Which way do the threads on the nofollow screw turn?

Somehow, I hadn’t thought of this: in comments on Rogers Cadenhead’s post (which I’m currently unable to get to for one reason or another), someone that I think was Jeremy Bowers brings up an interesting point. I’ve been assuming that nofollowing a link is the equivalent of not having it there: if you have three real links, and one nofollow link, then each of the three real links gets a third of the PageRank that the page has to pass on. Jeremy suggests that instead, it may be that each of the three remaining links gets a quarter of the PageRank.

A quick review of basic PR: give every page that you know about PR 1. Then, start at a page, count up all the links, and divide that page’s PR (still 1, for now) by the number of links. That’s how much it can give to each: a page with four links can give .25 to each. To keep things honest and working, apply a damping factor to all the links, higher for external than internal links: let the external links give .21, and the internal links only .18. Then all you have to do is iterate around through the whole intarweb, at times looping back around to your start page and refiguring what it gives based on more incoming links for it, until everything stabilizes. So, the basics to remember are that the more links you have, the less you give to each one, and links to other sites leak PageRank: you can’t give quite as much to yourself as you can to others, but especially if they don’t link back to you, an external link sends it to someone else, an internal link keeps it for you.

So, what’s the difference? Well, if as I assumed a nofollowed link just doesn’t exist, then by nofollowing all of the external links on a site, it’s possible to conserve all that PageRank for yourself. This isn’t something that most people will do: it’s mean, and cruel, and clearly breaks the web. However, the people with spammy sites that don’t actually include any content at all, only crap that’s designed to suck in search engine users and either show them ads or get an affiliate payment when they click out, don’t care. If they can keep every bit of their PR, they will, and thus they will rise in the rankings relative to sites that actually have external links.

On the other hand, if nofollowed links count against link numbers, but then drop their allocated PageRank on the floor, then Google and friends have handed weblogs a pretty poison pill. If your entry has three internal navigation links, a link in the body of the post, and twenty comments amounting to thirty nofollowed links between links in the comments and commenter URLs, then rather than split that page’s PR in four pieces, Google will split it 34 pieces, three for you, one for the site you linked to, and 30 that it essentially keeps for itself.

If the first case is true, then PageRank really must be dead, and Jeremy Zawodny just declared its death a little too soon. I doubt that’s the case, since I still see expermental evidence that PageRank matters, though it’s hard to tell exactly how much it matters. But in a situation where you can subtly avoid giving any PageRank to others, without having to be obvious about not linking (and after all, how many people realize that in terms of search engine awareness, Wikipedia no longer links to anyone but themselves?), PageRank is dead.

If it’s the second case, well, expect to see not only your commenter’s weblogs, but also the sites you link to, sink in the rankings, because then nofollow is just a way to persuade weblogs to pour out all their PageRank on the floor.


Comment by Shelley #
2005-02-03 21:16:06

Congratulations, you’ve just discovered the real reason for nofollow — to compensate for weblogs disproportionate influence. This was never about spam, but what a great meme to use to spread the use.

Comment by BillSaysThis #
2005-02-03 21:37:33

Also, if this analysis is correct, then comment and trackback spammers have even more incentive to keep up their blather, perhaps increase it, since the lost PR devalues all that given to other links; since rank in a search result set is relative, whatever they can take away means their pages compare better no matter what their pages’ PR.

Comment by Phil Ringnalda #
2005-02-03 21:49:01

Yes indeed, spammers or just blogwars: don’t like what I have to say about something? If I’ve got nofollow turned on, send a steady stream of comments linking to anything, the current changes.xml from weblogs.com will work fine as a source of URLs, and diminish my voice. If I don’t have it turned on, well, there’s no way to hurt a site by linking to it (other than googlebombing), but it’s always been well known in SEO circles that linking to a ”bad neighborhood” (link farms and spamming sites stuffed with keywords and the like) will harm the linking site. So, send me a stream of links to… well, to the same stuff the comment spammers are already sending, since they are bad neighborhoods.

Comment by BillSaysThis #
2005-02-03 21:58:57

Like many other things lately, people who value money over other potential rewards are using the ease afforded by technology to drown out these other rewards. Then again, from my perspective, most of us (bloggers) want good PageRank for either ego (that’s me) or because we want our writings found at the top of any relevant search (which is ego too, I guess). I just never bothered having comments or trackback on my site.

