Links-Do they have too much weight?

[deepsand]

Originally Posted by Peter (IMC)

There isn't really a no vote in PR, but I don't know if you have noticed. People don't link to sites if they don't like them. In fact, when a site is considered really bad, and they want to show them anyway, they just write down the domain name, but make sure it is not a link.

Are you saying that you've never seen an OBL whose purpose was to say, for example, "This is a steaming heap of BS?" The web is awash in OBLs whose purpose is to denigrate rather than compliment.

[Peter (IMC)]

imagine a site has 100 backlinks.

Your logic is:

1 + 1 + 1 + 1 ..... + (N) = 100

Google's logic is more like:

for keyword xxxxxxx:

1 + 0 + 0 + 1 + 1 + 0 + 0 .... (N) = 32


That's the anchor texts making some links not count.

It's more complicated than this of course, but in this simple visualization it should be clear.

[janeth]

Originally Posted by deepsand

But, that weight has nothing to do with it being a "yes" or "no" vote.

If you read my earlier post referred to, then you should now understand that the PR algorithm presumed that, as is the case with academic works, citations would be to other works which supplemented and/or complemented the citing work; i.e. that, in general, such citations would be "yes" votes." In that case, a sum total of votes was equal to the no. of "yes" votes.

On today's web, OBLs may be "yes," "no" or "neutral" votes. Here, the sum total is an absolute total only. I.e., in PR, 1 + (-1) = 2, rather than equalling zero.

Just because you wrote it I am suppose to forget what I know, what I've seen and what I think. Interesting lol

The weight has nothing to do with it being yes or no, correct. But there is weight.

[Peter (IMC)]

Originally Posted by deepsand

Are you saying that you've never seen an OBL whose purpose was to say, for example, "This is a steaming heap of BS?" The web is awash in OBLs whose purpose is to denigrate rather than compliment.

Never seen that one specifically but Of course those types of links exist. But that's not common,. and besides, that link at best helps to rank high for:

steaming heap,
BS,
heap of BS,
steaming BS,
...
...

well you get the point I hope.

That link won't be helpful for the real keywords of the site. These links don't have to have a negative impact on rankings, all they have to be is not counted.

[janeth]

Originally Posted by Peter (IMC)

Never seen that one specifically

lol

Originally Posted by Peter (IMC)

That link won't be helpful for the real keywords of the site. These links don't have to have a negative impact on rankings, all they have to be is not counted.

It will pass pr on to the page that it is linking to giving a boost to the pr for the pages that site links to as well as helping to rank the content on that page higher.

It would be a helpful link.

[deepsand]

Originally Posted by Peter (IMC)

imagine a site has 100 backlinks.

Your logic is:

1 + 1 + 1 + 1 ..... + (N) = 100

Google's logic is more like:

for keyword xxxxxxx:

1 + 0 + 0 + 1 + 1 + 0 + 0 .... (N) = 32


That's the anchor texts making some links not count.

It's more complicated than this of course, but in this simple visualization it should be clear.

You're addressing weight, while I'm talking about the semantical value of that which is weighted.

[deepsand]

Originally Posted by janeth

Just because you wrote it I am suppose to forget what I know, what I've seen and what I think. Interesting lol

Your meaning here is unknown.

Originally Posted by janeth

The weight has nothing to do with it being yes or no, correct. But there is weight.

Correct. My point is that PR does not measure the semantical value of that which is being weighted, owing to the fact that, as already stated, the PR algorithm presumes a positive value of 1 prior to weighting.

[deepsand]

Originally Posted by Peter (IMC)

Never seen that one specifically but Of course those types of links exist. But that's not common,. and besides, that link at best helps to rank high for:

steaming heap,
BS,
heap of BS,
steaming BS,
...
...

well you get the point I hope.

"Steaming heap of BS" was a description of the links grantor's opinion of the linked-to content, its author and/or its publisher, not anchor text.

Originally Posted by Peter (IMC)

That link won't be helpful for the real keywords of the site. These links don't have to have a negative impact on rankings, all they have to be is not counted.

