Forum Moderators: Robert Charlton & goodroi
Matt Cutts mentioned the change at SMX Advanced last night. Excellent summary in Search Engine Roundtable [seroundtable.com] post.
I'm sure there's going to be a lot of howling over this change.
The random surfer is a nice image, an analogy.
Yes, I know, I was trying to use the analogy to explain my point without having to actually do the math. It's been a while since I took Linear Algebra ;-).
PR does not have to normalize
That I didn't know, and it makes this a non-issue. That makes sense though. If it's not normalized, then it's not conserved. It can be created and destroyed. So all you have to do, like you said, is create some more to compensate for what is destroyed.
We are actually implementing nofollow links on our pagination of category pages now. I'll see if we see any of the sculpting benefits...
I don't see how.
The inline nofollow is the equivalent of a nofollow meta tag.
The nofollow meta tag makes pagerank die, period.
If you inline nofollowed every link on a page, the effect would be the EXACT same as using the nofollow meta tag.
The inline nofollow tag makes a dead end, just like the nofollow meta tag, and just like linking through a robots.txt denied page.
He said what nofollow is like, both things kill pagerank. Those statements are clear. In contrast, he never, ever, ever, ever said anything like this notion that nofollowing half of a page's links increases the power of the other links. Never... even though that is an extremely simple concept to say, and is the clarity you want. But he didn't say anything like that, ever.
(But again, what they actually did could be a million things. It's just that he described it clearly as the same as ways that pagerank does, and never explained it in the way the FUD spreaders have been insisting.)
Exactly. There wwill hardly ever be reason to use nofollow on a granular level for your own site rather than noindex,follow... except for some obscure reasons like not wanting to run the bot into a page it will have problems with, or nofollowing jpgs you don't want in image search.
Nofollow offers some nice functionality, but that is almost all for off-site links. Generally there are better tactics/tags for your internal links.
PR does not have to normalize
Er... Shaddows, are you sure about that? Go to Matt Cutts' blog and look at slide 15 of the slide deck he has in his May 30 2009 post. It gives the classic mathematical definition of pagerank. He explicitly states that the L1 norm of pagerank is 1. That looks like normalization to me.
As I said before, I think G has enough math skills to compensate for this -- but I don't think you can dismiss the issue of normalization offhand like that.
I'm not saying they couldn't do it and make it work, but I am saying that it would have to be done very carefully and I just question whether it would be worth it from their perspective. Maybe it is. It just seems odd and I'm not satisfied that we're getting the whole story here.
Given Wikipedia's ubiquity and authority, surely this would have been one of the things Google checked before rolling out any new scheme?
Let's Assume that I have a site with PR 4, which has 50 links to internal pages on it (noted as Page 1, Page 2, ...., Page 50). Let's also assume that Page 1 is an important page that I want to elevate ranking for. In this current scenario, link juice from my PR 4 index page will be spread evenly across pages 1 through 50, so Page 1 will only receive 1/50th of the link juice (in theory).
Now, let's suppose that I add a 51st page to my site, and restructure my linking so that my PR 4 index page only points to Page 1 (from above) and my new Page 51 ONLY. Page 51, in turn, is structured to point to all of my other original Pages 2 through Page 50. In this scenario, will Page 1 benefit from more link juice, since it is only one of 2 pages being pointed to directly by my index page ? OR, will my link juice of my index page still be dispersed to all 51 pages, since the Index page point to page 51, which in turn, points to the remainder of my pages ?
Does this question make sense ?
The dilemma, of course, is that the approach you describe would also completely de-emphasize most of your pages for the user. What you want is an arrangement that works both for users and for search engines.
This suggests intelligent hierarchical categorization, with occasional nudges for emphasis where needed.
People linked to stuff before nofollow existed. Nothing about nofollow impacts on that.
Sure, it's now getting half of the pagerank your index page has to give instead of 2% of it.
