Will links from the same site count less?

[WebWalla]

Given that links are from pages with the same PR in the following simple link structure, will domain1 get more PR than domain2?

example.com/page1.htm links to domain1.com
example.com/page2.htm links to domain1.com

example2.com/page1.htm links to domain2.com
example3.com/page1.htm links to domain2.com

i.e. do multiple links from the same domain count less than links from different domains?

Thanks.

[hakre]

hi WebWalla,

yes, links from different domains count more than links from the same domain. the pr will be calculated from all sites linking to the site/who link to 'the site'. the links from one domain you can think as 'one link only'.

[fathom]

In simply PageRank theory each link counts the same towards PageRank being passed (all other things being equal).

Weighing to determine rank position has 100 variables which are considered therefore answering your question is rather moot since in all likelihood all other things would not be equal.

do multiple links from the same domain count less than links from different domains?

It is believe that unique links (different domains/IPs) offer a slight advantage in weighing. More to the point if most or all inbound links resolve to the same domain/IP and this is many without many other unique domains/IP the chance of being penalized increases.

So yes - in all likelihood Google prefers different links from different sites.

[ciml]

PageRank flows the same whether there are five internal links, five links from some other domain, or five links from five domains.

As fathom points out, the effect on rankings is more complex. In each of WebWalla's situations, the links will give a better link text boost than if the two links were from other pages on domain1.com

I don't think it matters whether they both come from example1.com or if one comes from each of example1.com and example2.com (but I think a lot of people would disagree with me).

[aek]

There has been quite a few threads about this recently and no one has given any solid evidence to convince me that 10 links from different domains is better than 10 links from the same domain.

Common sense would suggest that 10 links from different domains should give you a bigger boost than 10 links from the same domain (all other things being equal) but from my own personal experience this is not the case.

[ciml]

aek, do you see a difference between the link text affect of ten links from pages on the domain they link to and ten links from pages all on some other domain?

[aek]

Sorry Ciml, I don't really understand your question, could you rephrase it so that an idiot can understand it :)

[NickCoons]

WebWalla,

Google's original PR algorithm does not know what a site is, it only knows what a page is, and PR flows the same from a page on the same site as it does from a page on a different site.

Google may have modified the algorithm since then to change the weight of links from the same site, but I have not necessarily seen that to be true yet.

[ciml]

aek, you've passed the 100 post mark on WebmasterWorld, you simply can't be an idiot. :-)

Three situations, the links are all from identical PR pages.

1) 4 links to www.domainA.com/page1.html from pages on www.domainA.com

2) 4 links to www.domainA.com/page2.html from pages on www.domainB.com

3) 4 links to www.domainA.com/page3.html from pages on 4 different domains.

page1, 2 and 3 all have the same PageRank, but which rank higher than which for the link text of all those incoming links?

[aspdesigner]

There is some belief that external inbound links might provide other ranking benefits.

But from a PR standpoint, ciml is absolutely correct, there is no difference.

[taxpod]

I think that the problem here is that it is virtually impossible to know exactly what you get from each individual link. The pure theory offers no suggestion that links from different domains are weighted more heavily than internal navigation links. The pure theory is simply page to page. But logis would indicate that there has to be more at play. You can work through an example where a domain has one inbound from say DMOZ creating a PR3 home page. Next you put 100 pages behind the home page and design navigation to push more PR up to the home page. Then you draft another example where this home page with one inbound links has a 10,000 page site. Mathematically the larger site's mhome page should rank higher for just PR.

Now give the smaller site's home page another inbound link from a page with PR3. The results might be that the smaller site's hoime page would have higher PR than the larger one but that doesn't indicate that the external link is better than the navigational links from the higher PR site.

All I'm saying here is that it is impossible to prove one way or the other unless you can control all the other variables and unless you can see the real PR of all these pages. You can work through examples using the pure theory and make a conclusion but then you have to ask if the actual PR algo hasn't been tweeked. There is no way to KNOW.

[aek]

Ciml, unless I'm reading this wrong the first two situations are identical except that they are different pages on the same domain so I think they would definitely be the same.

However I think the 3rd example is also the same with respect to ranking (although many will disagree).

If you are to believe the original pagerank paper, ranking is done on a page by page basis which would mean they would all rank the same, and I think this is still, for the most part, true.

Note: All the above examples are based on having the same link text on all incoming links and ranking for the same keywords.

[vitaplease]

Will links from the same site count less? (pr-donation-wise)

This thread: [webmasterworld.com...]

suggests that Google does take some site factors into account.

They probably average the Pr donation of all links from that same site and donate one.
Simple webmaster/client signing precaution.

[ciml]

taxpod, may I shamelessly put two parts of your last paragraph together in a different order to say what I mean?

There is no way to KNOW unless you can control all the other variables.

aek, it seems that our experiences don't quite agree then. I get page 2 and 3 equal, beating page 1.

vitaplease, the thread you mention is an unusual case, IMO.