If not "nofollowed," not counted how?

[deepsand]

Originally Posted by janeth

lol



It will pass pr on to the page that it is linking to giving a boost to the pr for the pages that site links to as well as helping to rank the content on that page higher.

It would be a helpful link.

Precisely my point! This, even though the link grantor holds the linked-to content, its author and/or publisher in disfavor.

[Peter (IMC)]

Originally Posted by janeth

lol



It will pass pr on to the page that it is linking to giving a boost to the pr for the pages that site links to as well as helping to rank the content on that page higher.

It would be a helpful link.

Precisely my point! This, even though the link grantor holds the linked-to content, its author and/or publisher in disfavor.

Just toolbar PR, but for searches it doesn't do much because of the anchor texts that are unrelated.

Anchor text is everything!

[janeth]

Originally Posted by Peter (IMC)

Just toolbar PR, but for searches it doesn't do much because of the anchor texts that are unrelated.

Anchor text is everything!

Do you have any proof that unrelated anchor text does not pass pr on to the site?

[Peter (IMC)]

They do pass PR, but only for the Toolbar. There is a total amount of PR of course. But when the search is done and the results are ranked, not all the links are counted, and with that not all PR is counted. So PR for Dell for the keyword computer is huge. But for the keyword "rembrand painting" it is 0. Even if Dell would put that phrase in a page, they still wouldn't rank high for it.

If anchor text wouldn't be used like this, all results would always be high PR sites.

[janeth]

If I have a thousand links to a page about marketing and all those links are for website design Google will still follow those links and index my page. Then if I link from the marketing site to another marketing site I will pass the pr from all those links pointing to the first site and give the second site a good strong link.

[Old Welsh Guy]

Originally Posted by AVC

OWG, your friends with the hot website will have to watch spammers on their forum too, just like any other webmaster has to watch, if spammers start dumping links on their blogs and forums pointing to bad sites, Google will not rank their pages, just like they will not rank others that allow the same to go on.

So if you want organic search traffic FROM GOOGLE, you have to meet GOOGLE GUIDELINES, it is as simple as that.

no no no no & NO! Google guidelines don't matter. if they like the site they are linking to they allow the link to stay, if the site is not up to THEIR standards, then the site link goes. It has NOTHING to do with google.

Google guidelines are quality guidelines, the site I mention was built BEFORE google existed ! So how could it POSSIBLY have been built to satisfy Google?

Search engines wnat the same as people. good sites, with pleanty of good information delivered in digestible chunks, that is easy to find from anywhere within the site. Google guidelines have NOTHING to do with the quality of the site. Outbound linking is decided PURELY on one question " does linking to this site enhance my user experience?". period!

You summed up something in your statement though 'they will have to watch as spammers will start dumping links" EXACTLY! and watch they do, and edit they do. which is probably why it ranks so well and has done from day 1.

Build sites for people and by and large the search engines will love em as well.

[Old Welsh Guy]

Originally Posted by Peter (IMC)

They do pass PR, but only for the Toolbar. There is a total amount of PR of course. But when the search is done and the results are ranked, not all the links are counted, and with that not all PR is counted. So PR for Dell for the keyword computer is huge. But for the keyword "rembrand painting" it is 0. Even if Dell would put that phrase in a page, they still wouldn't rank high for it.

If anchor text wouldn't be used like this, all results would always be high PR sites.

peter, I normally agree with you on most stuff, but not on this one. PR is a pure numeric value, buand forms the 'importance' element of the ranking element, while semantic is relevance and is tex based which has absolutely nothing to do with PR. PR flows fbetween links based on a few eleents, whereas semantic relevance is a whole different ballgame. AFAIK, I have seen absolutely no proof that page relevance affects PR. that would make for one hell of a lot of data processing overheads. PR = numbers = value. Relevance = Alphabetical (converted to numbers)

[AVC]

OWG, and spammers dumping links that are followed are against Google guidelines and if not taken care of will effect your organic rankings.

