Forum Moderators: open
PR hopes to measure the likelihood that a random walk will find a given page. But that's the statistical walk of a bot - where a link is a link is a link. For a HUMAN user taking a random walk, some links are much less likely to to get a click than others.
Do you think Google may have some factor such as Link Prominence? Some algo that attempts to measure, even in a rudimentary way, how important a link is on the page?
For instance, one simple and easily doable measurement might be to upgrade the value of a link that occurs inline - within a paragraph surrounded by other text.
As I said, this is a purely abstract "thought exercise" for me right now - but one which I may test in the near future. If I were trying to create a good search engine that accounted well for the value of links, I think I'd look at something like this.
I tested the position of the link (i.e. is it the first link or the fiftieth link on the page) and it didn't seem to make a difference to PageRank transfer.
It would be a little surprising if Google already gave some links more PR prominence than others yet, but there has been talk of link text maybe couting more weight if it's in a heading.
Page Rank factors aside, if there's a link within a paragraph of text it seems it would have more contextual relevance. A paragraph telling about yellow widgets with a link inline to the yellow widgets page is more meaningful contextually than that same link within a site-wide navigation menu would be.
I don't know if it carries any different weight, but might if the words that are near a link are taken into consideration, particularly if there would be a difference in the value of an on-theme link as opposed to a random, unrelated link.
As you said, the bot can't know what a human would do. But, suppose you had something that could tell you what the human actually clicked on? Say for instance a browser addon utility that reports clicks back to homebase. aka: an ie toolbar [toolbar.google.com].
Links within a paragraph structure would be easily identifiable. Some of us think they do receive a bit more weight.
Links within a paragraph structure would be easily identifiable. Some of us think they do receive a bit more weight.
If that's the case, Google should rethink its policy. Links that are integrated into body text may be less important than freestanding links, depending on the Web site's editorial philosophy.
On my own site, links that I regard as important are generally presented in this format:
Name of link in bold text
Annotation annotation annotation annotation annotation annotation annotation.
Links from within body text are likely to be citations or nonessential links. (For example, in a paragraph about a hotel designed by Helmut Jahn, I linked from Helmut Jahn's name to a biography of the architect--but that link is far less important to the reader, or to the subject that I was writing about, than the freestanding annotated link to the hotel's own Web site.)
in a paragraph about a hotel designed by Helmut Jahn, I linked from Helmut Jahn's name to a biography of the architect--but that link is far less important to the reader, or to the subject that I was writing about
Looking at the page the link is on though, if Helmut Jahn's name is in the page title, an H tag and also has that link to his biography within the right page density, your page that the link is on could end up ranking for a search for Helmut Jahn, depending on how competitive it is.
- position on the page
- the display format of the text (font size, boldness and color emphasis, etc)
I assume that this might carry over to link prominence as well... ie, a link in very tiny type at the bottom of the page might be less prominent that a bold link at the top.
I would also guess that repetition of links on a page might carry some weight... ie, if a link to a page is on both the top or side nav bar (in text or as a graphic), and is also in the bottom global links, that this might be seen as a more prominent link.
Whether Google, or other engines, have the resources to split hairs as fine as we do is another matter. I try to cover all bases... within reasonable design parameters... as long as I'm not driving myself crazy.
I was analyzing it the other day and discovered the reason why it was ranking:
It had a set of text kw links on top of the page, with h2 tags.
Conclusion: I do think this is in some way part of the algo.
There are two different valutations of links:
1. The "Pagerank-gift" of a link
2. The "Ranking-gift " of a link
I think you mean to discuss the Pagerank-gift of a link relevant to its prominence? as the ranking-gift prominence, in my opinion is already in place.
I would say that not all links are equal (Pagerank-gift wise) already at this moment.
Imagine the indexpage of a webmaster signing all 1.800 pages of all his clients webpages with a link towards his indexpage.
If all links are equal Pagerank gift-wise, that could be quite a Pagerank boost towards his indexpage. While there is nothing wrong with signing your work, I would suppose that every webpage has a properties file checking incoming links and neutralising too much self-complimentary Pagerank inheritance.
