Forum Moderators: open
I really don't know why Google hasn't figured this out yet--reciprocal links should be a penalty, not a boost, at least as far as "normal" searches go.
Consider this: Let's say we have a "widget" website--we'll call it website A--this website contains a lot of information about widgets and has an internal web of links, however, it has few links to other sites about widgets. If website A has a lot of inbound links then they will have a high PR because they're thought of as authoritative/popular concerning widgets.
Now we have another "widget" website--call it website B--which only, for the most part, contains links to other sites that have widget information. The PR of website B should be calculated on the inbound links, sure. But its PR should be negatively affected by the amount of outgoing links it has to "widget" sites...and certainly should not be positively skewed.
This is my point: website A has information on widgets...and that's what Google should be returning in its RPs; website B has *lists* of websites with information on widgets, and therefore for our SE purposes is competing with Google--remember if I type in "widget" on Google I get a *list* of sites concerning "widgets".
Implement the algorithm outlined for website B and reciprocal links lose value; inbound links become more valuable; and content returns to the fore.
Peter
[edited by: Critter at 6:01 pm (utc) on April 11, 2003]
It simply figures that a site specialising on selectign and reviewign other widget sites through a human webmaster, will always be better at it then the automated, global, general google engine. Therefore Google wants to pass the searcher for widgets on to the site with links to widsget sites, cos it'll provide better results.
In a way it does compete with google, and in its humbleness google asumes humans are better editors and reviewrs then itself and therefore concedes to this "authoritative" site.
Authority isn't neccessarily content... Just think of yahoo or google. Google has hardly ANY content at all.. it just links to it well...
SN
Unfortunately your view goes against the entire premise of PR--that the web dictates what's relevant and popular and authoritative by linking to pages, not individual web pages ("link" pages in question).
The algorithm would have to take into consideration the number of inbound links for a particular subject and the number of outbound links: for a "link only" site that has 1,000 incoming links but only 25 outgoing links (the "list") the outbound links would not negatively affect its ranking too much.
However, for a site that had 1,000 inbound links, and 800 outgoing links for a particular subject the ranking would take a (proabably big) hit...because it's just a reciprocal link site.
In this way, outbound links do still negatively affect a site's ranking, but if others link to you as a "link" site you'll still get PR.
With the weighting of the outbound links against the inbound links on a per-subject basis you can see that this still alleviates the "reciprocal link for PR" problem, but leaves good lists pretty much alone.
Of course, outbound links should *always* negatively affect your PR for a subject, because the website in question is not the source, it's simply linking to the source...and secondary sources should *always* come afterward in the SERPs.
At the end of the day this is the main point: when I type in "widget" in Google, I want to go to a site about widgets, not to a site that says "click here to go to a site about widgets".
Peter
How does penalizing a site for similar inbound to outbound numbers solve any problems?
Many informative sites have tons of inbound links because they are the place to go for information about a particular "subject". If the "subject" is something you can purchase, they typically have many outbound links on where to go to purchase it.
I agree that many recip sites have a high PR, but they are usually sending people to other sites anyway. So it makes sense to find those recip sites and get a link on them.
The key here is "for a particular subject"...
If a site has 1,000 links that's fine, but if a site has 1,000 *incoming* links from "widget" sites, and points back 1,000 *outgoing* links to the same "widget" sites it would be penalized.
You could even bring this to a higher level, so that mutual links (links that point back to each other on a per-domain basis) are cancelled out. But the broader subject-based algorithm would server better because it wouldn't be as subject to outside manipulation.
I'm looking at it from the end-user's perspective. If Google implemented this algorithm it would be one more step towards being the "authoritative list" for any particular subject itself, thus replacing the function of the individual "link only" web pages.
Your analogy of "items for purchase" pages falls short. A page like this would most likely *not* have inbound links from the vendors it points to, because in linking to the "list" page, they're giving their current users a resource TO FIND THEIR COMPETITORS.
Peter
PR is just a mathematical representation, it has NO inherent value. What it DOES represent is the the structure of good content on the net. After all links don't grow like trees, they are put there by people. So PR is simply a method to quantify this linking activity of the REAL people. Simply because Google admits to not beeing any good at really understanding the web... so it uses PR to basically leave the "understanding" of the web up to the people making the links.
If I have a website on shoes, I only get links form sites who want to give their visitors info on shoes. Now If I am a site with links to all websites with clothing products in my city, than I'm authorative, anybody (including those who link to shoes) who wants to link to ANY item of clothing can now link to me. So if my site is great for people, it's great for google, cos Google wants to show in its results pages that are good for people.
SN
[edited by: killroy at 4:43 pm (utc) on April 11, 2003]
Again you're making my point :)
If you're a website that has a list of sites that sell shoes, then, like you said, you're going to get inbound links from website that want to provide their users a "shoe resource".
That means that your inbound links probably don't have anything much to do with shoes, right? Or why would they be pointing their users to you? Why would sites that already provide information on shoes link to your list of sites that provide information on shoes? It doesn't make sense except if it's a reciprocal link; and this adds *nothing* to SERPs except clutter.
Please notice, this reciprocal link cancelling algorithm is on a per-subject basis.
I know this is a tough pill to swallow, but this is the point for this posting: If Google wants to be the resource for finding information, then sites that proclaim to be a resource for *finding* that same information are competitors of sorts, and detract from the SERPs at Google; contrariwise, if Google continues to include reciprocal link variable informaion in its algorithm (i.e. does not punish reciprocal links) then Google becomes *less* of a resource for doing so.
