Are the "linking" factors the core of GoogleRank?

What was the helpful response, huh? The guy pointed to the best article on PageRank which clearly shows that there is a PageRank leak. That is a fact. Denying a simple fact will be much more confusing, than simply stating the obvious truth about the leak but explaining it does not matter in the real world which is the 100% truth. And the "there's no leak" thing is just nonsense.

I am all for the truth.

 

As you wish... I'm not going to beat this to death. We obviously have different views. C'est la vie.

 

Stay calm!
Different people have different views.

As far as what I know, here's my 2 cents on PageRank (I may be wrong since my understanding is based on articles on some SEO experts).

PageRank is relaitve in nature. That means a rise in A's website means a drop in B's website, or something similar (eg rise in A = drop in B & C). It is somewhat like the system of foreign exchange rates. An apreciation in A's currency means the depreciation of B's.

Due to the above nature, the sum of PageRank(PR) must be always constant. Now let's assume the sum of all PageRank(PR) always has a value of 100, for the sake of easy explanation.

Situation 1
Assumption:
- There are 2 websites only
- Website A's PR is 51 as a start; while Website B's PR, 49. The total is 100 (51+49).
- Now only the linking factor is in active to affect the PageRank.

Now web-A votes web-B by linking to B.
This will increases the PR in B. OK, let's assume it rises by 2.
After the change, it is:
Web-A PR: 51
Web-B PR: 51
--------------
Total PR:..102 (Hmm...?!)

Wait! As we all know the sum of "PR" is constant (ie always 100). So where does the "2" point go away from? It must be Web-A since it is the only web in this situation.

So the correct PRs are:
After the change, it is:
Web-A PR: 49
Web-B PR: 51
--------------
Total PR:..100

But don't ever attempt to make any conclusion now! You may get it wrong or misinterpret the above result. Let's read on!!

In reality, linking is only 1 factor. Google claims it has more than 100 factors when calculating the PageRank. Also it is not going to have 2 websites only in the world wide web. Just one "vote" is going to have negligible effect on the PageRank.

In reality, you may find "website 1" links to no website; but its PR is even lower than "website 2" which links to many websites. However the result means little since we haven't tried to control any relevant variables in the observation.

That's why we need to carry out "variable-controlled study". To see if a variable has an effect on the subject, we need keep "other relevant variables" constant in both subjects. Then we let the tested variable to go on either way (eg subject XX has that variable in effect; subject YY doesn't). Then we observe and see if it leads to change.

The nature will usually deceive us. The above technique is what scientists ususally do when they do their research (eg examine if smoking causes lung cancer). Real world examples usually have a very low reliability in proving or disproving something.

 

As far as ONLY your ranking in Google is concerned, you should not "vote" to others. What you did is to boost others' ranking. When they get higher and even higher than you, they will replace your position. Thus voting others has no benefit to you: you just help others, NOT you.

So what will you do? Not going to link to others anymore. No, at least not to me. It is just as selfish as in the case where you will never donate $$$ to help others.

We should be good webmasters. We should never use dirty techniques to boost our websites. To me it is equal to a salesperson who exaggerate a product to deceiving poeple into buying it. Never a good idea in my opinion.

Quality and content of a website is our prime concern. I will not sarcifice content/quality in echange for higher ranking positions!

PS: My valueless 2 cents ONLY! Don't treat it seriously.

 

Wai_wai,

1. Your example of 2 sites and PR is wrong, because the classic PR is not host-to-host, but page-to-page (it operates on a graph of pages, not sites). Other than that, the sum of all PR is the sum of the number of pages. Every link results in some pages gaining and some losing PR. Ian Roger's paper is great on explaining it.

2. Forget about "variable-controlled" study. All SEO tests I've seen are meaningless. I'll give just one reason: based on the competitiveness of a query, Google may use different ranking algorithms. So, a SEO test based on keywords like "alkdasldjalkdjaslkjdaslk" won't mean sh** about competitive queries (where actually on-page factors except the title tag may be ignored).