[Old Welsh Guy]

Originally Posted by AVC

OWG, and spammers dumping links that are followed are against Google guidelines and if not taken care of will effect your organic rankings.

Are you reading what I am posting here?

I will repeat it. There are NO SPAM LINKS ON THEIR SITE AS THEY ARE MONITORED CONSTANTLY

There are ABSOLUTELY no outbound unrelated links. Spam is only an issue if it is not dealt with, and no following high quality links is not best practice in my experience.

[AVC]

I agree, but you did mention that they have users that post a lot on a hot forum !!!

[Peter (IMC)]

Originally Posted by Old Welsh Guy

peter, I normally agree with you on most stuff, but not on this one. PR is a pure numeric value, buand forms the 'importance' element of the ranking element, while semantic is relevance and is tex based which has absolutely nothing to do with PR. PR flows fbetween links based on a few eleents, whereas semantic relevance is a whole different ballgame. AFAIK, I have seen absolutely no proof that page relevance affects PR. that would make for one hell of a lot of data processing overheads. PR = numbers = value. Relevance = Alphabetical (converted to numbers)

Perhaps we´re not understanding eachother. To me PR is also a purely numeric value. Actually it is the sum of a whole bunch of small PR (transfered) values. Each link passes a certain amount of PR to the total.

All I'm saying is that for each search PR is recalculated in such a way that only the PR of links that are counted for that search (i.e. in a simple way, those links with the keyword in the anchor) is used.

That doesn't take that much processing power as it just needs to make a quick comparison. There are no iterative processes in play here. For each link the amount of transfered PR can be held in the database, so it's just a matter of adding up some numbers. (Googlebot or some other algorithm working with data from Google bot would update the amount of transfered PR every now and then.)

It's not so difficult to say: This site has a PR 4,... the query is widgets. 20% of all backlinks of this page have the keyword widgets in the anchor text. Add up the transfered PR of each of those links and there we have the new PR value...... oh that's interesting, for this query we use PR 1 for this page.

Remember they say in their own site they use advanced keyword matching techniques together with PR.

[kgun]

Originally Posted by deepsand

... their flagship product is so very much a proprietary one, one whose inner workings and hidden mechanisms are as carefully guarded as are Crown Jewels!

My bolding.

Irene, note that you are talking about The Crown Jewels of Her Majesty. I am not quite convinced.

[deepsand]

Originally Posted by Peter (IMC)

Perhaps we´re not understanding eachother. To me PR is also a purely numeric value. Actually it is the sum of a whole bunch of small PR (transfered) values. Each link passes a certain amount of PR to the total.

All I'm saying is that for each search PR is recalculated in such a way that only the PR of links that are counted for that search (i.e. in a simple way, those links with the keyword in the anchor) is used.

That doesn't take that much processing power as it just needs to make a quick comparison. There are no iterative processes in play here. For each link the amount of transfered PR can be held in the database, so it's just a matter of adding up some numbers. (Googlebot or some other algorithm working with data from Google bot would update the amount of transfered PR every now and then.)

It's not so difficult to say: This site has a PR 4,... the query is widgets. 20% of all backlinks of this page have the keyword widgets in the anchor text. Add up the transfered PR of each of those links and there we have the new PR value...... oh that's interesting, for this query we use PR 1 for this page.

Remember they say in their own site they use advanced keyword matching techniques together with PR.

SERP is calculated at time of search, by the search engine. Link information is captured at time of spidering, and becomes part of the data handled by the indexing engines, whose results are used by the search engine as part of the SERP algorithm. PR, which is indeed a reiterative calculation, is based on the indexed data, as opposed to real time data gathered at time of search.

If I can relocate it, I'll pass along a link to a Google engineering paper that describes the indexing process; while it's quite a tome, it is most informative.

[deepsand]

Originally Posted by kgun

My bolding.

Irene, note that you are talking about The Crown Jewels of Her Majesty. I am not quite convinced.

Well, look-but-don't-touch seems to be the rule for both.

