Multiple links on a page & PR transfer

Have things changed at all?

[More Traffic Please]

In the past I have always understood that an external page that linked to my page, and had ten total links on it would transfer 10% of it's PR to my site. If I had two of the ten links, I would receive 20% of the PR the page had to offer etc. Is there any evidence that this has changed? For example, the first link transfers 10% but each additional link placed on the same page that points to my site will transfer a lesser amount of PR?

[hetzeld]

I'm not even sure that a second link on the same page would transfer ANY PR...

Just a guess.

[Marcia]

It would probably be better if there are two links to have them pointing to two different pages on the site.

[johnnydequino]

What if you had 5 links, and had every link on every page on the site? Good/bad or does not matter?

jd

[ciml]

As hetzeld and Marcia point out, if two of the ten links are to the same URL then they count as one and each link gives out 1/9th (so no PR is lost).

Oddly, PR is lost if there are /robots.txt excluded links, or 404s (in the case of 404s, that only started last Friday).

johnnydequino, that would give you extra PR, assuming that there are other links too.

[Marcia]

>>PR is lost if there are /robots.txt excluded links, or 404s

I know if there are 10 outbound links on a page and 5 are to robots.txt excluded pages it's still divided by 10 (not 5) so half is going to waste, but Calum, what are you referring to by PR being lost?

[More Traffic Please]

Thanks ciml,

Thats interesting. I guess if 9 of the ten links pointed to one URL and only one link pointed to another URL, then the PR would be split evenly between the two URL's.

[ogletree]

It all depends on the PR. I have one site that points to 5 of my sites and it is a PR5. It also points to lots of other sites. This weekend all my sites went from PR2 to PR4 on the Google toolbar. The new PR5 link was the only change in my backlinks. I think that a PR5 site can make a bunch of sites a PR4. At least that is what I am saying. It does not take much for a site to get to a PR4. That changes the higher you get. Maybe the splitting of PR does not happen until PR6 or above.

[johnnydequino]

CIML - just to clarify:

I want to link to my top 12 highest ranking, highest relevant links on every page of my site.

So for 60 pages, those 12 links will show up on all 60.
(In return I will be on each one of the PR5 and above sites on their homepage)

What I am hoping is this will generate extra PR for both me and my link partners. True/false?

jd

[eztrip]

what a second.

If have say 3 PR5 pages that link to my home page and each of those PR5 pages points to 50 other sites with PR values from 2 to 10 on them, does that make my site a PR4 automatically just because it's getting pointed at by a few PR5 pages or does the PR5 when being passed onto to the next page get divided by 50?

Confused...

[ciml]

Marcia:
> I know if there are 10 outbound links on a page and 5 are to robots.txt excluded pages it's still divided by 10 (not 5) so half is going to waste,

That's it.

> but Calum, what are you referring to by PR being lost?

The half going to waste doesn't recycle within your link network. I imagine it ends up in the rank source (due to normalisation), but looking at a PR feedback loop situation quite a lot of raw PR can be lost from your pages by a REP excluded URL or a 404.

The interetsing thing is that the behaviour seemed to change last Friday. Letting 404s suck PR from a site does give an incentive to keep an eye on link rot, but I suppose that's not really a Google objective.

More_Traffic_Please, yes the raw PR would be split evenly and the Toolbar PR would be only slightly less than the linking page.

Ogletree
> It does not take much for a site to get to a PR4. That changes the higher you get. Maybe the splitting of PR does not happen until PR6 or above.

I recommend "The Anatomy of a Large-Scale Hypertextual Web Search Engine" by Sergey Brin and Lawrence Page, then "The PageRank Citation Ranking: Bringing Order to the Web" (same authors, and then some research into logarithmic scales. Only then, I recommend considering the affect of aspects like REP/404s on PR and the analyses of PageRank feedbank loops using log scales (which also causes confusion). I haven't come across any rigorous testing that demonstrated a change in the log scale with PR, or a change in the general flow of PageRank from the original Google papers.

johnnydequino, if I understand the structure correctly then it's PR nuetral. Some pages gain a little, and others loose a little but it averages out. You're not creating PR there, just moving it around. (Marcia once drew a nice parallel with thermodynamics.)

eztrip, the raw PR is divided by 50, but n^5 / 50!= 0.5 (between 3.5 and 4.0 by my calculations). Three links from Toolbar PR5 page (5 >= PR < 6) should give Toolbar PR4 or Toolbar PR5.

It is very important to remember that we're talking purely about pages; a link from a PR2 page on a PR10 site is just a link from a PR2 page (else we'd all be putting up geocities sites!).

<added>
Unless stated otherwise, I talk about a destination URL's PageRank without taking into account any PR feedback loop within the site.

[edited by: ciml at 7:15 pm (utc) on Aug. 14, 2003]

[Gateway]

speaking of all this PR calculation, has anyone made a tool where you could enter the # of links in/out and their PR's to do a rough calculation of what your site could get.. and then further on that, would it would take to get the next PR rank and so on?

the formula seems there, im sure someone could write something, im a non coder tho..

[More Traffic Please]

"speaking of all this PR calculation, has anyone made a tool where you could enter the # of links in/out and their PR's to do a rough calculation of what your site could get.. and then further on that, would it would take to get the next PR rank and so on?"

Because of the logarithmic nature of PR and the fact that we only see Toolbar PR, I think it would be pretty tough to come up with anything very accurate. In fact, I think the good old SWAG method may be more accurate. (Scientific Wild A$$ Guess)

[ciml]

If your analysis is right then you could be entirely accurate, just not precise. :-)

[More Traffic Please]

"If your analysis is right then you could be entirely accurate, just not precise. :-)"

I guess the part I don't understand is this. If you have two PR5 pages and one page has the ability to pass along anywhere from 4-8 times the amount of PR than the other page, depending on Googles log scale, how can you be very accurate?

[ciml]

You could be accurate in that situation by stating your result with an error margin of 800%.

The Toolbar PR scale is very inprecise, but from my analyses rather more accurate. In other words, the Toolbar doesn't tell you if a URL has PR 4.05 or 4.95, but it won't tell you it's 4.95 when it's really 5.05