasis placed on what that PR number actually is…. So, if people are trying to look at what we’re doing and their idea is based on that single thing from 1 to 10, then… well, they’re not going to be effective in figuring out what we’re doing at all.

Dulitz also volunteered if people are casually surfing the Web and they’re looking at the PR of a page to see what Google’s impression of it is, “I’m all for people doing that.” Casual surfers with a passing interest in a page’s PR score certainly suggests novelty value to me.

That was then. This is now.

I don’t use the Google toolbar and have never, ever give a thought to PR when I’m talking strategy with clients.

I simply don’t need a little meter to help me figure out a link from the BBC, CNN, Yahoo, or the busiest, most visible sites in any given community are vastly more important (and valuable) than a link from Joe’s Pizza Delivery Web site. I can figure that out on my own.

If I’m paying for links, I want a lot more tangible evidence from the site owner. I want stats that tell me how visible the links are across all major search engines, how much traffic they send, and how much traffic they attract overall. I want to see the site owner is a savvy online marketer and is an authority in his community or is developing a presence as such. I need to know he understands and uses analytics to provide tactical data. This is sound, useful marketing intelligence. It’s a lot more important to me than a meaningless 4 or 5 in a little sprinkling of green fairy dust above the pages.

A Survey on PageRank Computing

Yahoo Lift Off – Data Conference 1 was a three-day event held at the Sunnyvale, CA, campus two weeks ago. I was delighted, honored, and privileged to be personally invited as guest speaker by Usama Fayyad, chief data officer and SVP of strategic data solutions at Yahoo

This was strictly an in-house event. Speakers consisted of experts from the industry, academia, and various Yahoo business units. The audience was made up of 190 Yahoo employees, executives, data power users, marketers, product managers, and engineering leads.

I had the honor of being introduced to Dr. Pavel Berkhin, author of “A Survey on PageRank Computing,” which he completed last December. It’s a fascinating read. After my presentation, Pavel readily agreed nobody actually uses PR.

For sure, the PR vector exists in the mix. But actually making it work on the global scale it was created for is a very tall order. As Pavel’s paper explains: “It’s the largest matrix computation in the world. Computing even a single PageRank is a difficult computational task. Computing many PageRanks is a much more complex challenge.”

When the original algorithm was written, it was based on the fact the Web was served as a static set of human-edited pages carefully placed in file systems. But the Web has evolved to have more database-driven sites. We’re talking about URLs in the billions.

Because it’s an internal document comprising proprietary information, I’m not at liberty to disclose further content. However, it’s certainly been an eye opener. In just under 40 pages, it references no fewer than 105 research papers and articles related to PR, written by dozens of leading experts from the various disciplines.

After I finished reading it, I was left with this: if PR actually works in the practical sense it was conceived for, why are so many scientists and researchers still slaving away trying to fix it?

Google is a fabulous company and a great search engine. I personally know a number of great people who work there. I’d never knock them. But I would openly, publicly ask them to do me this one small favor: dump the toolbar meter. Some people actually take it seriously!

Meet Mike at Search Engine Strategies August 8-11 in San Jose, CA.

Want more search information? ClickZ SEM Archives contain all our search columns, organized by topic.