PR0 Sites and Historical Data

[Marcia]

I've seen some sites (not new ones) go PR0 that have several fairly decent PR inbounds, which should give them PR3+ or at least PR2 on the homepage, and also have other IBLs that have been around for a while. Still PR0 across the site.

Yet, I've also seen a couple of cases of newer sites with fewer links, including not having any from pages that have decent PR (like PR4) in a couple of cases, that are PR2 or PR3 just about across the site.

The difference between the two groups is that the first sites, the older ones that are PR0, haven't had any new IBLs for a long, LONG time; whereas the others, the sites which are newer than those, do have some PR, while having very, very few IBLs.

I'm thinking that maybe it isn't a *bug* at all causing the PR0, but that historical data might be used for time lapse of link acquisition, and that it might be hurting to not have gotten any link for a long time. It sure looks that way.

Anyone seeing sites go PR0, and if so, how long has it been since new links have been received from pages that are indexed by Google?

[edited by: Marcia at 3:59 am (utc) on Nov. 26, 2007]

[Robert Charlton]

It may be impossible to know with much certainty whether a given page has received new inbounds in a while... or what the Toolbar PR snapshots of the pages in question might mean at a given time.

That said, if enough lightbulbs to go on among site owners who've seen PR zeros and have a sense about the age and activity of their pages, then we may learn something.

[edited by: Robert_Charlton at 5:56 pm (utc) on Nov. 26, 2007]

[tedster]

Just to clarify - as understand it, Marcia is not trying to resolve a "problem" with "a website" here. These are long dormant domains that she began to look at in the light of Google's patent on using historical information, and she feels she seeing a pattern -- so she is floating an idea by us for discussion.

For me the question becomes this: Is anyone reading here who has a long established site that went from a formerly decent PR to PR 0? Such a site owner could check to see if the site recently got any new backlinks.

I do have some previously developed domains that have now been dormant for years, so I thought I could check out Marcia's idea on my own. However, in every case of mine, the sites have attracted recent backlinks - and they all retained their PR. So no corroboration from me, and no negation either.

[edited by: tedster at 8:50 am (utc) on Nov. 26, 2007]

[stgeorge]

We have a couple of sites that have gone PR0 and there is no explanation for it.

The question is interesting....

[edited by: Robert_Charlton at 5:54 pm (utc) on Nov. 26, 2007]

[tedster]

Stgeorge, have your newly PR zeroed domains recently acquired any new backlinks?

[MarkWolk]

I'm thinking that maybe it isn't a *bug* at all causing the PR0, but that historical data might be used for time lapse of link acquisition, and that it might be hurting to not have gotten any link for a long time.

I have one site hit by the PR0 "bug". It has acquired several new links recently (1-3 months ago), before being hit. Several links were from PR4+ pages.

[glengara]

Is it just TBPR 0 or have the sites lost pages from the index?

[Pico_Train]

I've wondered about this too since I've slacked a bit on some sites and as would be expected over time, lost some rankings, probably due to others increasing links more than my site.

At the same time, I've wondered if perhaps over a set time, links although still "live", "die", so to speak, that is they no longer give the boost they used to. I also suspect links from "reputable", "authoritative", "respected" and "high-end" domains do not die - Yahoo, ODP, New York Times, etc...

Different from them being removed on an expired domain, change of ownership or page revamp.

[MarkWolk]

Is it just TBPR 0 or have the sites lost pages from the index?

Only the toolbar PR is affected (sitewide), but the rankings are still OK on that particular site.

[tedster]

The parts of this puzzle may be difficult to pull apart. But let's note these factors.

1. With the April PR Update, there definitely was a PR0 bug for the home page that did not effect inner pages or rankings. Google's Adam Lasnik confirmed this [webmasterworld.com] as just a reporting bug,

2. In November there were three distinct PR updates in quick succession. The first two were toolbar "demotions" for some sites that sell links that pass PR - but so far rankings appeared to be unaffected. In every case I've seen, there is still some PR being reported. I've also not seen a verifiable case of the link BUYER showing lowered PR, only the link seller.

3. The third November PR Update was across the board - a long-await export of more recent PR data to the toolbar. It's possible, but not confirmed, that in this mix we're also seeing a similar PR0 home page bug to the April PR update.

Are we sufficiently dizzy yet?

4. Marcia noticed "PR0 across the site", and not just on the Home Page. If this idea of historical data influencing PR holds up, then I would also expect to see PR0 on ALL the internal urls, and not just on the domain root as we see in the "bug" variety of PR0.

I would also expect to see rankings and spidering affected - all the urls going supplemental, for example. Otherwise, what's the point?

[nippi]

My experience with backlinks, is once you reach a certain level of backlinks, there is nothing you can do to stop further ones being added.

Either by cold calling link exchangers, people who simply copy other site's outbound links, software that scans dmoz and search results etc,

I monitor the links of dozens of sites which I have not manually added a link to in some years, yet their inbounds continue to grow.

Marcia have you checked inbound links for the sites using a reliable tool and can say for sure no links added in a long, LONG time?

[glengara]

If the PR0 was based on "historical data" I'd have expected a gradual drop in rankings for inner pages followed by some of them dropping out of the index....

[tedster]

Google filed a new link-related patent [appft1.uspto.gov] in November 2006, and freshness of backlinks is a frequently mentioned scoring factor in there. Here's the abstract from the patent filing:

A system may determine time-varying behavior of links pointing to a document, generate a score for the document based, at least in part, on the time-varying behavior of the links pointing to the document, and rank the document with regard to at least one other document based, at least in part, on the score.

