|Question about Links, Link Farms, Google Rank|
Please forgive me if I overstep- I am not a webmaster, only a small company administrator trying to understand the world of SEO, and Google. I have read a lot about how Google works, from Google's site and here on this very informative forum.
But I still don't quite understand some things. Is it true that Google will consider a site more "popular" if there are many links to it, regardless of the "quality" of those links? For instance, many competitive sites have pages of "links" that are wholly unrelated to the search category, yet as a result, their rank would be much higher than ours, which does not participate in reciprocal links? Maybe it is the term "popular" that throws me. So in reality the actual sites may have no more real value to the visitor than our website as far as content goes, but it will be ranked higher because they have many category unrelated links? Is this an accurate description of how Google works?
Thanks in advance for clarifying this for me.
Google Pagerank calculations are comprised of a differential equation which must be iterated to solution.
Basically, the equation takes into account a few factors, the most notable being the quality and number of incoming links (as well as the quantity of links outgoing from those pages linking in) and the quantity of outgoing links.
This main linking paradigm comprises the modifier in pagerank. The set-structure is made of up the number of actual (physical) pages within your site and how you link them internally.
What you need to determine is who is actually linking to your competitors, how they then link within their own site, and then, who they link to.
As a rule, a small number of high PR ranks are more beneficial than a large number of low PR ranks; in fact, based on the differential equation, a link from a PR site which is low enough can REDUCE the over pagerank of the site it links to.
In order to successfully increase your own page rank, focus on the following, in this order: (1) does your site contain an adequate number of content driven pages; (2) are these pages linked properly between each other to maximize internal utility (and thus Page Rank?); (3) don't try to get too many links from link farms--focus on one or two HIGH PR ranks, and give it a month.
Also, something most people don't realize--google gives a PR preference modifier to pages who UPDATE on a continual (i.e., daily/weekly, etc.) basis. This preference translates into increased PR as well as Search Engine search term preferences.
The two sites for which I'm webmaster, (PR8 and PR7, respectively) have updates done on a consistant basis, which tends to enforce, if not pad, pagerank.
[edited by: esaslo at 7:15 pm (utc) on Mar. 28, 2003]
[edited by: Marcia at 7:57 pm (utc) on Mar. 28, 2003]
[edit reason] sorry, no off-site solicitation please [/edit]
Psmith0000: yes, pagerank is based on all links, even ones which might appear to be unrelated to the site's content. "Popularity" is one way of looking at it; another is to say that a link to a website is kind of like a "vote" for the quality of the site's quality/value/usefulness.
However, pagerank is just one of many elements that Google considers when listing the actual search results.
>> a link from a PR site which is low enough can REDUCE the over pagerank of the site it links to
Can you explain how this can happen? I have never heard this.
>> google gives a PR preference modifier to pages who UPDATE on a continual (i.e., daily/weekly, etc.) basis. This preference translates into increased PR...
Never heard this either, and I don't believe that freshness translates to higher PR.
>>the "quality" of those links?
Accepted wisdom appears to be that the text in the link is , in part, a measure of the quality to google.
In other words if your site is about "widgets" and the link to your site on the other site mentions "widgets", then Google gives it more credence than if the link ignores mentioning widgets and only refers to "sprockets" or some other word(s) that do not in fact occur on your site.
A link from a PR site which is low enough can REDUCE the over pagerank of the site it links to
Can you explain how this can happen? I have never heard this.
This is simply a matter of the differential equation (math-guy, here); let me demonstrate:
Take the most simplified form of the pagerank equation:
PR(A)=(1-d) * d(...) extend to PR(A)=PR(A) + (1-d) + d(...)
PR(A) = (1-d) * (P(A)/(P(B)))... + ... (P(An)/(P(Bn)))
Where the d is the modifying factor.
So, let us say that P(A) is the page's pagerank and P(B) is the number of outgoing links for any given page P.
If page P has a pagerank of 2 and has 50 outgoing links that's (2/50) * d. D is less than one (nominally set at .85); so, do a series on your own: compare the following:
P(A)= (1-.85) * (2/30) + (2/40) + (2/50) ... + p(An)
P(B)= (1-.85) * (8/2) + (7/2) + (6/4) ... + p(Bn)
Now, this is just for instructional purposes; this is the basic representation. Although the google equation might not be exact, this representation is good enough.
You can clearly see how P(B) >> P(A).
If you want to prove it, expand it with the taylor series about (1-d)*P(An)
This comes from three basic facts, the most prevalent of which is my experience logging every single time I'm indexed, the dates and times, on each site.
However, adding pages consistantly is a good force against the overall number of pages on the web. I.e., overall pagerank is based on the total number of pages on the web. Increasing your internal pagecount helps to offset the increase in overall pagecount.
Its really not a "freshness paradigm" as much as it is an ACTIVITY paradigm.
>>regardless of the "quality" of those links?
