| This 72 message thread spans 3 pages: < < 72 ( 1  3 ) > > || |
|Do outward links increase pagerank?|
will adding outward ranks help?
I have a site which has only a few of the pages spidered by google.
I have submitted my site to about a hundred other directory sites, where most of them have an individual page setup for me, where a link to my home page is included.
If i create a sitemap of sorts where i list all these pages on those sites where my site is linked back to, will that increase my pagerank?
As technically the googlebot will find those many more inward links to my site?
Thanks in advance
I agree, we disagree.
However, my main arguments are not only taken from theory but from practical examinations. This means not only studying the normal pages of some sites but building numerous pages exact for this purpose. For example, one could do the following thing: create some pages with the same PR and examine what happens if these pages link to one, two, ... N pages. If your assumption about internal linking would be right this would result in the same PR for each level independent of the number of pages (as long as N < 100). However, that's not what you'll find ... (just try it)
In a similar way you can study all other PR problems.
|There are people that have "have studied and veryfied" the other side of the coin too. This is why the statement "Truth is no-body here really knows whether outbound links decrease PR or not." is the only *fact*. |
I already gave an example how to examine PR in a scientific manner. I have made numerous tests on different domains and comparing the results with the data from other studies by different people. The results coincide and lead always to the same conclusions. These results are reproducible.
If someone else made tests (in a scientific way) which yields different results, I would be very interested in seeing and discussing them - please contact me.
|How do you explain directories, having far more outbound links than inbound, having good PR? |
... because of numerous high quality incoming links and a logarithmic toolbar scale with a large base value.
To find out if outgoing links decrease PR or not you have to compare two sites with identical page structures with identical incoming links but one with outgoing links and one without at the same time. That's the only way to investigate this question.
Put up a brand new directory with a pr7 site linked to the homepage, no other links in anywhere..... the internal pages have predicatable pr throughout the site structure, irrespective of the number of links out on each page. Over the course of time new deep links in effect certain pages, but it proves the theory that links out does not effect pr.
I believe there are many other penalties that effect pr, and it is possible that these factors are what you are seeing. Links are a fundemental foundation of the web and will never be penalised. Google likes links, that is how they find pages to index.
"To find out if outgoing links decrease PR or not you have to compare two sites with identical page structures with identical incoming links but one with outgoing links and one without at the same time. That's the only way to investigate this question."
Also you will need the same content, if you believe some words (such as sex words) trigger penalties. So then you have two duplicate sites.... another problem. Thus it is impossible to do 'scientific' tests.
This is where too much annalysis becomes unhelpful and can give the wrong results. Returning to basic logic is better, and concluding that links out is not a sign of a good page or a bad page is a sensible assumption. I'm sure Google would and does agree with that.
Why does Brett tell us to link to a high page ranking site from our main site in his golden rules bit if outgoing links have no effect?
|Also you will need the same content, if you believe some words (such as sex words) trigger penalties. So then you have two duplicate sites.... another problem. Thus it is impossible to do 'scientific' tests |
You don't need the exact same page if you are only interested in PR distribution. PR is based on link structure not on content. Also, even with simple modifications, the page won't be affect by any kind of dublicate entry problem. Also, the dublicate entry problem should only affect the SERPs for a specific keyword but not PR.
Therefore, tests can be easily done without any problem.
|This is where too much annalysis becomes unhelpful and can give the wrong results. |
'annalysis becomes unhelpful' - I haven't heard this before ...
There is no problem with scientific tests - that's how science works: developing a theory and look if it's in agreement with reality or not. The results of those PR tests coincide with theory as well as analytical and numerical calculations and they are in agreement with logic.
Just because the results aren't in agreement with your 'basic logic' doesn't mean they aren't helpful. How are you verifying your asumptions? Or isn't it necessary?
"PR is based on link structure not on content."
How can you be sure? Have you 'scientific proof'? What about pr penalties? This is a huge assumption you are making which undervalues any further 'scientific' conclusions.
"'annalysis becomes unhelpful' - I haven't heard this before ..."
Can't see the wood for the trees? - Lies, damn lies and statistics?
