6 links from PR x gives you PR x+1?

[chrissi]

Hi,

just say the google toolbar is based on a log(6):

So I need about 6 links from PR 4 sites to get a PR 5 for my site.

Or 6 links from PR 5 sites to get a PR 6?

Or 36 links from PR 4 sites to get PR 6?

Is this calculation correct?

Thanks,

Chris

[James_Dale]

Something like that. Do a search for "PageRank Uncovered" in Google...Naturally there is more to consider than volume of links.

[edited by: James_Dale at 12:53 pm (utc) on Oct. 19, 2003]

[Marcia]

>>So I need about 6 links from PR 4 sites to get a PR 5 for my site.

No, that would more than likely make it a PR4 or possibly a PR3 if they were "low" PR4's with a lot of other links on the pages. It would take a lot more than just that to have a PR5 - and not all PR4 links are equal, nor are all PR5 links equal.

[DerekH]

Just to remind you that a link from a page of Pr (n) has a "strength" that goes down the more links there are from that page...

I think the analysis you've done is far too simplistic - the more links from a page, the less each is worth...

You need to read the paper that James_Dale refers to!

Regards
DerekH

[chrissi]

ok, it seems, that I thought it were so easy, I downloaded the document "PageRank Uncovered" James_Dale mentioned ... and will read it now :)

Thanks,

Chris

[ciml]

The following assumes a Toolbar log base of six (not a figure I'd choose), that you don't recycle any of the PR back to that page (unusual), that none of the pages are penalised and that all have been indexed. Where I write PR, I mean Toolbar PR, and to make things easy to follow I'm ignoring the decay due to the rank source. (You could add 0.091 to the Toolbar PR of the incoming links to compensate.)

To get PR5.0, you need six links from PR5.0 pages, each with six links on it.
Or six links from PR4.0 pages, each with one link on it.
Or six links from PR6.0 pages, each with one thirty six links on it.
Or one link from a PR5.0 page, with one link on it.
Or one link from a PR6.0 page, with six links on it.
Or thirty six links from PR5.0 pages, each with thirty six links on it.
Or thirty six links from PR4.0 pages, each with six links on it.
Or thirty six links from PR3.0 pages, each with one links on it.

You can play with the figures:
LOG₆(6^^PL * NL / LPP)

6 = Toolbar log base
PL = Toolbar PR of the links you're getting to the page.
NL = Number of links you're getting to the page.
LPP = Number of links per page that links to you (assuming no two links to the same URL on a page, and no links to /robots.txt excluded or HTTP status 404 URLs)

[pegaweb]

Isn't there also a "damping factor" that needs to be considered?

I think it's 0.85

Meaning that a page passes on 85% of its own PR through links.

[ciml]

That's right pegaweb, but when comparing the relative Toolbar PageRanks we're dealing with a log scale. Note that in the figures I use I'm saying that there would be an 0.091 absolute difference on the Toolbar to compensate for a 15% difference in raw PR.

[James_Dale]

Hey Guys, from now on, just 'James' please. I feel like a complete twat being referred to with an underscore ;)

[DerekH]

James wrote
>>Hey Guys, from now on, just 'James' please. I feel like a complete twat being referred to with an underscore ;)

You're right, James - and indeed I recall that GoogleGuy himself recommended the hyphen rather than the underscore <grin>

DerekH

[James_Dale]

Marcia, thanks for the sticky reminding me of the ...ahem, alternative meaning for 'twat'..

Geez, now I really do feel like a ...