Forum Moderators: open
This subject particulary interests me because I am interested in developing a Web site which would by its nature contain very many links to pages found on external sites. I am now worried that my site would, in S.E.O. terms, be like a bucket with a hole in it!
Are my concerns justified?
[edited by: martinibuster at 10:08 pm (utc) on Aug. 22, 2004]
[edit reason] Edited for specifics [/edit]
No, it is plainly false.
Linking off a page deducts nothing at all from the page. Zero. It's simply the case that if 100 people say you are is cool, you get to say 85 times that other people are cool... without deducting one little bit from your own coolness.
What confuses people is that among that 85 cool votes, you can vote for your own coolness, and you can do it in more ways than one... and you can do it in a way that while direct is seemingly indirect.
With my cool rating of 100, I could direct links to the Disney main page, to the Adobe main page, and to the NASA main page. I just gave them each some cool votes, but I still am 100 cool though. No change whatsoever.
The exact same effect would occur if I had no links off my page, it just dead ends, but I stay 100 cool.
Now instead of me linking to Disney, Adobe and NASA, I instead link to three new pages on my domain: page2, page3, page4. These pages now will have pagerank. If we just stop there, and let these pages dead end, I still am 100 for the original page.
But now let's be sensible and link back from to the original page from page2, page3, and page4. Each of these now send *some* pagerank back to the original page, and now the original page gets a value of higher than 100.
So, linking off your domain didn't reduce your pagerank so much as it prevented you from raising your pagerank.
The basic concept is very simple: once you have the right to vote, you can vote for yourself.
But linking off a domain can also raise your pagerank, if what you link to links back to you. Pagerank doesn't care about the domain, only the links. If instead of linking to Disney, Adobe, and NASA, you link to the dmoz, google directory and Yahoo Directory pages that link to your one page, this will also increase the pagerank of your one page. Off-domain is not the point. Getting pagerank back one way or another from where you link is.
The real-world situation is obviously that linking patterns are uneven, so the RAW PR gets distributed unevenly too. The Google PR calculations you can do for your own pages might come to, say, an overall site PR of 30 - if you have 30 interlinked pages, and if you have incoming links from pages with PR, your aggregate site PR goes up even more (and go down if you link out).
And then Google adjusts all the raw PRs of the pages in its index so the maximum stays at 10.
I suppose the upshot is that an outgoing link does reduce the linking page's PR (as the Google Page Rank Calculator shows), but maybe only slightly, and there could be an unrelated and outweighing benefit from the other factors that Google uses to rank a page - plus, of course, there may be links in.
As a matter of interest, following a similar thread discussion here a few months ago (where BigDave stepped in, as I remember it) I stopped caring about PR for my sites. Now I just link however the fancy takes me. But then no-one is paying me as a Search Engine Optimiser (thankfully).
Everyone agrees that a high page rank page vote will raise the page rank of a low page rank page ( i hope).
So, if you link to a low page rank page on another site, then this is some rank that would have raised the page rank on another page on your own site had you linked to it instead.
So, linking out does "leak" page rank from your site, but not from the page you are linking from directly.
However, if you consider that your links , had you kept them on your own site, would raise the PR slightly of other pages on your site that in turn link back to the page you link out from , then it (in a roundabout way), does actually cause page rank leak from the page you link from.
However, as mentioned earlier, there may be advantages to linking to other sites, so its a balance you are trying to achieve.
For a site with more than just a few pages it is more or less impossible for anyone who is arithmetically challenged (like me) to try to PR-optimise the linkages with a pencil and paper. One thing I remain certain of (despite what steveb says, who I acknowledge to be an expert in this field) is that an outgoing link, all other things being equal, reduces the linking page's PR. By how much and what effect this has on the page's ranking is another discussion, but I believe it to be a fact. I don't see how linking to DMOZ from one of my pages (which I happen to have done, partly because they linked to mine - and yes, I managed to get a site in the Open Directory) can do other than cost me some PR.
Of course if an outgoing link brings a corresponding incoming one, then there might be a gain - or there might not be, depending on the relative PRs of the pages concerned.
Search Google for "Google Page Rank Calculator" and try the results.