Comment by Jacques Distler #
2005-02-04 00:16:26

If I’ve got nofollow turned on, send a steady stream of comments linking to anything, the current changes.xml from weblogs.com will work fine as a source of URLs, and diminish my voice.

Sorry, but I don’t quite see how this follows. The links that you don’t nofollow get exactly the same fraction of your available PR as before. The random links to ”the current changes.xml from weblogs.com ” (or whatever), which would have received some PR from you, now don’t.

So what? How does that diminish your PR? Only if those (would-be PR-recipients) have links back to you, does your PR suffer.

More likely, the loss of PR gets diffusely spread over the web as a whole. Since only relative PR matters, who cares?

This is not the case for the specific case of Commenter-URLs, which (except for spammers) invariably point to the main page of other weblogs. Then the loss of PR specifically hits weblogs as a group, rather than being more diffusely spread.

OK. But I’m not necessarily convinced that that’s a horrible thing.

More acutely, my understanding is that nofollowing every link on a page is exactly the same as putting a

<meta name="robots" content="nofollow" />

at the top of the page. There are lots of pages I’ve put that on (for sound reasons) with nary a thought for the PR that I just dropped on the floor.

Why should I start fretting now?

Comment by Phil Ringnalda #
2005-02-04 08:00:05

Correct: that was fuzzy thinking on my part. It only makes a difference if you leave comments that you would otherwise remove, and if the total is divided by the total number of links before removing the nofollowed ones.

Of course, since some people leave spam, letting their every post and discussion wind up with a discussion of poker, erectile dysfunction, and incest, there will certainly be some, probably more, who leave things they would otherwise remove, thinking that nofollow keeps them safe, not realizing it only keeps the search engines safe. Well, maybe they deserve to have less rank for themselves through their internal links and less for the things they link to.

Comment by Rogers Cadenhead #
2005-02-04 06:14:26

”This isn’t something that most people will do: it’s mean, and cruel, and clearly breaks the web.”

I think within the commercial subset of the Web, companies will adopt nofollow as a technique to maximize the benefits of Page Rank. Why should the New York Times care whether its external links bestow Page Rank? Most large corporations hate external linking anyway.

Also, I think we’ll see some individuals and non-commercial publishers adopt the practice because Google has offered tacit approval for it. Wikipedia isn’t even commercially motivated, and they adopted nofollow across the board so eagerly they’re just now getting around to wondering whether it’s a good idea.

As for this being a Trojan Horse to devalue weblogs, that seems far-fetched to me. They could identify and devalue weblogs in other ways much more easily without messing with Page Rank — the company’s bread and butter.

Comment by Jacques Distler #
2005-02-04 07:15:00

As for this being a Trojan Horse to devalue weblogs, that seems far-fetched to me. They could identify and devalue weblogs in other ways much more easily without messing with Page Rank — the company’s bread and butter.

Not only is it far-fetched, it’s the wrong question. Google doesn’t care about who gets high PR and who doesn’t. What they care about is, ”will this change improve or worsen the quality of our search results.”

Maybe they figure that getting rid of comment spam does more to improve their search results than SEO hijinx with nofollow does to worsen them.

Or maybe they goofed. Time will tell.

Comment by Phil Ringnalda #
2005-02-04 22:33:00

Of course, it may be quite some time before the idea of using nofollow to tune your PageRank flow filters, well, down.

Comment by Jeremy Bowers #
2005-02-07 09:32:07

Yes, it was me.

It’s also worth pointing out that this need not be a binary thing; along with a ”nofollow” link being counted as 0 links (the current model everyone is operating under) and 1 link (the model I suggested is at least possible), it could also be a fractional link, or even a function of the number of links on a page; as Jacques Distler points out, Google only cares about improving results. I would expect it would take a number of iterations for them to get it right (if they are trying and aren’t just doing it the way everyone thinks they are), since it’s too complicated to guess in advance… and they may never tell us exactly what they are doing…

Trackback by Musings #
2005-02-04 22:43:25

Who Do You Trust?

PGP-signed Comments and rel=”nofollow”

Comment by Douglas Karr #
2007-10-13 11:32:55

The Dofollow plugin was actually conflicting with Brian’s threaded comments on my blog… some comments wouldn’t even display. I found how to disable nofollow in the core code in WordPress


Sorry, the comment form is closed at this time.