In contrast, the pagerank of pages 2 though 50 will be gutted (and even more so if Page 1 doesn't link to pages 2 through 50, or page 51).
(1) i am thinking many pages on my sites, are suddenly going to rank better for the term "more". When showing article snippets, i normall have an ordinary link on the articles title, and a rel-nofollow link on the more... at the end of the snippet. Now? if I don;t want PR to fall into a black hole i need to remove the rel-nofollow, or remove the link entirely or look like I am spamming the anchor "more"
(2) I usually have a rel-nofollow on the text link to "Home" with proper link on the sites main keyword which link to the home page, elsewhere in the naviagation. makes sense I would prefer to rank for this term, than "home" or lose pr simply for telling google not to follow this link.
Google. Really. What is this going to achieve? Those who are pages sculpting are going to now find another way, but weve lost a valuable tool for letting google know what links and pages are important, and which are not.
its a mistake
so, i wont do that then
In the meantime, why make changes unless you are seeing a new ranking problem that seems related? Your site, using nofollow, is already a kind of test, isn't it? If something really did just change at Google, then any changes to your rankings should be there, and if so, that's when I'd consider making a change.
This is the kind of "big buzz" issue that causes some people to jump too fast and that can create problems rather than pre-empt them.
on the internal links side I believe I have the tools to figure it out myself, just haven't updated my graphs yet. probably will also need to make some changes and track the results.
Won't webmasters selfishly just eliminate as many links as possible from their sites, to preserve link juice for the internal benefactors?
Some may, but--as was suggested earlier in this thread--they could find themselves at a disadvantage if Google gives a ranking boost to pages that have relevant outbound links for a given query.
If you have 10 links and 5 are nofollow, there is this assumption that the other 5 get page rank. That might have been partially true at one time, but that’s less effective these days.
Initially if you had 10 links, and 5 were no followed, the other 5 would get the remaining page rank, it’s not that way these days.
Very vague, but official enough for me. I doubt they'll get into it much further.
It doesn't seem to matter much if you add or remove nofollow to/from old links, what is on the page when it is originally published seems to be weighted more heavily when it comes to links. I'm not sure a link added to an existing page long after it was published will result in passing pagerank 100% of the time anyway.
[edited by: JS_Harris at 8:09 am (utc) on June 8, 2009]
I'm not sure a link added to an existing page long after it was published will result in passing pagerank 100% of the time anyway.
I don't see how that could possibly be right. Links added after publication might be weighted differently, but there's no way they don't pass pagerank. What possible justification could there be for that? It would just be silly. There would be no point in adding new pages to an existing site, just as one example.
I can think of dozens of counterexamples from my own experience, and I have never seen anything to suggest this might be true.
Exactly. With my sites, I use ASP or PHP scripting (include files) for footers, headers, etc. for showing those utility links.
Not exactly. ASP and PHP include files generate HTML. Those links are now on every page. Google, for all intents and purposes, doesn't know or care that that is a PHP include (it can usually tell from headers, but that's irrelevant here). The point is, this changes nothing from a static page.
What Tedster is taking about is altogether different.
Your utility links are on a page that is in and of itself a complete page. It's not a server-side include at all.
You then build your page template such that in the appropriate spot on the template, you have an iframe. That utility page is shown as in the iframe. Effectively that's a sort of client-side include.
The difference is that the links on the utility page are *not* part of the main page.
How G handles this is up to them of course. They could decide that anything in a frame counts as part of the page and each link counts just as though it were a server-side include. But Ted is saying that for now he's not seeing that to be the case. Rather, Google sees it as one isolated page that has no inbound href links (just the iframe source attribute) and thus doesn't leak off juice.
So I suppose under a current Google implementation, you could create a Wordpress theme where the comments are in fact an iframe and nofollowed and it would have a similar effect to what nofollow used to have.