"The probability that the random surfer visits a page is its PageRank."
Obviously, the more links that point to that page will increase the probability that the page is found.
PageRank has nothing to do with the length of time that the surfer stays on that page - as suggested by FreeJung in an earlier post. Therefore, the PageRank of the page is unaffected by its links out.
I did leave out one key element to my statement, page rank does leak from a site. A site being a collective of pages that you are in control of that each have their own PR.
In response to cbpayne:
3) Add 300 or so outgoing links on the page
Though it's true that this won't have an effect if the page is an island, if it's part of a site it will have an effect. However due to the logarithmic nature of PR and the wide range within a single digit in TBPR (the only thing we can measure) the effect would be so small that it wouldn't cause a change. Try taking a low PR website with a bunch of pages that interlink to each other and add 300 outbound links to every single page. Then you'll have a better chance at seeing the TBPR change.
I'm not trying to suggest that the sky is falling because PR leak is real. I'm just answering a question :) Jusr because PR leak is real doesn't mean that you shouldn't link out. It just means that you have to factor it into your decisions.
There are a lot of things in SEO that have positive impacts and virtually no negatives, but other components are a balance. Linking out is one of them. You need to balance the traffic that will leave your site, the PR leak and other factors with the benefits you get from it. It's a different equation for everyone.
PageRank has nothing to do with the length of time that the surfer stays on that page - as suggested by FreeJung in an earlier post
My apologies, I wasn't being clear. What I mean by "time" is time within the simulation, not actual physical time. Time within the simulation simply means the number of times the random surfer visits a page. So while linking off-site will not reduce the PR of a page that doesn't link in-site anyway, it will reduce the overall PR of the site if the page does link in-site. This is simply because the PR passed by each link is inversely proportional to the number of links on the page. So linking off-site reduces the PR that is passed to your other internal pages, reducing your overall PR.
However, I still think linking out is good, as I said. Just don't overdo it.
If I create a page with no outbound links and link it from my home page then, after a PR update, it will assume some PR value, say PR(x).
If I then add a new link from this new page back to the home page then after a PR update the PR of the new page will be PR(x+i), where i is due to the internal vote of the new page.
If instead my outbound link from the new page went to a page that in no direct or indirect way linked back to my new page then my new page would still have PR(x).
No matter how many out bound links I add to this page - the probability of it being found can never fall below x .
Added:
And as i is probably very, very small, there is no reason to stop outbound links.
And as i is probably very, very small, there is no reason to stop outbound links.
Floods are most often caused by tiny, negligible drops of rain :)
Just suggesting that looking at the bigger picture of your website makes a difference in figuring out just how much/where you want to link out. Adding a single outgoing link to a single page is not going to affect your PR at all. But what about tens, hundreds, thousands... At some point it's no longer a negligible amount.
Now if outbound links are what your main attraction is (think directory) then it makes all the sense in the world to leak your PR every chance you get. For others it's a different story. There's no blanket answer as to whether or not PR leak matters, it's got to be taken on a case by case basis.
I hope the discussion going on here will get people who are unsure about these things to take a long hard look at their sites in a global manner and really think about what's best for them and their visitors.
George Bush was "elected" President via votes.
He appoints Cabinet Members (main subdirectory pages).
The Cabinet Members hire workers (individual html pages).
Nothing in this process lowered the power of the President.
Bush takes care to appoint cabinet members, who will support him, and they in turn hire workers, who support them (and Bush).
Suppose the opposite were true: If Bush appointed Democrat cabinet members, who hired Democrat workers, then his power would surely be lowered.
So we have:
PR(A) = (1-d) + d (PR(T1)/C(T1) + ... + PR(Tn)/C(Tn))Still waiting for Doc_Z and Big_Dave to restore sanity to this thread :)
In most cases this effect is small and the toolbar won't show any change. However, even this 'small' effect can be measured.
As already said, this doesn't mean that adding outgoing links has a negative effect to the SERPS.
(1) 'Normal' means that the site has more than one page and there are no dead ends.
(2) For simplicity, 'higher order' effects (due to complicate connections) are neglected.
