Forum Moderators: open
I went out to page 67 and at the bottom of the page is www.semconf.com (the page for pubconf) that is a PR6 with only 28 links. On pages in the 20s there were PR4 guestbooks with hundreds of outgoing links.
Just because something "appears" to be true from a quick glance does not make it true.
Many of the higher PR pages get crawled earlier since they are linked from higher in DMOZ aned Yahoo. If they are listed in the order they were entered into the DB, they might appear to behave in a manner that could be mistaken at a glance to be something like the PR passed through the link.
<Many of the higher PR pages get crawled earlier since they are linked from higher in DMOZ aned Yahoo. If they are listed in the order they were entered into the DB, they might appear to behave in a manner that could be mistaken at a glance to be something like the PR passed through the link.>
This seems like it would be make sense, and be consistent with all of the backlink checks I've done. Good observation.
link:www.webmasterworld.com was just an example which I have not examined in detail so far, because i don't want to post my own sites.
Of course, I can not prove my assumption, because I have to show that it is valid for all domains - while you just need one example to show a violation of the rule.
However, if you have a closer look at www.semconf.com you will see that this site has only one relevant incoming link and that this page contains also a link to www.webmasterworld.com. Perhaps Google is giving for one of these reason less weight to this incoming link? (I know that this is speculative).
At least for some of my own domains I can rule out that the results are ordered as they entered into the DB.
Also I can't believe that the fact that there are only high ranked backlinks for google.com is random.
My thoughts on the order would be that they are not in PR heirarchy. It would make it very easy for competitors to look up your baclinks and solicit links for themselves. You can do it anyway, but ranking them would let you know who which of the sites carried the most weight.
You are right, you cannot prove a truth, but I can disprove it. The point is that it was disproved easily.
I have no doubt that you can do the same thing for what I postulated. I just showed another example of the reason it might be able to look that way in some cases.
I am still convinced that google is just reading the information directly out of the databse in order. There is no reason for them to go to the effort of ranking them in any particular order. They are concerned with search results, Most people who run link will just go ahead and dig through the entire list.
Google supplies the link: command as a service. They don't make any adword income off of them. It is already somewhat crippled, so I don't see a reason for them to go to any extra effort to make the information you get from it more useful.
I think the only reason that there are only PR8, PR9 and PR10 pages for the first 1000 results (of course I didn't checked all of the sites) could be that the pages are order by transfered PR.
However, this proves at least that the results are not randomly ordered.
Perhaps there is no reason for Google to give this additional information (also it would be very usefull), but there is really no effort for them to do this. They have to check the PR anyway, since they show only pages with PR>=4.
--------------------------
10,9,9,9,8,8,9,9,8,8
--------------------------
8,10,10,10,8,8,8,8,8,8
--------------------------
Even if there would be a PR7 site with just one or two links, I would not see a violation of the assumption.
A site with PR <= PR6 or a PR7 site with many links would be a violation. Even in this case the statement that the backlinks are not randomly ordered would be still valid.
The 8th place is a PR7 employment page from infoworld. On the 16th page there is an infoworld PR4 page that has 76 links, followed by google's own PR10 /options/ page that has a PR10 and 59 links.
If you are going to make a statment about all 1000 backlinks, at least look through them and check that there are no obvious failures to your hypothesis. I don't check them all, but buried articles in a magazine generally don't rank a PR8-10. I also found several 6s.
Making authoritative statements like you did, without qualifying it as a theory, can mislead newbies since it is on Webmaster World, therfore it must be true.
Ten minutes ago, the first link of infoworld (the 8th place) was shown in my Browser with PR9. Now it is showing PR9 for some seconds and then changes to PR7. The link on the 16th page (infoworld) is coming up with PR8 and changes then to PR4. Also there is an additional infoworld on page 4 which shows only a grey bar. To my opinion this shows some strange Tollbar behaviour as reported [webmasterworld.com].
I didn't assume that there is a correlation between backlink listing and PR, but a correlation between backlink and transfred PR.
As already said, even if the theory is not correct the order is definitively not randomly as can be easily shown. Since I can rule out for one of my sites that the backlinks are ordered by first entering in DB (Dmoz is at #8, a new site with high PR at #2), they are ordered by an unknown parameter.
The order means not much of anything. Aside from parent domians tending to be on the left margin and sub pages indented, there is no there there to ponder.
sorry I misunderstood you. I thought you were suggesting that the results are in the order when the were spidered the first time from Google. (That is what I can rule out.) To see if the results are ordered by the date of the last crawling, I have to look in several log files. I'll do this the next days. Assuming that Google starts with his own domain this could in principle explain the large number of high ranked sites which normaly appear first.
steveb
I'll have a closer look to the changes of incoming links for the next update. (Unfortunately one can neither determine the exact value of the PR nor the changes compared to last month.) However, the reorder of your profile rank could be caused by PR changes and especially by transfered PR changes through modifications of the algorithm.