Have you heard of those? :) I'm just suggesting that with so many variables in the Google algo we run the risk of believing we know what is going on. The safer and probably more helpful approach is to frequently return to basics.
e,g, Give google good quality pages with, if required, loads of links out. If it is good for the user then have faith that Google is trying to find these types of pages. You can annalyse everything until your blue in the face, but if that results in you modifying the pages to suit googlebot rather than the user, then you run the risk of losing sight of both their objectives and your own... namely a good website. If you restrict links out at the detriment of the page just because of seo reasons, then you will ultimately lose out on user satisfaction and possibly rankings.
The three things certain in life are Death, Tax's and the google algo being scientifically annalysed :)
|Have you 'scientific proof'? |
|If you restrict links out at the detriment of the page just because of seo reasons, then you will ultimately lose out on user satisfaction and possibly rankings. |
You should read my posts. I already mentioned that I have numerous outgoing links and that outgoing links can improve ranking.
|.. but if that results in you modifying the pages to suit googlebot rather than the user, then you run the risk of losing sight of both their objectives and your own... |
I never said that I've modified my real content pages in any way. Of course, building a good, useful website for users is the main thing - and I didn't said anything against it.
However, the original question was about the influence of outgoing links - neither about reciprocal links nor about something else. The original papers, any kind of calculations as well as measurements (in principle the only important point) give a clear answer. What else can one say/argue? I haven't heard any kind of argument or proof against this. If you are able to give any arguments, I would like to hear them. (By the way, I'm still waiting for a mail of someone who derived different results.) All I have heard so far was "NONSENSE", "basic logic", "annalysis becomes unhelpful" and "blue in the face". That's not my way to discuss.
As long as no arguments are given, I'll stop this 'discussion'.
If anyone derived data which disprove my results, I would be happy to see them. I never had a problem correcting my statements if there were disproven. However, you need reproducible data.
"However, the original question was about the influence of outgoing links - neither about reciprocal links nor about something else."
Which to remind you, this thread is about reciprocal links. Either focus on this thread, or start a different one.
|As long as no arguments are given, I'll stop this 'discussion'. |
The Ostrich said... :o)
Looks like ive started a controversy here :)
So what the final verdict guys?
Do i set up that links page or not?
Please Vote :)
Better ask doc_z, he/she knows how Google operates better than Google themselves :o)
Seriously though, I would go for it. Google thrives on links.
What doc_z said
|Outgoing links decrease PR. However, they can improve your ranking even if PR is decreased. |
is basically correct and in accordence with my own experience and analysis. Although as martinibuster says
|I think we have to be careful how we state this |
But the answer to your question
|Do i set up that links page or not? |
is yes because PR is only one consideration of many.
At the top of my main index page, I have several graphical links. These are to Google and Hotbot plus several major companies in fields such as anti-virus technology, firewalls, etc. i.e links to sites that I think visitors may find useful. I also have another specific links page that discusses sites useful to software authors. Again, these are mostly high PR sites.
All links on my site are plain HTML.
I cannot say whether outgoing links do any good, but I am certain that they do no harm. People keep saying that outgoing links reduce a page's PR. Whilst I am no expert on Google technology, this seems irrational to me. I do recall that their algo divides the PR passed on to other sites (equally?) between all the links on a page, but I don't see that adding outward links on page is likely to decrease its PR. Of course, high link-density might trip a filter if the page appears to have no other content but that's all.
My site has never suffered from spidering problems.
Folks, read doc_z's posts. Reread them. And keep in mind that PageRank and rankings are two different things while reading them all over again.
All of doc_z's statements comply with my calculations and observations. And guess what, I have outgoing links on my sites, too.
It's a shame to see people who give away well-founded information being confronted with so much ignorance.
|Outgoing links decrease PR. |
Must we believe this unproven *theory*?
Novasoft, you basically asked some very difficult questions, although they sound simple. There are a lot of dependencies in them - and they (obviously) touch upon subjects of great importance (hence the discussion).
It might help to consider that the amounts of distributed PR can be very-very small and that any given PR measure is never really a constant factor for more than a limited period of time, as the www changes constantly. One can never say "Yes" or "No" except for special cases, as the "Yes" (and "no") will never really be an absolute "Yes" (or "no") rather, it will be a "Yes" (or "no") within some tolerance level.