Just to clarify: An average raw PR of one is only realized if there are no dead ends. And PR calculation is only an eigen value problem for a damping factor d=1. For d<1 it's a set of linear equations solved by inverting the transition matrix.
Most pages gain PR by having some external inbound links plus linking out to child pages which in turn link back. If you make the amount of PR you pass to your child pages less by linking out more to pages which don't link back directly you will get less PR back from your child pages --> your PR decreases.
PR is like a vote, but the votes can flow back, even over an arbitrary number of indirections.
Steveb has a good explanation
That may be so, but it still seems to cause confusion where PR leakage is concerned. No doubt the effect of an outgoing link could be an increase in the PR of the page with the outgoing link (or other pages interlinked with it) but that would only be by the effect of an incoming external link that is a direct result of the outgoing link - a reciprocal. I believe it is a fact that all other things being equal, linking out reduces the PR of the page doing the linking and also the other pages it links to.
Oddly (until one is able to understand it by calculation) if there are two interlinked pages on a site and one links out, they will both lose PR but less PR will be lost by the page with the outgoing link. For me, it simply goes to show that the vote metaphor is misleading and that it's very hard to guage the effect of a linking strategy unless it is calculated in some way. For a site with a large number of pages, calculating off a micro-version of the site could be a useful exercise because it can point to a sound strategy for the whole site. Otherwise it can be a guess that turns out to be wrong.
Oddly (until one is able to understand it by calculation) if there are two interlinked pages on a site and one links out, they will both lose PR but less PR will be lost by the page with the outgoing link.
This is true - and not just for 2-page sites. The reason is that the page that links out only suffers indirectly.
This makes it even harder to evaluate whether Google will "reward" a page for linking out.
So can we agree that there is a leakage? :)
Sure we can. Providing that you and I are the only ones left in this thread :)
IMHO it's a matter of semantics: You can say that a site/page is "leaking" - or you can say that it's "missing an opportunity to increase its own PageRank".
At the risk of incurring the Wrath of the Mods, let me add that the site in my profile takes a closer look at PageRank leakage.
So, a reciprocal link increases PR?
So, a reciprocal link increases PR?
Only if the page that links to your page has higher PR and/or fewer outbound links than your page has.
As was stated earlier in this thread the average of the RAW PageRanks for all pages in the GoogleVerse is exactly one*. With your links you can funnel the PageRank around, but you cannot "generate" PageRank.
* This is not quite true because of dead-end pages and penalties.
If a website is a closed loop the rank vote eventually returns and boosts the original pages' final rank. With outgoing links some of this "potential" rank is leaked.
I put this to the test with a new website which did not link out at all. After three months of gathering one-way links the site achieved a PR8 with all sub-pages being PR7. This type of rank generally cannot be created through reciprocal link programs -- because of the leak of 'potential' rank.
So the gist is that a page does _not_ lose rank by linking out but it _does_ lose the potential of getting some additional PR boost by having it's 'vote' reflected back.
If you can get a PR5 through recips only then why not a PR8?
It appears to me that the following are valid operating statements based on my limited experience (my assumptions refer to visible toolbar results only):
A)
1) lets say I have a page, www.striped-widgets.com/index.html. It is a brand new page. There are no subpages, just the one page. There are no outbound links.
2) several pages on several sites with PR5 and no other outbound links link to my page.
Expectation - G spiders my page and assigns it a PR, based on the votes I get from the other pages. If the pages linking to me are actually solid, full 5s, I might be a 4. If there are enough of them, I might even get a 5, but no higher.
B) none of the pages linking to me will drop in PR because they are "voting for/passing PR to" my page.
C) If my page is now, in fact, a full, real PR 5, and I add a new page on my site with only one link (from www.striped-widgets.com/index.html) internally and no inbound links from any external source, when G spiders and indexes it, the toolbar should show a 4. If I add a hundred more internal pages, depending on the linking structure I have established, those pages may have various PR values, but my index page will remain a 5.
D) For passing outbound PR, the amount I can pass is equal to .85 of 5, divided by the number of outgoing links.
Please let me know if/where I am confused. Thanks in advance.