3. Matt Cutts (a guy that works at Google) has stated that Google detects sites that use internal linking that artificially boosts PageRank (for example by not linking to other sites, generating a lot of dummy pages to generate PR etc.)

4. If you don't vote for others, they won't vote for you.

5. If you don't vote for others or use PR hoarding internal linking, Google can detect it easily and penalize your site.

Anyway, you are working here on very old PageRank information, that was used years ago. Google has already found ways to punish PR hoarding schemes and unnatural linking strats.

 

Oh! I see.
This is used for simplicity purpose. That's why I keep saying web-A, web-B. Forget to clarify it, sorry!

I have one question.
You know, there's Google Toolbar available which can show GoogleRank (0-10, right?)

Say when I go to the homepage of web-A, does the GoogleRank shows is the rank of that homepage ONLY, or the average rank of all pages in web-A (simply it means the rank of web-A on average)?
I tihnk it should be the former, right?

Yes, you are right.
"Variable-controlled" study only focuses on rare (sets of) keywords.
So it can't really reflect the competitivity nature of search psotions.

Now here's my 2 cents about "Variable-controlled" study:
At least it is better than nothing, or better than just taking a real-world example (and claiming something).

Sometimes these kinds of tests have some values. Eg: One tries to build different pages with different contents. The results shows Google like a page to be complete (eg have titles; metadata: Keyword, Description etc.; headings). Then at least we know these things help to boost your PageRank although we can't know "the size of the effect".

No test is perfect and we have to live with it. And we can still learn something form "variable-controlled" study, can't we?

It seems you are saying if there's no linking out, Google will punish you.
That's a good sign.
So evil webmasters can't even gain on these dirty tricks.
But I think they may workaround by adding only a few links, and bypass the punishment.

Old PageRank info... True. That's why I ask this question, to see if some major changes happen.
By the way, some changes I realise are:
- links (votes) are counted only if the page has some PageRank (fame). It seems to be at least 4. Is it correct?
- Google no longer uses the "classic" way of sorting pages, ie sort pages by PageRank ONLY. There are other factors which play into effect (eg HostRank, SiteRank, IR Value and so on.

Any good and updated articles/websites etc. which talks about Google ranking and SEO?

 

No. That's incorrect.

That was never true. PageRank, according to Google, is "only one of more than 100 factors" determining ranking.

Yes.

No.

www.seobook.com

 

Then what is correct?

It seems PageRank is used to be more important.
Now it is less important.

 

I am sorry but unless you work for google and have the actual formula for pagerank you cannot say this is a fact. Every decent search engine marketer will tell you that pagerank is a myth and to stop wasting time worrying about it, proper linkage is so, so much more important. How many people outside the world of SEO gives a s**t (or even knows) about pagerank?

 

All links are counted, unless they are "bad neighborhood" type links or subject to some other Google filter. It doesn't matter whether those links are PR0.0000001 or PR10. The higher PR ones have a greater impact but all "count". If you use a Google link: query, Google will show you only a small and seemingly quasi-random selection of the backlinks it knows about. That isn't based on PR either.

That may be and probably is true. My point was that it was NEVER the only factor.

 

It is what I heard recently from one website.
The main idea is:
- to prevent link farms and bad linking systems, links which have low PageRank (I forget the value exactly, but it seems to be lower than 4) will not be counted at all.
Anyway it must be wrong after what you said. Thanks for your clarification.

Yes, it is true.
See this: http://www.searchenginejournal.com/index.php?p=2133

Because at one time, Google’s ranking algorithms were based in large part on the PageRank calculations. Too soon, however, webmasters realized that in order to get high rankings, all you needed was a bunch of links. Google caught on however, and adjusted the algorithms to ensure that only relevant links were counted.​

 

Yes, outsiders will never know exactly. So what we try is to learn as much as possible.

Sometimes Google will publish some articles about its ranking algorithms. This article is based on Google papers to explain how PageRank works. Thus I think it is much more reliable at that time.

However the article is a bit old. Algorithms have been changed. Although the main concept of PageRank and links still applies, some details may be outdated and so inaccurate.

 

thanx for sharing this ebook.

I would surely get my hands on this book.