36 replies to this topic

[bwelford]

I agree, Haystack. Any backlink from another page is worth having and they will all add up. Presumably the fact that link:www.mydomain.com will only show me PR 4+ backlinks is just to cut down the bandwidth used on this exercise.

The PR scale is said to be logarithmic, like the way we measure earthquakes. I have seen a view that each PR point means that the PR is 6 times more "powerful" than the one below it. So perhaps the logic is that 6 PR3 backlinks are worth as much as one PR4 backlink.

Barry Welford

Edited by bwelford, 27 August 2003 - 04:58 AM.

[Haystack]

Hi Barry,
and as Jill has pointed out in either this or another thread, today's PR3 could be tomorrow's PR8, so link away.

[Peter]

Asz,

If you like to understand PR a bit better without all the complicated math behind it,.. you can read an article about it here: PageRank Explained

Regards,

Peter

[Jill]

Umm...Peter, don't you think you should change the name of your PageRank article to avoid confusion with other articles of the same name?

Jill

[Jill]

Peter, where'd you learn that the dampening factor has something to do with how likely a person is to be bored with a page? I've never heard that before and can't help but wonder if that's accurate. I thought it meant something completely different. Are you sure about what you're saying?

[This was in his article.]

Jill

[Debra]

azs:

Your last question was a good one but a little off the topic, so I moved it here:

http://www.highranki...p?showtopic=567

[Mel]

I think the idea that the damping factor is something about being bored comes from the original Google paper at Stanford, which uses as an intuituve justification for the damping factor the concept that a user is given a random page and keeps clicking in the links in random order until he gets bored clicking on links and decides to load another page.

In the mathmatical formula I believe it is used to insure that the PR reiterative calculations needed to compute the PR of all web pages reaches a tolerable error in a reasonable amount of time.