| 5:51 pm on Oct 2, 2003 (gmt 0)|
Currently, PR doesn't depend on the placement of the link.
| 6:20 pm on Oct 2, 2003 (gmt 0)|
no offence, i, too, understand this to have been true in the past. are you sure this is the case today?
| 8:23 pm on Oct 2, 2003 (gmt 0)|
|are you sure this is the case today |
| 9:25 pm on Oct 2, 2003 (gmt 0)|
again, no offence intended, just making sure. thanks a bunch!
| 9:32 pm on Oct 2, 2003 (gmt 0)|
leanweb, while it may not affect PR as such, are you noticing anything different with links at the top of pages that would bring your attention to it?
| 9:49 pm on Oct 2, 2003 (gmt 0)|
nothing to report yet, but i have suspicions that i need to investigate further. i will certainly report here if i find something out.
| 6:39 am on Oct 3, 2003 (gmt 0)|
I doubt very much that Google would consider the links position on a page in any way, shape or form, many link pages are alphabetical. No reason why a site beggining with the letter "A" should get more weight than a site starting with "B".
| 7:17 am on Oct 3, 2003 (gmt 0)|
i respectfully disagree. some properties of a link, such as its location, size (both, font size and length of text, or size of image under anchor) and color would have affect on how probable it is that it will be clicked. which is the premise of PR system.
its a whole another story that its tough to accurately account for these properties, so i suspect google isnt. or wasnt, which i am beginning to notice things that lead me to believe that this may be changing. but then again, it could just be my paranoia, so let me do more investigating before reporting something.
can GG confirm and save me the trouble?
| 7:23 am on Oct 3, 2003 (gmt 0)|
That is exactly the type of question that you should not ask GoogleGuy.
| 7:42 am on Oct 3, 2003 (gmt 0)|
he seems a reasonable guy. PR system is open. so i dont see why he shouldnt disclose this type of stuff. granted he'll prolly qualify that it is subject to change w/o notice.
i am not trying to circumvent the system, just understand it so that i don't do anything to inadvertly hurt my site's ranking, thus my business. thats not unreasonable, is it?
it would be doubly unfair to retaliate (ban my site) for my asking questions here (the url to my site is available in profile). which i _really_ doubt. i side with Einstein on this one, Gods are not malicious ;)
| 7:46 am on Oct 3, 2003 (gmt 0)|
Actually, we don't address to GG, who doesn't get into areas of scoring, never has - for obvious reasons. We'd all drive him daffy for sure, he'd be bombarded. And there's just so much he or anyone could tell.
But here we have something you're bringing up that touches on an important point:
|some properties of a link, such as its location, size (both, font size and length of text, or size of image under anchor) and color would have affect on how probable it is that it will be clicked. which is the premise of PR system. |
PR aside, and it doesn't logically follow from how we've understood it that it would affect PR as such but could affect other considerations. What GoogleGuy has said more than once is that we should just make good sites and are good for our users and they'll do well with Google.
What you pointed out about links being done in a certain way in certain places making them more likely to be clicked is just plain good marketing sense. If there are pages we want emphasized to sell products it makes a lot of sense to place them strategically to increase the chance of getting more clicks.
Darned if I know exactly how they'd do it, but if WE consider pages on our site more important than others, wouldn't that indicate that they'd be more important to searchers looking for what's at the end of those links when they click on them?
Certain kinds of items are selling so fast now people can't keep up with shipping. Would links to those items be buried in size 1 Verdana font in bottom or left navigation, or would a sensible site owner place those links prominently where when someone hits the page they can see it plain as day and where it's easy to click on.
Placement location on a page should indicate importance from a site owners viewpoint, so following through logically if it's proportaionately more relevant it should rank well, and that would have a naturaly consequence of higher click-throughs. We're more likely to click on something that's right in out faces than hidden in the site's template navigation.
Philosophically, the premise of PR and the random surfer concept, even if it's not mathematically counted in, aren't incompatible with how clickable links are or how likely the are to be clicked on.