If you consider changes below, say, just an example, 0.0005% in value to be nothing, then it's a "no". But then again, it's not really a no, as a change is really taking place, it's just a small one. Add all these minuscule changes along multiple runs through the entire 3 bio dataset and you get some quite impressive effects, but that's beyond any individual webmasters realm.
Now for the questions:
You have a lot of incoming links to a (directory?) site. Your problem is that it is not deep spidered. You want the index page of that site to get higher PR hoping that this will get the spiders to go deeper. You have two questions - i'll try to answer reversely:
>> technically the googlebot will find those many more inward links to my site?
If these directories suffer from the same problem as you and hence the pages containing links to you are not spidered, then: Yes.
Reverse problem - that "external sitemap" (placed on your site) will need to get spidered first. Then again, it will not be so unless it's there in the first place, so go ahead and try it.
>> If i create a sitemap (...) will that increase my pagerank?
Here, as i interpret it at least, "pagerank" refers to the PR of your index page.
Yes and no. It will increase your index page PR by a minuscule fraction of next-to-nothing just by featuring a link back to your index page. Adding a lot of outbound links on it will give it content and keywords as well, but it will not change the PR of your index page more than adding the page has done already.
On the contrary, the PR that this link page distributes on to the total 3 bio. page system is divided among all the links on it (which include a link to your index page) so that small fraction of added PR to the index page you got just by adding the links page will be even smaller now. That's a "very close to no" type answer.
It's valid in the first round. Here comes the second, and the reason why i reversed your questions. If the discovery of this site map (supposing it gets spidered) will lead Gbot to discover new pages featuring a link to your index page and spider these pages so it "sees" that link then ...still no.
Enter third round. Here's another dose of...ehm, controversy. Because, what does Gbot do when it discovers a brand new page? It can hardly give it any PR as it's not possible to run through the entire 3 bio. set on-the-fly. So, let's just skip speculation for the moment and note that this new page that links to you will get a "white Tbar PR" (which is not really zero, but not really anything either.)
The PR that this page "recieves" is derived from the on-site-links plus your added link page. So "incoming PR" to this page is a minuscule fraction of next-to-nothing divided by the number of links you have on your links page plus (the unknown value of) PR from the internal pages of that other site.
Then, it has to pass PR on as well - not only to your index page, but also (i presume) to other pages on that site. The PR you get in return will then be (hold on tight now): "a fraction of a minuscule fraction of next-to-nothing divided by the number of links you have on your links page plus (the unknown value of) PR from the internal pages of that other site. "
Add 100 links to the links page and this times 100 is what you will get in return. This, in turn, will perhaps be lower than the original "minuscule fraction of next-to-nothing" that you got in added PR just by adding the page (before you put any other links on it than the one to your index page). It might also be equal to. Regardless, some PR is added to your index page from this page (if it's equal to zero or not is up to you to judge, but it is a positive (very small) number)
This is extreme fine-tuning, and it's probably not just what you need right now, but i'll get to that. Enter forth round:
Gbot has now discovered some brand new pages. And time goes by. What does Gbot do? Well, it crawls. Not only the brand new pages, but other pages on those sites as well. And time goes by. The big 3 billion Google machinery works, it chews on all those pages and spit them out again. Voila - some of them might get PR (meaning: More PR than the minuscule fract...whatever).
Here's the good stuff, now: If any of these newly discovered pages now have more PR than they had in rounds 1,2,3, then there's more PR to pass on to your index page. Congrats!
It might still be very little amounts though, so as to not even move your Tbar PR one notch. OTOH, you might also be lucky and G might find that one of them is a PR7-8 which is very unlikely but not totally impossible. Again, we are talking very small numbers here, probability-wise.
Yes, do it. No, don't expect a lot from it (*). As i said:
|it's probably not just what you need right now, but i'll get to that. |
Get deep inbounds. That is what you need, and three words was really enough. Anyway, you asked for it ;)
(*) Strictly PR-wise. As others have pointed out, such a page might have other effects that are not PR-related.
[edited by: claus at 12:24 pm (utc) on Sep. 18, 2003]
Must we believe this unproven *theory*?
It's not a theory. It's the normal behaviour of PageRank distribution.