[kgun]

Originally Posted by deepsand

SERP is calculated at time of search, by the search engine. Link information is captured at time of spidering, and becomes part of the data handled by the indexing engines, whose results are used by the search engine as part of the SERP algorithm. PR, which is indeed a reiterative calculation, is based on the indexed data, as opposed to real time data gathered at time of search.

If I can relocate it, I'll pass along a link to a Google engineering paper that describes the indexing process; while it's quite a tome, it is most informative.

Google's "Crown Jewels" can not be so carefully guarded after all

[deepsand]

Originally Posted by kgun

Google's "Crown Jewels" can not be so carefully guarded after all

Here, I believe, the real jewel, the one that is of true import, would be an engineer's description of the SERP algo.; PR is fool's gold compared to that.

[kgun]

Originally Posted by deepsand

Here, I believe, the real jewel, the one that is of true import, would be an engineer's description of the SERP algo.; PR is fool's gold compared to that.

Agree to that.

[Peter (IMC)]

Originally Posted by deepsand

SERP is calculated at time of search, by the search engine. Link information is captured at time of spidering, and becomes part of the data handled by the indexing engines, whose results are used by the search engine as part of the SERP algorithm. PR, which is indeed a reiterative calculation, is based on the indexed data, as opposed to real time data gathered at time of search.

If I can relocate it, I'll pass along a link to a Google engineering paper that describes the indexing process; while it's quite a tome, it is most informative.

That paper you´re refering to is describing the Pagerank algorithm. The pagerank algorithm is the base of all algorithms of Google. There is a lot of stuff you can do once the values are calculated. Once you know how much PR is transfered in each link, you can choose which links to use and calculate a query specific PR at the time of the search. I don't really understand why this is so difficult to understand.

Keep in mind that if full PR would be used in every search, high PR sites would always beat low PR sites. You could argue that on-page factors reduce the effect of PR, but if that would be true, you would still be able to rank for a keyword that only exists on the page. That's not the case (unless there are no sites with that specific keywords in the anchor of its backlinks). You just can't get a high ranking without the keywords in the anchor texts of at least some backlinks. That fact together with the statement that PR is at the heart of all Google algorithms, tells you that anchor texts are related to PR.

[janeth]

Originally Posted by Peter (IMC)

Keep in mind that if full PR would be used in every search, high PR sites would always beat low PR sites. You could argue that on-page factors reduce the effect of PR, but if that would be true, you would still be able to rank for a keyword that only exists on the page. That's not the case (unless there are no sites with that specific keywords in the anchor of its backlinks). You just can't get a high ranking without the keywords in the anchor texts of at least some backlinks. That fact together with the statement that PR is at the heart of all Google algorithms, tells you that anchor texts are related to PR.

When that same pages pases pr to another site or another page all the pr for that page is counted.

[deepsand]

Originally Posted by Peter (IMC)

That paper you´re refering to is describing the Pagerank algorithm.

The paper to which I refer is a technical one, written by 1 or 2 of Google's computer engineers, which describes in detail the actual indexing process itself.

Originally Posted by Peter (IMC)

The pagerank algorithm is the base of all algorithms of Google.

The values generated by the PR algorithm are but one factor included in the SERP algorithm

Originally Posted by Peter (IMC)

Once you know how much PR is transfered in each link, you can choose which links to use and calculate a query specific PR at the time of the search.

I'm not sure precisely what you mean here. The SERP algorithm doesn't select which links are to be used, but rather which of the pages already indexed should be used, based on the correlation between the search query string and the keyword data indexed for each page.

[Peter (IMC)]

Originally Posted by deepsand

The paper to which I refer is a technical one, written by 1 or 2 of Google's computer engineers, which describes in detail the actual indexing process itself.


The values generated by the PR algorithm are but one factor included in the SERP algorithm


I'm not sure precisely what you mean here. The SERP algorithm doesn't select which links are to be used, but rather which of the pages already indexed should be used, based on the correlation between the search query string and the keyword data indexed for each page.