[edited by: Marcia at 8:04 am (utc) on Oct. 3, 2003]
| 7:52 am on Oct 3, 2003 (gmt 0)|
the location of the link can cause differences, e.g. the ranking of the linking page. However, currently PR doesn't depend on the location of the link.
| 7:55 am on Oct 3, 2003 (gmt 0)|
doc_z, it could make a lot of sense for some links to transmit more PR to pages linked to than others depending on where they are. But I'm not so sure it's possible to tell whether it's being done or not, most of us aren't that sophisticated to be able to tell. We CAN see certain effects of different placement of links, but probably not the numerical PR transmitted.
| 8:01 am on Oct 3, 2003 (gmt 0)|
agree, Marcia. what i think i'm seeing is, some links get "penalized" and seem to not transmit PR value. or transmit a drastically reduced PR. i have not been able to confirm that, tho. not sure if i will ever be able to ;)
| 8:15 am on Oct 3, 2003 (gmt 0)|
I still say that Google would never give more weight, or PR to a link that is higher on page than another. As mentioned above, most sort their links in Alphabetical order and it's makes no sense at all for Google to view sites beginning with "A" as being far more important than site starting with "Z". Sure, sites starting with "A" will get more clicks from the links page, but that's all. Google even sorts its DMOZ directory by Page Rank and NOT alphabetically.
I am happy to be *proven* wrong, but PR application is one of Googles closely guarded secrets.
| 8:41 am on Oct 3, 2003 (gmt 0)|
there is a whitepaper on PR, just google for it. it may not be as detailed as we may wish, but it has all the important info, including the math.
now, your alphabetical point is well taken. but consider links such as credits at the bottom of page. or really pale small font links. you'd think those get clicked less often, no?
and the probability of click is at the root of PR algo, no matter how you tweak it.
| 9:08 am on Oct 3, 2003 (gmt 0)|
of course Google could have changed the PR algorithm. For example, one can think of an algorithm where the transferred PR depends on the size (location, type) of the link. I started a thread about this topic long time ago, but nobody could gave a definitiv answer. Therefore, I started my own research. Currently, I'm seeing no dependency on the location of the link. Of course, Google can change this in the future.
| 9:36 am on Oct 3, 2003 (gmt 0)|
|there is a whitepaper on PR, just google for it. it may not be as detailed as we may wish, but it has all the important info, including the math. |
Sure there is, I have read many of these theories before, I have yet to see an official one from Google though, have you?
|but consider links such as credits at the bottom of page. or really pale small font links. you'd think those get clicked less often, no? |
Yes off course they get clicked less often. The number of times a link is clicked on any specific link page has little or nothing to do with PR though. Google doesn't know how often a link on a link pages get's clicked.
IMO (and it's so logical) If there are 10 links on a PR5 page, sorted A:B and each of these linked pages have all others PR factors the same, Google would not give more weight, or PR to the top one over the bottom one simply because it starts with the letter "A". Think about it, it would make no sense at all.
| 9:55 am on Oct 3, 2003 (gmt 0)|
doc_z seems to've confirmed that what i think google is doing (accounting for ling geography) is not the case. well, so be it, i am paranoid, nothing new.
but, Dave, just fyi, google does have an "official" paper on how PR works. here it is:
this is as official as you going to get. they have few more papers, for example i've seen recently a whitepaper on how their file system works. and there was one on their solution of clustering. google runs off of COTS computers, believe it or not.
all in all they are very open about what they are doing. they sprung off stanford for pete's sake ;)
| 9:55 am on Oct 3, 2003 (gmt 0)|
No, the paper was not released by google. It was released by a couple of graduate students named Larry Page and Sergey Brin. It is almost spooky how a couple of graduate students could figure out how the top search engine would be able to calculate pagerank. And they even had the great precognitive ability to do it before Google even existed.
Yes, it might have changed a little bit, but several Google employees have stated that PageRank is basically the same algorithm that is outlined in the paper.
| 9:59 am on Oct 3, 2003 (gmt 0)|
Dave, there's a difference between a page with a list of outbound links, like a directory or a links page, and a content page on a site, like the homepage that would lead to product pages.