(I'll just sit over there on the prorch and pluck at my banjo while waiting)
Expectation - G spiders my page and assigns it a PR, based on the votes I get from the other pages. If the pages linking to me are actually solid, full 5s, I might be a 4.
Two things to remember:
Thus a single high TBR 5 page (raw PageRank 49,999) with a damping factor of 0.85 might beget (49,999 * 0.85 /5,000) – i.e.: 8 low TPR 5 pages.
If there are enough of them, I might even get a 5, but no higher.
If there’s enough of them, the sky’s the limit. But it will take a lot of PR5’s to make a PR10 :)
B) none of the pages linking to me will drop in PR because they are "voting for/passing PR to" my page.
That is the very subject of this thread :)
It depends on what those pages were doing before they started linking to your page. If they were linking to "foreign" page (i.e. one that didn't link back) or weren’t linking to anything at all – then they had nothing to lose.
But if they used to link to pages that linked back, they are now (as I wrote) "missing an opportunity to increase [their] own PageRank".
C) If my page is now, in fact, a full, real PR 5, and I add a new page on my site with only one link (from www.striped-widgets.com/index.html) internally and no inbound links from any external source, when G spiders and indexes it, the toolbar should show a 4.
No, that’s only a rough rule-of-thumb. As I just wrote, a TPR 5 page may spawn several new TPR 5 pages.
If I add a hundred more internal pages, depending on the linking structure I have established, those pages may have various PR values, but my index page will remain a 5.
Again no. Your inbound links determine your site’s total PageRank. With your internal links you are distributing this PageRank. The more pages you add to your site, the less PageRank there will be for each page. (Let’s ignore the subject of intrinsic PageRank since this is outside this thread).
D) For passing outbound PR, the amount I can pass is equal to .85 of 5, divided by the number of outgoing links.
Surprisingly this is not the case. For every iteration the linking pages are leaking (or missing an opportunity or whatever ;) ) and for every iteration the linked-to site is increasing its PageRank. The net value may be as high as:
PageRank * 0.85 / (1 – 0.85)Even more surprisingly (are you there Doc_Z? ;) ) the formula d/(1-d) is not set in stone, but depends on the linking structure of the linking site.
(I'll just sit over there on the prorch and pluck at my banjo while waiting)
Do you like J.J.Cale? :)
“Sit on the porch without no shoes,
Pickin' the bass and singin' the blues. ”
Surely that is central to the idea of generating "internal" page rank - giving you an opportunity to distribute this within your site?
No, period.
If a page has no outgoing links, a dead end, it has X PR.
If that same page adds a link to www.sustefdtrstfbicuehsj.com, the page will have the exact same PR as if it had no links.
Exact same.
Again, this is extremely simple and should not be made unduly complex. Perhaps the easiest way to understand it is this... if the above mentioned page has one single link off of it which goes to the root source of its PR, that is ideal from a PR standpoint. (Other linking structures could also be ideal, like linking to two pages that link to the page.)
Any linking that is not this ideal is simply that, not ideal for the PR of the page in question.
i created a few months ago a new page absolutely no outbound links only inbound links from PR 5 and
PR6 pages ,now it has a PR5
I created a few months ago a new page with several outbound links and only one inbound link from a PR4 page. It has a PR4.
Shouldn't we build pages for visitors?
If there are two interlinked pages on a site and one links out, they will both lose PR but less PR will be lost by the page with the outgoing link.
This has been said to be true and more or less sums up the theoretical scenario I'm talking about. One makes a site with, say, all 30 of its pages interlinked so that the sum total of the raw PR for those pages is maximised as 30 (raw PR 1 per page). An outgoing link (to an external page that is not within the 30 pages, and does not link back) will reduce the PR of all the pages though, as I say, the page with the link will lose less than others. This surely means that an outgoing link - in principle - leaks PR.
I realise that outgoing links can also (theoretically) produce an increase in PR because the scenario requires all pages within the calculation to be interlinked in order to build the PR up to its maximum, which obviously over the whole of the world wide web is never going to be the case. So the total raw PR over the whole of the world wide web is not maximised and can always be added to by more links. In that sense I can see how an outgoing link might not cause a loss of PR, but in reality the web is so huge that there is more likely to be a loss than a gain.