Google says they look at the quality of the page linking:
Google Technology - Page Rank Explained [google.com]
|Google interprets a link from page A to page B as a vote, by page A, for page B. But, Google looks at more than the sheer volume of votes, or links a page receives; it also analyzes the page that casts the vote. Votes cast by pages that are themselves "important" weigh more heavily and help to make other pages "important." |
That's in explaining PageRank, which is numerical and can be interpreted to mean that the linking page is more "important" because that page itself has higher PR.
Link farms are traditionally groups or pages of links passed around to share and increase link pop for all. Then there's also "domain farms" - what I call Page Rank Ranches - sites set up that may or may not be topically related, but whose primary purpose is to pass the Page Rank around among them.
Higher PR doesn't necessarily mean higher rankings, but it's a good part of marketing sites.
|So in reality the actual sites may have no more real value to the visitor than our website as far as content goes, but it will be ranked higher because they have many category unrelated links? Is this an accurate description of how Google works? |
Yep, it's not insurmountable but it sure can and does happen. Some speculate about emergence of topic or context sensitive Page Rank, which wouldn't automatically eliminate rigging the score since those who know how to set it up often do it within topical areas. It's the other things that go along with it that determine whether there's actually quality, but in itself it's technically not breaking any rules.
|Also, something most people don't realize--google gives a PR preference modifier to pages who UPDATE on a continual (i.e., daily/weekly, etc.) basis. This preference translates into increased PR as well as Search Engine search term preferences. |
Does this mean an update of individual pages (i.e. continually add content to existing, indexed pages) or does it mean continually add new, fresh pages to the web site?
[edited by: Marcia at 8:15 pm (utc) on Mar. 28, 2003]
[edit reason] fixed text formatting [/edit]
Just updates done to the site as a whole, from my experience, not, necessarily, to individual pages. Although I do notice that updates or modifications to my homepage often result in it getting re-indexed very quickly.
<<a link from a PR site which is low enough can REDUCE the over pagerank of the site it links to.>>
I'm no math whiz, but this really, really doesn't make sense and I strongly question the accuracy of the statement.
|A link from a PR site which is low enough can REDUCE the over pagerank of the site it links to |
I believe you have misunderstood the math.
The equation ADDS at least some value for every page that links to the target page. There is no reason to plug in a PageRank of 2 for every page in the calculation, as if every page that links to the target page wuold only be a PR2.
But even if you do that, still each added link will only INCREASE the total PR "vote" for the target page, even if by only a small amount.
You have also misquoted the equation. The damping factor is first subtracted from 1, and then that result is added to the product of the damping factor times the sum of all the individual page calculations:
PR(A) = (1-d) + d(PR(T1)/C(T1) + ... + PR(Tn)/C(Tn))
In a real situation, there will be many pages linking to a given target, and each page will have a different PageRank. It would look more like this:
P(A)= (1-.85) + .85*((2/30) + (4/20) + (3/10) ... + PR(Tn/C(Tn))
You can see that no matter how small the page rank, and no matter how many links on the page, still the value of the entire equation can only increase with each added link. The worst case, a PR0, simply adds nothing. But it surely can't reduce the total value.
[edited by: tedster at 8:48 pm (utc) on Mar. 28, 2003]
|google gives a PR preference modifier to pages who UPDATE on a continual (i.e., daily/weekly, etc.) basis. This preference translates into increased PR as well as Search Engine search term preferences. |
That may be a brand spanking new concept we're hearing, but it certainly isn't a feasible one in view of the fact that PR is accrued by the value of the votes to a page. I fail to see any possible relation between modifying the content of a page and that page's PR, other than an alteration in the number or value of the incoming links, which is a different matter altogether.
A "Fresh" listing can be triggered by addition of a new page to a site if it's linked to from pages that freshbot knows about and visits, but any subsequent alteration of Page Rank causative by changes to pages within the site would strictly be a reflection of the internal PR distribution in the site rather than being affected by the freshness factor.
Changes in PR can happen for any number of reasons, but there can't be any empirical demonstration that "Fresh" affects PR one iota, other than the possibility that a frequently updated page may attract inbound links. But it would be those links that would affect the PR, not the change in the page's content.
Also, any subsequent increase in rankings due to the additon of new pages could certainly happen, but certainly not because of "Freshness." I can think of a dozen factors right off the top of my head of how adding pages and/or modifying pages can help a site's rankings, but being fresh crawled isn't one of them.
A fresh-crawled listing may *appear* to shoot to the top, but that doesn't mean it'll stay there when the real scoring is actually computed.
a link from a PR site which is low enough cannot reduce the pagerank of the site it links to!
Google doesn't give a PR preference modifier to pages who update on a continual basis. Of course, ranking is influenced by updating a page, however, PR is not changed!
By the way:
- (for d=1) is calculating eigen vectors
- (for 0<d<1) is solving a set of linear equations, whichs corresponds to invert a matrix
There is no differential equation.
