| 9:34 am on Nov 25, 2003 (gmt 0)|
I thought it was logrithmic.
4 -> 5 =10
5 -> 6 =100
4 -> 6 =1000
| 9:44 am on Nov 25, 2003 (gmt 0)|
Hi davidpbrown - many thanks for your reply ... that is the same thing, I believe - do you remember where you came across this explanation?
| 10:01 am on Nov 25, 2003 (gmt 0)|
Not from one place, just gathered from little bits speculating on what's needed for pagerank increase.
Personally I don't pay any attention to PR and just deliver a good rounded site.. give users what they want and the like of PR and search results follow.
Here's one talk of PR.. High PageRank is easier for the Big Fish [webmasterworld.com]
| 10:27 am on Nov 25, 2003 (gmt 0)|
Hi, thanks for that recommended thread - one of the posts there by danny suggested that the increase in PageRank was by a factor of 6, rather than 10.
And discod posted this guide:
Pages who have: fewer than 100 backlinks will be PR 5 or below
100-500 range = PR 6, occassionally 5 or 7
500-3000 range = PR 7, occasionally 6 or 8
3000-12000 = PR 8, occasionally 7 or 9
12000-50000 = PR 9, sometimes 7 or 9
above 50000 = PR likely 9 or 10, occasionally 8
again ... many thanks,
| 10:35 am on Nov 25, 2003 (gmt 0)|
|And discod posted this guide: |
This isn't correct - the PR algorithm is a combination of inbound links and the PR of the referring page.
One link is sufficient in some cases (for example a high PR7 referring page is likely to give the recipient page a PR6).
It has absolutely nothing to do with raw number of inbounds.
| 10:54 am on Nov 25, 2003 (gmt 0)|
Hi trillianjedi - many thanks for that clarification. It's good to know ....
| 1:02 pm on Nov 25, 2003 (gmt 0)|
To get a good idea about PageRank, I'd still cite the original papers as giving the best understanding about how it works (without modern changes, such as penalties and how to handle 404s etc.)
"The PageRank Citation Ranking: Bringing Order the the Web" and "The Anatomy of a Hypertextual Search Engine", both by Sergey Brin and Larry Page.
| 1:43 pm on Nov 25, 2003 (gmt 0)|
Links to above papers:
The Anatomy of a Large-Scale Hypertextual Web Search Engine [www7.scu.edu.au] by Sergey Brin and Lawrence Page
The PageRank Citation Ranking: Bringing Order the the Web [citeseer.nj.nec.com] by Sergey Brin and Lawrence Page
The first is in HTML, the second, alas, in PDF; it looks as if it had been reproduced from a very poor typewritten carbon copy. Does anyone know of a better online version of it?
| 1:47 pm on Nov 25, 2003 (gmt 0)|
* The Actual PageRank is what is used internally by Google to help calculate your rankings.
* The ToolBar PageRank is just a rough estimate of that value, based on a 1-10 scale.
* Actual PageRank increases exponentially for each Toolbar "step".
* The exact exponential factor is unknown, subject to some disagreement, and may change, but some have estimated it to be roughly around 6.
* The Actual PageRank is the sum of all of the PageRank credit from each of you page's inbound links (including internal links from other pages of your site)
* The amount of Actual PageRank credit is based on the total PageRank the linking page can provide, divided-up equally among all of the outbound links on a page.
* So, a link from a high-PR page will give your page more PageRank.
* And for two pages with similar PageRank, a link from a page with fewer outbound links will likely give you more PageRank than one with many outbound links.
* Also, for example, a link from a PR4 page with only 5 outbound links would typically give you more PageRank than a link from a PR6 page that had 50 outbound links (because it has to be divided-up among 10 x as many outbound links).
* You can also adjust your internal linking and site structure around to "concentrate" more of your total PageRank on certain pages on your site.
* Contrary to popular misconception, your outbound link structure can also have an impact on your total PageRank, due to PageRank feedback through your site's structure.
The details get a bit more complicated, but you get the general idea.
| 1:54 pm on Nov 25, 2003 (gmt 0)|
If you look at the underlying maths in the original papers, it is certainly a logarithmic scale. But what is not known are some of the parameters, such as the size of the damping factor, and the actual base of the scale.
By base, I mean if an increase in PR of 1 corresponds to an increase in 'importance' of b, then the base of the log scale is b. There's no particular reason for the base to be 10, it may be larger, or smaller than this. Who knows? But my bet is on it being larger.
For example, my site is apparently PR 6, the BBC is PR 9. Now by what factor do you think the BBC is more important than my small e-commerce site? If the base is 10, it is 1000 x more important - doesn't sound right to me. Instinctively, the BBC megasite is surely much more than 1000 x more important than my site! So maybe the base is 20 or even higher!
(N.B. a PR of 9 could of course represent a true PR of anywhere between just over 9, and almost 10 due to the integer rounding)
| 2:22 pm on Nov 25, 2003 (gmt 0)|
|Instinctively, the BBC megasite is surely much more than 1000 x more important than my site! |
I think I understand what you mean by "instinctively" -- i.e. in human terms, it might seem that the BBC, as an organisation and media giant that is using the web effectively, is *way* more important than your own mini-business and accompanying site.
However, an algorithim is just an algorithim, and doesn't have this human understanding of scale. So a factor of 1000 seems a perfectly reasonable number in this context.
And human understanding differs from one person to the next. Saying the BBC site is 1000 times more important than your e-commerce site does, to me at least, seem a pretty significant difference in scale.
|Small Website Guy|
| 3:32 pm on Nov 25, 2003 (gmt 0)|
If PageRank runs from 0 to 10.9999 (with the toolbar always rounding down), and there are 3 and a half billion pages on the web, this comes down to a logarithmic scale with a base a little bigger than 7.
| 4:52 pm on Nov 25, 2003 (gmt 0)|
If you have a "low" 6 and BBC has a "high" 9, the difference could be as much as -
where x is the exponential factor.
Using link:, Google reports about 339,000 decent inbound links to BBC's home page.
If you have a PR 6, that is pretty good, you quite likely have more than a handful of good inbound links to your site as well. I'm guessing at least several hundred, unless you got a few really CHOICE links.
How much more PR do they have?
If x = 6, that's 1,296 times, if x = 10, that's 10,000 times.
In the public's eye, their site might seem MUCH more important.
But from a purely PR standpoint, that their page would have a PR of 1,200 - 10,000 times higher than your pretty decent PR 6, seems a reasonable value.
| 1:08 pm on Nov 26, 2003 (gmt 0)|
I did actually point out in my previous post that a nominal PR value is actually rounded up and represents a range of possible values :)
| 2:20 pm on Nov 26, 2003 (gmt 0)|
You are correct. That was why I used it as the basis for the calculations.
BTW, am curious, those Pr6's don't seem to be as prevalent as they used to be, how many inbounds you have on that PR6?
| 2:50 pm on Nov 26, 2003 (gmt 0)|
Hi Aspdesigner - I don't have that many inbound links, but I have some good quality ones, a few PR5 and 6s, but nothing spectacular. My quals are in astrophysics, and one day I simply got Word to convert my Masters thesis to html, I put it on the web and thought I might as well point it to my commerce site.
It is such an unusual piece of work that several universities and institutions linked to it.
So that's the key to high PR - a degree in astrophysics ;)
[it's nice to get some benefit from it, the qualification itself has never earned me a cent because I'm simply not smart enough to be a proper professional astrophysicist :)