If we were working on a search engine and sat there looking at a web page, what would we look for on that page in order to decide how to construct an algo to best reflect the page's relevance? There's more to scoring than just PR.
| 10:08 am on Oct 3, 2003 (gmt 0)|
|Dave, there's a difference between a page with a list of outbound links, like a directory or a links page, and a content page on a site, like the homepage that would lead to product pages. |
Well yes there is. But not as far as Google assigning PR to the links on that page go. Besides this, why do you bring up this point? The subject is "importance of location of link on page" not "importance of location of link on a content or homepage" :o)
|If we were working on a search engine and sat there looking at a web page, what would we look for on that page in order to decide how to construct an algo to best reflect the page's relevance? |
Lot's of factors, but I doubt that all the ones beggining with the letter "A" would be given more weight that the rest. In others words, IMO a links position on a links page would not be taken into account. It it were, all links that have anchor text starting with "A" would all have the higher PR.
| 1:25 pm on Oct 3, 2003 (gmt 0)|
all links that have anchor text starting with "A" would all have the higher PR.
My question is: do users (on average) click more often on links at the top of a list (unordered or ordered by any rule)?
*If* it's more probable that users click on a more prominent/visible link, then a good distribution model should give more weigth to the PR passed by that link.
I'm not saying that Google takes in account prominence of links, but if it will happen in the future it could be a logic decision. IMHO
| 8:35 am on Oct 4, 2003 (gmt 0)|
I don't see how Google could give more weight to links at the top of the page even if it wanted to.
What do we mean by "top"?
It's quite possible for a link to appear near the top of the HTML source file yet appear at the bottom of the user's screen because of graphics, formatting, alignment, tables, CSS, etc, etc. And that's before thinking about frames (spit).
I don't believe that Google makes any attempt to render the HTML, it just grabs the bits it understands from the source file. As such it simply doesn't know where different links will appear when the user views the page.
Of course link placement IS important to users and it's them we should be designing for first, not Google.
| 8:43 am on Oct 4, 2003 (gmt 0)|
I agree. I doubt very much that Google cares about a links position on the page when to it comes to PR. It just makes no sense and what Google does usually does make sense.
It will take some pretty solid evidence (and the reason behind it) why Google would consider sites starting with "A" are more important that sites starting with "B", "C" etc.
| 4:14 pm on Oct 4, 2003 (gmt 0)|
What do we mean by "top"?
I mean "at the top of the page rendered by a rendering engine".
I completely agree with you; Google does not have the technology to render billions of pages and to determine the position of the links.
The actual "random surfer" model guarantees equal probability to the page link's clicks. This is because there's no technology capable of determining whichever link has more probability to receive more clicks than another until now.
Whenever this kind of technology will be created, it will be useful, because it will generate more accurate results of which pages receive the most hits.
| 5:34 pm on Oct 4, 2003 (gmt 0)|
>>Besides this, why do you bring up this point? The subject is "importance of location of link on page" not "importance of location of link on a content or homepage" )
Oh, it does make a difference. There have been changes specifically affecting link pages lately; there are posts around to that effect. A links page is basically nothing more than a shell containing the site navigation and links going off the site, with any text being supportive. It represents the content of other sites, which may or may not be related. Link pages aren't the same as actual content pages on a site.
>>Google would never give more weight, or PR to a link that is higher on page than another
It doesn't necessarily have to be PR that's being affected. There's a big a difference between being higher or lower on a page in a list of links or higher or lower on a regular site page, and whether or not links are part of the site's global navigation or the main body text and/or content of a particular page.
Higher or lower could mean nothing. Technically, site navigation could be put smack dead center on a page with all the actual content surrounding it.
Global navigation is relevant to the site as a whole, and the particular site links in the navigation are relevant to the pages they correspond to and serve to connect up the pages on the site that relate to each other, but strip away the shell that contains the common site template containing the global navigation and what we're left with is the meat of what each page is actually relevant for.
It isn't just PR that's important with where a link is geographically on a page. Even if it's got nothing to do with PR, there are more important things and there are other factors that could possibly influence it.
| 1:11 am on Oct 5, 2003 (gmt 0)|
Sorry Marcia, but I can't figure out what your point is and why you are telling me what you are?
I'm simply saying, I doubt Google gives more weight to a external link 'higher' on a page than a link 'lower' on a page.
| 3:04 am on Oct 5, 2003 (gmt 0)|
There's one place it matters for certain, and that's beyond 101K. I don't know why Google wouldn't judge link placement differently, and they could do a million things to do it currently: size, contrast, position, etc. It's consistent with the random surfer model, and I think they are doing it now, but that's my opinion and feeling based on my results.
| This 47 message thread spans 2 pages: 47 (  2 ) > > |