(Also, iteration is not the only way to solve the set of linear equations. It is just the simplest/fastest way for 3 billion pages. Iteration is just a technique.)
[edited by: doc_z at 9:27 pm (utc) on Mar. 28, 2003]
|You have also misquoted the equation. The damping factor is first subtracted from 1, and then that result is added to the product of the damping factor times the sum of all the individual page calculations: |
PR(A) = (1-d) + d(PR(T1)/C(T1) + ... + PR(Tn)/C(Tn))
When you expand the equation, (take a taylor series expansion, for example), the initial subtraction becomes a delta, which is just added to another gamma. Basically, the inital subtraction doesn't really matter. All it does is set a minimum pagerank saying that the page does, indeed, "exist".
And yes, I forgot the first key set of parenthesis in my equation; however, it really wouldn't make any sense withouth them.
|P(A)= (1-.85) + .85*((2/30) + (4/20) + (3/10) ... + PR(Tn/C(Tn)) |
Distribute "d"; it doesn't really matter if it is .15 or .85, all that matters is that its less than one. This is where our understanding of how the equation works diverges. If you DO distribute "d" among the individual pages, THEN add, you get a decidedly different result from ADDING and then applying D.
Now, from Marcia:
That may be a brand spanking new concept we're hearing, but it certainly isn't a feasible one in view of the fact that PR is accrued by the value of the votes to a page. I fail to see any possible relation between modifying the content of a page and that page's PR, other than an alteration in the number or value of the incoming links.
Extend your thinking of pagerank into a two-way street. Pagerank has a sufficient effect on search engine indexing and ranking. For example, each week I add 50 or so pages (articles) to my non profit website; these are all indexed by google by the following monday. In an manner of speaking, google has created links for these pages of a designated "1" from its own indexing process. This pads the number of actual pages in the website; possible is a pagerank increase through this.
Now, develop a plan to do something like this on a regular basis. Do you ever notice how some sites aren't regularly re-indexed? That is, new pages sometimes take a month to be added to the google index for them? This even happenes on PR6/7 sites without much content change.
However, sites that DO have content development and DO have an intital pagerank see their pages indexed more quickly. I have first hand proof of this, comparing exact changes to a PR7 and a PR4; one gets the changes indexed in a few days, the other, more than a month. (No external linking present on any of the new pages; the PR7 being updated weekly, the PR4, monthly).
In my mind there is a premium placed, however hard it is to focus down upon, on regular development; and this is seen directly through indexing and then, second hand, through pagerank.
>>Distribute "d"; it doesn't really matter if it is .15 or .85, all that matters is that its less than one. This is where our understanding of how the equation works diverges.
To be simplistic, if someone gives me 2 cents it's far less than a dollar. But if I've got $2.00 I'll then have $2.02, not $1.98.
>>>comparing exact changes to a PR7 and a PR4; one gets the changes indexed in a few days, the other, more than a month.
I have a PR5 site with interior pages PR4 that's been getting fresh listings constantly for ages without changes or additions being made at all. The only change has been a minor change now and then to a page that links to the site, possibly some other sites that link to it. That linking site first got freshed after getting linked to from a homepage that gets fresh updates, and then the third site was next in line. Site one gets changed frequently and that's how it started. There seemed to be a chain reaction. The PR has not gone up one iota on any of those 3 sites mentioned, in fact it's dropped some in spite of one them getting a new ODP PR5 listing plus also the Google Directory.
I've been getting freshed on a brand new site every time I upload a new page, and that site is linked to from a page that's constantly freshed - the same page that's linked to the site that never gets changed and still gets fresh listings.
The possibility of increase in PR from adding internal pages has been discussed from a number of vantage points in the past.
Yep, a site can be improved by adding pages, but not demonstrably in any definitive fashion because of getting fresh listings. The fact that additional pages are added and Freshbot summoned isn't, in and of itself, a causative factor for increased PR. In fact, you can add pages to a site, link out to all those new pages from the homepage, and actually hurt PR and rankings some. That's what I believe happened with site #1 above - overall site page PR dropped when a number of new pages were added and linked to from the homepage.
If there's an increase in PR there's a likelihood it's because of additional PR votes being cast from added interior pages back to the homepage and other key pages on the site, depending on how it's done - which is what makes site navigation and internal PR distrubution so important to consider.
Absolutely, adding to a site is helpful, but we can't be conjecturing too much without some kind of reasonable corroboration about why it's helping or we'll end living with misconceptions.
|If you DO distribute "d" among the individual pages, THEN add, you get a decidedly different result from ADDING and then applying D. |
d*(a + b + c) is exactly the same as d*a + d*b + d*c
Wow. Some of this is very difficult for me to understand, but overall, I get it. I think.
Thank you all very much for furthering my understanding of page rank. It still doesn't seem to be a completely uncorruptible algorithm, but I guess we all have to live with it.
Thank you again.