I have no doubt that freshness of backlinks is a factor today - and has been so in some fashion ever since the first historical data patent. But how that scoring factor is implemented is the question for me. I can't quite wrap my head around the idea of sending all the domain's urls to PR0 for stale backlinks. It just seems like it would be so clunky and inelegant, compared to the kind of algorithmic work Google does in other areas these days.

[glengara]

Might it not equally be read as an attempt to monitor "link churn"?

[webmasterworld.com...]

[nippi]

i am curious about this patent. Now its been thought of by google, and patented.... does this prevent other search engines using it?

[tedster]

That's what it is. The question that remains is how those link-churn scoring changes are accomplished. I somehow doubt the PR angle, and rather think Google's got another algorithm factor going, somehwere in the query-specific link juice calculations instead of the query-neutral PR calculations.

[stgeorge]

tedster to answer your question, yes, there have been new links acquired, but at no greater rate than in the past. They are all on-topic links from "clean" sites. Some are reciprocal and some are one way.

The site has lost all PR, but rankings have remained. The number of indexed pages has not changed and neither have the link: results.

I am completely baffled by this one.

[Marcia]

link churn*

In this case, if it's genuinely something that's happening as cause and effect, I don't think it's so much "link churn" as looking at the number, quality and time of appearance of links as they relate to the inception date of the domain in question.

...another algorithm factor going, somehwere in the query-specific link juice calculations instead of the query-neutral PR calculations

It wasn't mentioned in relation to PR as such, but there was specific mention made in the Patent application about the probability of older sites being expected to have a good number of links, compared with newer sites that are expected to have very few, and additionally, that new sites with brand new links may appear to have a better growth rate, resulting in the possibility of having an advantage for either one or the other. It goes on to say that when inception dates are considered, there may be appropriate modifications made to the scoring (either positively or negatively).

While that's referrng to ranking score rather than PR calculations, there's obviously some mechanism allegedly in place that can accomplish normalization for link weighting and scoring based on historical and inbound link data.

From their Webmaster Help section on PageRank Zero [google.com], disregarding being too new or guidelines violations (different issues entirely), take note of the mention of numbers of links:

There are no or few trusted links to the site.

Those are only a few points they're revealing, but the key I'm looking at is the use of the word "few" and how that relates to the other myriad factors that relate to what the whole linking profile and history of a site are.

Even if, hypothetically speaking, a site were to have a couple of trusted links dating a couple of years back, and otherwise not have a strong link profile, how much in the way of sending signals of quality is there, if the site hasn't acquired any new inbound links with a minimum level of quality, relevance or trust during the period of time since?

*BTW, the patent application linked to in that thread isn't at that location any more, the URL is now an app related to local search. It seems to me that all (or most) patent applications that have anything interesting or revealing about linking seem to have a rather ephemeral presence once word of them gets around.

[edited by: Marcia at 12:15 am (utc) on Nov. 27, 2007]

[glengara]

It's the absence of all the "normal" PR0 symptoms I find baffling....

[randle]

I don't think it's so much "link churn" as looking at the number, quality and time of appearance of links as they relate to the inception date of the domain in question.

We do believe in this, but not in so far as it ultimately results in a PR of Zero, (which is a rather dramatic event in itself). It would appear to us, that a lack of new inbound links, these days results in a continual decrease in PR, (or PR as displayed in the tool bar).

Perhaps there is cause to think about how pure age of the site is being factored in, (which used to be quite significant) and the time line of links acquired. Old ones were more valued than newer ones, but now newer ones seem to put the luster on the apple.

[CainIV]

I have a domain that was recently pr0'ed. It was previously a pagerank 4.

The website gets news links about every 2-3 days, often from higher pagerank on-topic authority websites and has been for about 4 years.

[Marcia]

A site that went PR0 also has lost all rankings. Interesting, no buying or selling links, and white hat.

[Marcia]

Another factor is the freshness of the pages the inbound links are on. If there aren't new or even recent backlinks, and the links that have existed for a long while are on pages that are "stale" and/or have a poor freshness score, then the value of the site's link profile goes down, and it might just explain going from a PR4 to PR0.

That wouldn't be link churn, that would be a quantum leap downward with regard to link-based scoring for the site, once all the historical link (and freshness) factors are calculated.

[webastronaut]

I had a 8yo site go from PR4 to PR0 in two stages in Nov. I used to do a lot of link exchanges on that site back in the day. The site still gets linked to from other sites and still has great search results from Google. I've been really confused about this and a little nervous because it is my main site for income.

[carminejg3]

it can easily be that a competing site reported you for selling links... even though you only exchanged.

I believe this happened to a few of our sites.

[Marcia]

Just to update, I'm still seeing this on stagnant older sites with no recent backlinks, while much newer sites still show the normal PR they would be expected to be, given the links they have. In fact, by all rights one site in particular should be around a PR4 based on some IBLs - and that's what it did have until it dropped down to a grey bar (and lost rankings).

And no, there's no question about links being bought or sold because it's never been done.

[Miamacs]

How's the cache of the linking pages?

I too see this sometimes, but the links aren't just old... but also from pages that ( while may 'show' medium-high PageRank ) usually have cache dates well over two months old.

...

On sites where there's a lot of PageRank circulating, a single link to a subpage can keep it with a semi-green bar, even if all of its other parameters are below of even those in the supplemental index.

Links from such places don't worth much.
I doubt there's a separate script to check fresh links/old links but...
Perhaps there's a time-out to the validity of a link.
I mean, a link will only worth anything if it's discovered every month or so.

But if not, it's assumed it's not really there.

Let's say the pages that have the link are there, and the link hasn't been removed, but... the pages get crawled/cached so seldom... that the link(age) checking script considers the association between the source and the target pretty much broken. ( doesn't know the reason but didn't get fresh data on the existing link, so it considers the link to be... gone )

...