But your question reverts more to on page factors:
Could it be that in the future, a sponsored link in graphic button with "sponsors" in the surrounding text would discount the Pagerank gift if the Pagerank of that page is rather high (anti-Pagerank-buying-measure)?
Could it be that Google plays with different dampening facotrs per link?
That is, a page also linking to other resources outside the website, gets more advantageous dampening factors for its internal links because the webpage is being hubby and offering the visitor more "independant" external sources of information?
I am convinced links deep within a paragraph carry more weight.
I was setting up a site a few months ago and have not been able to finish it. It has an index page with a list of links to 10 seperate pages... all with just a title and one line of identical text (whoops). Also on the index page is a link within a paragraph to an 11th page also with the same text...... this page gets very good rankings! The interesting thing is that it is not the anchor text of this link which is important, as the page ranks well for different phrases, often used in the anchor text of other links to other pages. This link seems to have become relevant for the 'theme' of the paragraph it is set in, thus appears on many searches.
I am under the firm impression that google is trying to figure out a page's 'template', and strip it of common elements.....i.e. links in a nav bar, common setup at the bottom of pages is weighted less (if at all).
I honestly think this is one of the things that went into the new algo last month. It would make sense from a user's perspective, as a surfer we just want the 'meat' of the page.
It kinda sucks for those of us who worked hard to get proper keywords for our templates though.:(
"Detecting common templates in pages, and separating out the common structure from the individual content."
I'm sure the team that developed the news search spent a lot of time creating code that would achieve that goal.
Once again doing good for usability can do you good for google. In this case, context sensitive navigation (e.g. breadcrumbs, "Other's shoppers bought", etc.) can help greatly.
I think you mean to discuss the Pagerank-gift of a link relevant to its prominence
Yes, exactly that. Some adjustment to the PR equation so that not all links pass on the exact same gift. An adjustment like that would stand a chance of increasing relevance and undermining artifical linking schemes, at least to a degree.
Prominence was just the best word I could come up with at the time - but
I am under the firm impression that google is trying to figure out a page's 'template', and strip it of common elements.
That would be a logical - if challenging - thing to try. It's right along the lines I've been considering.
I'm not sure that you get different points from a link (apart from the balance between outbound & inbound links), but maybe I'm just confused about the question.
I'm not sure this is relevant, but one thing I've noticed is that when you get too many links from the same domain, only a few links count and all the other links are dropped from the PR calculation.
I would confirm that in a way: Having lots of links from one domain and its detailpages (portal linking to you from index-page and x-thousand detail-pages by a layout-footer or inclusion-file)does not bring the same benefit know than before the last algo-shift.
I have a little different interpretation though: IMO it is not the inheritance of the simple PR that is cut off but more the relevance for the specific keyword in the anchor-texts of these links. Don't know if I could express it clearly. ;-) This could be corresponding to the difference between pagerank-gift and ranking-gift Vitaplease just made.
I am seeing sites with high PR (7) (obviously they have their high PR only from one or two portals with multible inboundlinks from detailpages) but they aren't ranking well for their keywords whereas pages with PR5 with most of their inbound-links from different domains appear in front of them on the SERP.
As far as the link-prominence discussion is concern:
- Links having less prominence if they showup on the bottom
- Links having more prominence if keywords surround them
>> interesting points
Why should google not move a step further?
=Ranking a page by it's complete benefit for the user. So inbound+outbound is relevant.
Imagine you have a page that receives links from pages with the same topic deals with this topic and even links out to pages on other domains that are dealing with the same topic. All involved anchor-texts contain the major keywords of this topic.
So this page is basically recommended by "specialists" of this topic gives further information about this topic PLUS it links out to provide even off-domain ressources about this topic to the user.
Isn't that the page with the highest benefit for the user?
Hence, isn't that the page that deserves the highest ranking?
I believe google has already moved into this direction or is beginning to.
I'm not sure this is relevant, but one thing I've noticed is that when you get too many links from the same domain, only a few links count and all the other links are dropped from the PR calculation.
And does it apply if the links from the domain all go to separate pages? There is a link from their red widget page to my red widget page, a link from their blue widget page to my blue widget page etc. I only have one link to their index page.