I am sure that you don't know what I mean, and I'm sorry, but I don't know in how many more ways I can explain the same thing. And it's even possible I'm completely wrong, though I, myself, am pretty sure I'm not that far of. Though a Google engineer that really knows what's happening, might have a big smile on his face when reading my stuff...

[deepsand]

Originally Posted by Peter (IMC)

I am sure that you don't know what I mean, and I'm sorry, but I don't know in how many more ways I can explain the same thing. And it's even possible I'm completely wrong, though I, myself, am pretty sure I'm not that far of. Though a Google engineer that really knows what's happening, might have a big smile on his face when reading my stuff...

I wouldn't be surprised were it the case that a lot of Google's employees are laughing at the lot of us.

As for the engineering paper, not being able to locate it is beginning to annoy me, as I had thought that I'd either downloaded the PDF file, or bookmarked it somewhere. Given that the latter are rather well organized, and it's not there, that leaves only the former. I suspect that it had a cryptic file name by default & that I neglected to rename it to something recognizable and informative, such that it's effectively lost amidst 500 Gig of HD space.

[Peter (IMC)]

The Anatomy of a Search Engine

[deepsand]

Speak of coincidences.

While searching online for the paper in question, I came across what appears to be the same document as you linked to, but in PDF format, at http://infolab.stanford.edu/pub/papers/google.pdf .

I've located those portions which serve to demonstrate what I was saying re. PR being independent of the search query.

Section 4.1, "Google Architecture Overview," along with the related Figure 1, "High Level Google Architecture, describes and depicts how PR is calculated from the results of the Indexing Function, with the end result being that "The links database is used to compute PageRanks for all the documents."

Then, in Section 4.5.1, "The Ranking System," the end of the 1st paragraph states "Finally, the IR score is combined with PageRank to give a final rank to the document."

Of course, absent is any hint of how Page Rank and the Intermediate Rank are "combined" to yield SERP.

[kgun]

Originally Posted by deepsand

Speak of coincidences.

While searching online for the paper in question, I came across what appears to be the same document as you linked to, but in PDF format, at http://infolab.stanford.edu/pub/papers/google.pdf .

If you had looked in my link collection, you would have found both on the first page of the second link in my signature. You find the PDF doucument in this way.

Click "search" in the upper right corner to the left of the "Digg" bookmark link. (CTRL + F).

Advanced topics

Scroll down to PDF documents

You may find some other links in the "Advanced topics" category that may interest you.

[Peter (IMC)]

Originally Posted by deepsand

Speak of coincidences.

While searching online for the paper in question, I came across what appears to be the same document as you linked to, but in PDF format, at http://infolab.stanford.edu/pub/papers/google.pdf .

I've located those portions which serve to demonstrate what I was saying re. PR being independent of the search query.

Section 4.1, "Google Architecture Overview," along with the related Figure 1, "High Level Google Architecture, describes and depicts how PR is calculated from the results of the Indexing Function, with the end result being that "The links database is used to compute PageRanks for all the documents."

Then, in Section 4.5.1, "The Ranking System," the end of the 1st paragraph states "Finally, the IR score is combined with PageRank to give a final rank to the document."

Of course, absent is any hint of how Page Rank and the Intermediate Rank are "combined" to yield SERP.

That document is the original document that Larry and Sergey wrote for their university project. They´re not just 2 engineers, they´re the founders of Google,.

Since that they started their search engine, Google has advanced a lot. That basic PR algorithm is still the base of all algorithms (as Google says in their site) but you can not assume that the whole search engine still works exactly as described in that document. What I'm describing is similar to Topical Sensitive PageRank (TSPR) but a bit more query specific. Something that, in my opinion, is pretty easy to do logically, though it may have required a massive change in the infrastructure and indexing processes. (Big Daddy comes in mind)

Try to think in terms of seperating PR per page and transfered PR per link.

[deepsand]

Originally Posted by kgun

If you had looked in my link collection, you would have found both on the first page of the second link in my signature. You find the PDF doucument in this way.