[aek]

Vitaplease, that's a very interesting thread, I think it's very strange that BBS has a page rank of 4 considering the backlinks from WebmasterWorld.

I have high ranking sites purely because I link to them from every page of other sites I own and they get the expected page rank from it. Although I have a few links from other domains.

I'm wondering if perhaps google penalises/lowers pagerank, if all, or almost all incoming links are from the same site.

[edited by: aek at 5:00 pm (utc) on Feb. 22, 2003]

[figment88]

I think that multiple links from one site used to count the same as single links from multiple sites but this changed about 4 or 5 updates ago.

I think in the current system Google performs what is called in data analysis a right-censoring (i.e. they cap it). I'm not sure what the number is but say it's three links, then

1 is more than 0
2 is more than 1
3 is more than 2
but
4 is the same as 3
5 is the same as 3
.
.
.
n is the same as 3

Alternatively that could be apply some type of transformation, so
1 is a lot more than 0
2 is much more than 1
3 is slightly more than 2
4 is negligbly more than 3
.
.
.
n is negligbly more than 3

The world is non-linear; there is no reason Google's algorithms have to be.

[fathom]

aek, you've passed the 100 post mark on WebmasterWorld, you simply can't be an idiot. :-)

ciml - I must now shuutdown to cllean tthe cofffee out of the keyboard and off the monitor.

I can'''t evenn type ttxk

[aek]

100 post mark!

Does that mean I can put my spamtastic new website in my profile?

[Shadow]

"The pure theory offers no suggestion that links from different domains are weighted more heavily than internal navigation links. The pure theory is simply page to page. "

This is absolutely not correct. Google did distinguish between within same site links and links between external sites fairly early. It's appearant from what's available of Page's writings.

[aek]

Shadow, could you give me a link to where you have read this, as I don't remember this being stated in the original page rank paper.

[allanp73]

Just one more thign to consider about page rank. Do sites that are on the same IP server offer more PR via links than sites residing on separate IP's?
Does Google look at which IP is providing the link?

[aek]

IP is irelevant when it comes to links AFAIK.

[Shadow]

aek,
I'll have to dive into my saved stuff, but I'm right about this :) . Just to clarify, I was referring to internal links having a different weight than inbound external links. I don't want to conclude anything about several links from one domain to a second domain being capped or something, I'm still looking into this. It's difficult.

I agree with you on the IP thing.

BTW, using dmoz in PR examples/testing (as taxpod mentioned in post 11) will not provide adequate data because dmoz data is seeded to a lot of smaller directories. Though small sites, those seeded directories will transfer some additional PR to the listed domains.

[aek]

Shadow, I completely agree with you that external links carry more weight than internal links, I must of misinterpreted your post.

[Shadow]

aek, I suspected someone would, I was only replying to the
"The pure theory offers no suggestion that links from different domains are weighted more heavily than internal navigation links. The pure theory is simply page to page" and "Google's original PR algorithm does not know what a site is, it only knows what a page is",
both posted in the thread, and both wrong, - though maybe not from the very beginning, but at least from 98/99.

[NickCoons]

Shadow,

<This is absolutely not correct. Google did distinguish between within same site links and links between external sites fairly early. It's appearant from what's available of Page's writings.>

I would like to see a link to this as well, because I haven't seen anything definitive as far as how internal versus external links are weighted, and I know the original algorithm does not take that into consideration.

[fathom]

NickCoons your first statement was quite accurate -- you indicated flow - PageRanks flows from link to link - internally & externally the same.

The moment you mentioned weight and directly in context with PageRank -- weight implies scoring, and that being the case, if only internal pages you have no weight of PageRank. PR0 pages on an unpenalized site proves this.

The original thread topics (and posts before your msg#8) were discussing external links only and in the context of WebWalla
1st msg.

[NickCoons]

fathom,

<The moment you mentioned weight and directly in context with PageRank -- weight implies scoring, and that being the case, if only internal pages you have no weight of PageRank. PR0 pages on an unpenalized site proves this.

The original thread topics (and posts before your msg#8) were discussing external links only and in the context of WebWalla
1st msg.>

Then perhaps I didn't phrase it correctly. But the point was that two links from different pages on the same site pointing to a third site should not have a different impact than two links on two different pages on two different sites pointing to that same third site. I believe that answers WebWalla's question.

[fathom]

hmmm... I think I understand that and agree.

I actually directly avoided the analysis of the suggested hypothetical link situation because in the real world, the precise situation would never materialize. (without no other influences)

[NickCoons]

fathom,

<I actually directly avoided the analysis of the suggested hypothetical link situation because in the real world, the precise situation would never materialize. (without no other influences)>

It's interesting how we often include the phrase "all other things being equal" in our discussions, when all other things are rarely, if ever, equal :-).