Forum Moderators: open
So now a PR 5 page with more backward links is able to beat a PR 6 page with few links. Both pages are obtimised equally for the concerned kw's .
Also another observation is that the internal links from the same domain counts less...
Any one seeing the same?
[edited by: gopi at 7:07 pm (utc) on Dec. 3, 2002]
What matters is no longer the PR of the page, but rather the PR pointing at the page from pages with matching link text. For example, let's say you have 50 links pointing at your page, 10 of which include the searched upon keyword in the link text and 40 of which do not include the keyword in the link text. My theory is that what Google scores is the cumulative incoming PR from the subset of 10 links that contained the keyword phrase. The result is a dynamic PR score for each page that varies based on the word searched for. We could call this the PR-Plus factor.
I've done some analysis to believe that my theory may be one major element of the current Google algorithm.
I am curious to hear what others think of my theory.
The sites i observed have keywords in the domain name so all incoming links have kw's in it , its the case for both high and low PR sites....
I am seeing consistently that sites with more links beat sites with less backward links ...All other things like anchor text, onpage creteria are equal here...
I remember ciml proposing the same in an another thread...
Currently PR is calculated by google every month and is static for each webpage
Its really difficult to precisely analyze, with the crude approximations that the toolbar gives us.
You are essentially arguing that pure link count has become a major factor. I am I reading you correctly? What I don't like about such a theory is that it would be way to easy to spam. 10 PR4 links are easier to get than 3 PR6 links.
In my observation, the 3 PR6 links would boost a site much higher than would the 10 PR4 links. I have seen sites with fewer links beat sites with more links.
What's difficult about this is that you seem to be observing different things than I am. Maybe we could swap search terms by stickymail, and do some cross comparison of each other's keywords.
Like you, I have also observed that internal site links seem to count much less than before 2 mos. ago. Not sure if this is element of the algo, or if it is just a shift of the PR scale.
Example 1.
Domain1.com has 10,000 pages with link on each page to targetdomain.com.
This will give targetdomain.com a good PR and lots of links, but not so good placement withing results.
Example 2.
targetdomain.com has 100 sites with individual links to it.
This will typically have a lower PR but will place much better in the results.
I personally think this was done so people could not go and buy their way to the top of the results by simply purchasing a sitewide link from a domain that had a high PR.
I have seen a PR3 site holding out PR4 sites with far better keyword density and many internal links. The PR3 site is not at all well optimised.
The PR3 site has links from mainly PR2 & PR1 sites with no above PR3. Most links are very related.
This situation has been the case for the last 3 months, but Google clearly has not quite worked out the pattern as the results changed again over the weekend.
What this is showing is that deep linking into a large site is more important now than just lots of links to its home page. This means Google is giving the little sites more of chance and making life very hard for larger sites.
As a result I have changed the affiliate links to the main page and changed the prefabricated links to include the keywords...Guess we will see in a month.
Is the high PR because it is a new site? We are linking to 10 sites that are with in our subject, but as I said - no one is linking back to us....
Does this mean we will get a even better PR on the next update - if we get people to link to us?
Potter
[webmasterworld.com...]
Gopi, I would say there should be something to that theory.
Internal links counting less, I think was something that happened in the september update.