There is not need to "believe" in something that you can simply verify just applying PR formula to a set of pages.
The only problem is that many people think that they should avoid outgoing links to keep all the PR within their site. This is simply a bad move, since outgoing link are useful both for users and for SEO.
Been busy for a well... glad to see the discussions haven't got so everyone agrees! ;)
Just thought I would add... from a purely PageRank perspective this is a bit like playing the lottery.
Outbound linking is a gamble (instead of giving all your internal pages to the maximum they could get - you give a little to someone else in the hopes of getting some/more back as a result.
The article I wrote is reflective of sites high in standing in your specific industry, thus they link to others (as high, higher, or slightly lower in standings and even to other industries.
If there is 14 million possibly number combinations in a lottery, and you buy 13,999,900 of the numbers the chance that "you win" is extremely good.
Linking to authorities hedges the bet in the same direction.
All their links go somewhere - and many by proxy normally go to you - you are in the same industry -- right?.
Admittedly, the more connected you are (your inbounds) the more of these return, more to you.
[edited by: fathom at 11:54 am (utc) on Sep. 18, 2003]
PR leakage and thus less returning to a homepage is certainly so small that it is hardly worth worrying about. I suspect that recently internal links are treated so differently that there may be a completely different set of rules for them. Doing absolute links rather than relative seem to make a difference, but I can see no longer any evidence of this.
We put hard non recipricated links out on every page of a site about a year ago and no negative effect was noticed as far as pr/rankings were concerned, in fact it may have helped. However, with the toolbar being so primitive it is hard to say.
I'm still unconvinced links out lowers pr, even with miniscle leakage it is an overall 'no' as the benefits of anchor text and possible 'hub' status is far more important.
Markus - Sorry about the ignorance, but hopefully I am learning.
|It's not a theory. It's the normal behaviour of PageRank distribution. |
Until/if it becomes *fact*, it's a theory.
|...even with miniscle leakage... |
And this is a very big point.
If for example 20 links are on the page - 19 are internal and one links out - that 1/20th of 85% -- how much is that?
Do the incoming links to your site have a high PR? I am assuming you link only to sites that have a PR higher than 1 so you don't accidentally link to a "bad neighborhood" ...
Unfortunately good questions like this one end up getting sidetracked and hijacked by irrelevant tangents.
PR is a tool that every webmaster has available, and its not just a tool, but a variable and maleable tool.
If you own a domain and get a link from a solid PR6 site to your index page, you will (likely) have a PR5 index page. Okay, now you could stop, and the PR sent to you would just die. You would have a PR5 one-page site, but you would not be using the PR under your control now to make your situation better.
If you were to then make ten child pages which each have one link from your index page, you will have created ten PR4 pages all by your lonesome. If however those PR4 pages do not link to anything else, once again you squander your available PR power. If those ten PR4 page now link back to your index page, the index page will go from PR5.x to PR5.x+. There will be *some* benefit.
Likewise, if instead of dead ends, you link those PR4 pages to the type of off-domain directory site pages described in the first message of this thread, and those sites link back to your index page, then your index page will become PR5.x+-... better than if you did not make these off site links, but less good than if you directly linked back to yourself.
The bottom line though is absolutely clear: having off-domain links can positively raise your own PR. This is simply because Google's PR is not voodoo or close-minded. You must consider the effects of your linking. Put another way, even if those ten PR4 pages linked out, and none of the pages they linked to linked back to you, but *one* site linked to another site that linked to another site that linked to the PR6 site that linked to you, you would *still* get an utterly miniscule PR boost to your index page because the PR6 site went from a PR6.74321 to a PR6.74323. The links you send out into the world can come back to you in microscopic benefit after three or six or one hundred degrees of seperation.
On the other hand, if you link from those ten PR4 pages to the homepage of NASA, you will have given away your PR power instead of using it to increase your own.
Not considering where the PR goes after you send it is nonsensical. There is no point even talking about that. In the vast majority if cases the best use of your own PR power is to link to your own pages under your own direct control (where you can then vote again with the PR you have just distributed to yourself). In some cases however, linking off domain can have a similarly positive effect. Off-domain linking will usually be an inefficient use of your linking, but sometimes it will be an excellent idea, both for PR and for other related reasons (which will likely be much more helful).