Click "search" in the upper right corner to the left of the "Digg" bookmark link. (CTRL + F).

Advanced topics

Scroll down to PDF documents

You may find some other links in the "Advanced topics" category that may interest you.

I've a browser Favorites/Bookmark Folder named "_To be acted on;" the underscore in the Folder Name forces it and those similarly named to be at the top when sorted by name.

Reviewing your links is on my "to do list," but hasn't floated to the top yet.

"When you're up to your ass in alligators it's difficult to remember that your primary objective was to drain the swamp." - anonymous

[Webnauts]

Originally Posted by Peter (IMC)

Anchor text is everything!

If that is true, what is if your search term is not clearly visible on your site and the title and description tags are not correctly marked?

[Webnauts]

Originally Posted by Peter (IMC)

The Anatomy of a Search Engine

And what do they mean with this sentence?
"Using anchor text efficiently is technically difficult because of the large amounts of data which must be processed."

[full house]

yes there are, most of website owners focus on link building. Link building is one of the best way to get good ranking.

[deepsand]

Originally Posted by full house

yes there are, ...

"yes there are" what?

Originally Posted by full house

most of website owners focus on link building.

Do they? What, if any, proof of such exists? It is my observation that those who do so are in the minority.

[Webnauts]

Originally Posted by deepsand

"yes there are" what?

Do they? What, if any, proof of such exists? It is my observation that those who do so are in the minority.

I am the best example brother.

[full house]

why you?

[Webnauts]

Originally Posted by full house

why you?

I do not focus in the link building.

[deepsand]

[off topic]

Is anyone else not getting their automatic e-mail alerts re. subscription activity?

[/off topic]

[full house]

you just do your SEO in content? I know that content is the king but you still need to build links to make sure of the success.

[deepsand]

Originally Posted by full house

you just do your SEO in content? I know that content is the king but you still need to build links to make sure of the success.

Do you?

How then to explain the financial success of those who have ignored link building?

IBLs are not the only means of successfully marketing ones site(s). And, they are not necessarily the most cost effective means of doing so.

[janeth]

Originally Posted by deepsand

Do you?

How then to explain the financial success of those who have ignored link building?

IBLs are not the only means of successfully marketing ones site(s). And, they are not necessarily the most cost effective means of doing so.

Without inbound links you can't even get a site indexed.

[janeth]

Originally Posted by deepsand

Do they? What, if any, proof of such exists? It is my observation that those who do so are in the minority.

Yes and it is that minority that ranks for the competitive keywords.

[Terry Van Horne]

Originally Posted by janeth

Without inbound links you can't even get a site indexed.

Agreed, the difference is how you go about developing links. Some think the only way to do it is beg... others, myselff included, prefer to publish content. I do press/social releases to launch, find the key sites in the niche and try to devlop an ongoing relationship either through article syndication, write a guest article etc. Any moron can email a bunch of people (that's what I envision when I hear the phrase "link building") and annoy half the people in the niche while you are at it. No thanks... that's a suckers bet with no guarantees of longevity or that the SE's won't change the rules... like they did with IBLs being unable to hurt you.

[janeth]

Originally Posted by Terry Van Horne

Agreed, the difference is how you go about developing links. Some think the only way to do it is beg... others, myselff included, prefer to publish content. I do press/social releases to launch, find the key sites in the niche and try to devlop an ongoing relationship either through article syndication, write a guest article etc. Any moron can email a bunch of people (that's what I envision when I hear the phrase "link building") and annoy half the people in the niche while you are at it. No thanks... that's a suckers bet with no guarantees of longevity or that the SE's won't change the rules... like they did with IBLs being unable to hurt you.

I agree 100%, I do not do link exchages and I have three full time writers. There are more ways to get one way links than most people would imagine.

[deepsand]

Originally Posted by janeth

Without inbound links you can't even get a site indexed.

False. For becoming aware of new Domain Names SEs have a variety of means which pre-date the existence of the IBL craze.