[edited by: tedster at 12:43 am (utc) on June 9, 2009]
[edit reason] moved from another location [/edit]
will Google transfer PR if I use a click tracking script
Google reps have been pretty vague/silent about this. If it is easy to assume the target url (i.e the link uses the target url as a parameter) it is possible that will establish a virtual link in the webgraph - and then send PR even through the script. That IS what they do with urls the discover via form input.
So it "might" depend on whether the tracking script really obscures the target url. Or not - as I said, this is indefinite territory. It might well depend on factors we're not considering, or vary in different situations.
It looks like using no-follow is now problematic
It's effect on the available PR for OTHER links on the page is what is problematic. It is still a certainty that rel="nofollow" will not vote any PR through that particualr link.
Whatever you are doing now (whatever Google has already indexed) is indeed being treated in this newly communicated, or "revised" way. Perhaps you can analyze your current situation and see whether you are satisfied with the current results.
Perhaps you can analyze your current situation and see whether you are satisfied with the current results.
I was satisfied with my current results 2 weeks ago when my site had all links in the sidebar using the nofollow attribute on every page other than the index. I relied on crosslinking of relevant articles to determine overall article worth.
After listening to Matt Cutts and reading the opinions of many industry pros I decided to stop wasting PR to the category links in the sidebar by removing all of the nofollow links. If it was true that PR is "wasted" and not "redirected" to other links on the page I should have seen an increase in rank/traffic. The opposite happened.
Within 48 hours the index page began dropping in the serps for all major keywords and traffic to the index page fell too, articles remained consistent. Roughly 48 hours later key articles began to do the same thing and the index continued to fall though not as quickly. As of last night the site is once again stable though not as highly ranked as it was prior to removing the nofollow attribute, traffic is down as well.
This morning I reverted the sidebar to be as it was two weeks ago and will find out if it returns to where it was. I have a hunch that pages can be "grandfathered", meaning a page that is already indexed when changes occur is spared the full effect of those changes, but that new changes to the same page receive treatment that includes all changes.
If I'm right the site will not return to it's former glory, the nofollow will no longer do what it did before because Google "refreshed" it's view of the page.
Perhaps THAT is what Matt Cutts meant. Perhaps months (years?) ago using nofollow sculpting helped, and pages that got the treatment back then still benefit from it, but any new pages trying to sculpt pagerank will not be credited for the sculpting anymore.
It makes sense that when Google re-crawls a page it applies current standards, it also makes sense that if it finds nothing has changed that it leaves it alone.
edit: to be clear, I removed all of the nofollow attributes from links, the links themselves were not changed. Also, if I'm right and the site doesn't bounce back i'm considering going with a "no navigation on article pages" layout to force the crosslinked articles+related articles+search method of finding more content. A link can't waste PR (which hinders other articles) if it's not there to begin with...
I think by making all these changes I may be causing semi-permanent damage, which sucks, but how else am I supposed to learn Matt?
[edited by: JS_Harris at 10:29 pm (utc) on June 9, 2009]
This whole situation smells of FUD.
JS_Harris, I predict that within whatever cycle your site works on, it will return to previous positions. Given the two-week delay to damage, I would say 2-4 weeks to recover. I would appreciate any reports you have.
I suspect that NoFollow just got discretionary, shades of grey have been introduced. I further suggest that Google always thought it might happen, hence the VERY loose language deployed to describe the effects.
What might this mean? Well, legitimate use (sculpting, blogs, algo-defined UGC) will work as previously thought. PR hoarding will no longer work. As I've argued elsewhere, allowing proliferation of NF would break the google algo, at all levels. Google needed to check this trend, and now they have.
You can either link out or not- vote or not. You can't have your cake and eat it. You simply cannot hoarde PR- you either send it or waste it.
As with all discretionary measures, Google will not speak. They will not lay out the effects of NoFollow, as it will depend on circumstances. And I expect treatment will vary over time- maybe by experimenting with split wastage/retention depending on context.
Thoughts?