I like the "site authority" comment mentioned by? Forget PR, or at least in my industry there isn't anyone over a five. Just checked...toolbar must be broken again:( Gonna test this and see what kind of an effect it has with a two word phrase I've been trying to gain ground on.
It's been stuck at #19 with google and yahoo since moving up from #50 or so. Do you guys think 50 or so outbound text links with this two word phrase and their company name will do the trick? The phrase is commonly used with the names. Maybe I shouldn't place them all on the same page?
|On the other hand, if you link from those ten PR4 pages to the homepage of NASA, you will have given away your PR power instead of using it to increase your own. |
And you would be wrong!
If I have a space related website: (and I do)
I link to NASA, (I actually did)
NASA has a page for Hubble that links to Hubble.NASA, (they actually do)
Hubble.NASA has a credits page (PR8) that links to DMOZ (they actually do)
DMOZ has a Science category that contains, a sub-category Astronomy, which contains a sub-category Solar System, and (they actually do)
In the sub-category Solar System, I have a link from DMOZ coming back to my site which, (really I do)
just so happens to link to the page I linked to NASA on.
Names have not been changed to protect the naysayers! ;)
|Not considering where the PR goes after you send it is nonsensical. There is no point even talking about that. |
Am I to understand that PageRank Leak only exists at my site? You really can't be saying I am getting nothing from this?
I actually shorten this up... but there are precisely 18 pages that link in a circle... starting and ending at mine.
PageRank wasn't developed for the "less connected".
|More Traffic Please|
Aside from all the calculations mentioned in this thread about PR, I seriously doubt that Google would want to send the message that their algo reduces the PR of a page because it has links to external sites. That is, with the exception of their comments about a reasonable number of links such as 100. To do so flies in the face of what the Web is all about and would only promote more PR hoarding.
"I seriously doubt that Google would want to send the message that their algo reduces the PR of a page because it has links to external sites"
It is entirely straightforward, despite the confusion on one hand and the denial on the other. PR is about "voting". You can vote for yourself. You can vote for folks who vote for you. You can vote for sites that don't vote for you but give you some authority/hub/link text benefits. You can vote for sites that return nearly absolute zero to you *in terms of PR*, and that PR benefits those sites. This is not "reducing the PR of a page", it merely denying yourself the ability to use that PR again yourself. It is you dropping some of your pennies in the plastic thingee at the 7-11 which enables someone totally unrelated to you getting the benefit from them.
Linking to other sites gives them pagerank. That's it. That's the deal. Those sites may in fact return pagerank to you, but all single links distribute PR. If you distribute PR to a domain you do not control, you should have a good reason.
It may be true that linking to external sites is a waste of potential PR distribution (that could go entirely to internal links) however, it is only true if Google choose to make it true. I would imagine that no one is in any doubt that Google can distinguish between internal links and external links (within a domain) therefore Google may choose to allocate PR distribution (amongst links) separately for internal and external links. Indeed, this is exactly what I would do.
Given that published (and believed) information on Google algos is limited, I cannot see that anyone outside the organisation can speak with authority on the subject of PR distribution. Everyone has their own theories and which are correct may change from time to time as Google try to overcome the excesses of SEO.
|I cannot see that anyone outside the organisation can speak with authority on the subject of PR distribution |
This is the single biggest problem with this forum. Too many people *think* they can and then try to pass it off as fact.
One thing you should keep in mind is that the PR of your site, and any site you might be linking to, is just a snapshot in time.
I have a high PR site, and I don't have any problem linking out to brand new sites if they are in my niche. I do ask them to link to me from their homepage if they are open to it, and most say yes, and jump at the chance of getting a link from me.
Guess what. A month or two after I do that, almost every single site I link to, jump up to a PR4 or more, and the link on their homepage now helps me.
Just another thing to keep in mind when you are debating the whole linking/not linking arguement.
One man's PR0 site today is another man's PR5 site a month or so down the road.
Don't be stingy, and you'll find you have made a lot of friends with growing PR sites.
I vote for worker. :]
| This 72 message thread spans 3 pages: < < 72 ( 1  3 ) > > |