Peter
Here's why I link to sites and why sites link to me. My wife and I run two websites -- mostly as a hobby and a little bit to keep my self challenged. These websites are for people in my "city" in Asia. One deals with expatriates and the other deals with parents. They're resource sites and we link to websites and create directory entries for providers who don't have websites.
When we link to someone, we send them a fax / email saying that we've linked to them and we'd like them to review their listings. If they belive our site(s) are of any value to their visitors, they can link back to us. You know what, most of the businesses and commerical sites we link to, will give us a link as they belive we're providing a valuable resource to the communities in general.
The reciprocal link helps build a cluster of sites which link to each other and helps form a virtual community. I belive they should be dealt with just like pages that link to each other within a site and should result in a themed-cluster. Once you go several layers deep into the reciprocal links you'll realise that these sites form a valuable pool of resources and service. Has to be a formula to this.. thats why Googleplex is filled with PhDs.
Now I know.. this does not apply to most businesses and I'm perhaps voicing a naive thought here. But hey... dont' write of reciprocals. Thats how some of us get a LOT of our traffic from these links.
[added: I forgot to mention, many people who do not have websites and whom we list in our directory, tell us that they don't mind if we leave a stack of business cards or brochures for our site at their reception / counter. How would you penalize that? :) ]
[edited by: shrirch at 4:47 pm (utc) on April 11, 2003]
it's easy for a bot like Google to find all sites relating to shoes. what is REALLY HARD is to also find sites on socks, on trousers, on shoe bands, and on shoe cream.
Now if the inbound linking site has nothign to do with shoes why the heck would the send their visitors to an information resource about shoes?
Now if they're selling suits, they'll send you to the shoe resource to pick a nice pair, ergo they are related.
Unralted sites DO NOT LINK NATURALLY. now spamming and all that is a different story, I'm merely tryign to illustrate the way Google "thinks"
SN
Now tell me, why, if I'm entering "shoes" in Google is Google showing me pages with "socks"?
:)
Google not knowing about relationships between sites is neither an advantage or disadvantage...that's what linking is for--my point is links between *same-subject* sites.
Again...if you have a directory-type site that links to 100 other sites about clothing or all sorts, and you have 1,000 incoming links then you'll still get a (dampened) PR.
And, again, you're making my point about the "shoe resource" site...an incoming link from a "suit" site is not an incoming link from a "shoe" site, so that your outgoing links to "shoe" sites would incur no penalty.
Peter
<^_^>
[edited by: Critter at 5:00 pm (utc) on April 11, 2003]
I am sure that Google is not really happy about the fact that reciprocal links are used by unethical webmasters to manipulate pagerank. Unfortunately, the only way they can ever deal with this problem (i.e., entirely discount reciprocal links in PR calculation) is to recognize sites rather than just individual web pages, as they now do. Usually, www.his-site.com/links.htm points to www.her-site.com/, while www.her-site.com/links.htm points to www.his-site.com, thus these links are not seen as reciprocal by Google.
Many sites with good content link to other sites with good content on a reciprocal basis. Lowering these sites would push these good sites downwards. Think of how many Yahoo pages contain links to sites which link right back to that specific page.
I think that what we're talking about is the difference between a machine and a human editor. A human can look through a list of sites and say, this one has real content and this one is exclusively links. A human can say that this one seems to only link to pages which only reciprocate links and, aside from that, doesn't have a good "links directory."
I'm totally in agreement that this is a problem. I started with a links site myself but now I publish mostly content. As I move away from links, I watch others with just links move up in the serps largely because they create reciprocal link arrangements. It has gotten to the point that most of the sites in my space are purely link libraries. That is not good for the web. That's akin to having an authoritative book with 20 pages of content and 980 pages of index.
I just don't think a machine can be set to have the natural intelligence to apply the principle correctly every time. There will be tons of errors which will ruin the relevance of the search.
Look for info on widgets,
go to widgetsite1.com from google,
find a link to widgetsite2.com on that site,
go to widgetsite2.com from widgetsite1.com
find a link to widgetsite1.com on that site,
go to widgetsite1.com from widgetsite2.com
find a link to widgetsite2.com on that site,
go to widgetsite2.com from widgetsite1.com
find a link to widgetsite1.com on that site,
go to widgetsite1.com from widgetsite2.com
etc....
and never find information about widgets!
reciprical links should not be a penalty, but they should NOT contribute to PR
They link to each other.
Sounds good to me. Why should either site be penalized? Why should either link be discounted?
If there's no inbound links for Site X or Site Y then only X & Y think that the content's useful...
And they link to each other...
:) Hehe...
REMEMBER, Google's whole PR system is built upon what is POPULAR...why would they want to consider your scenario at all?
Peter
but on a more serious note,you are right, don't penalise for linking to each other, but also don't give them google-benefit from it
You're absolutely right about Yahoo...but Yahoo's directory becomes pretty useless in light of pagerank (as we've seen).
So Yahoo's directory would take one heckuva hit if all the sites they linked to linked back to them.
But for increasing Google's RP relevance, this is the way it should be.
Peter
Site x is about widgets so is site Y.
Both sites link to many other sites, and are linked to from many other sites. {added: Not all of the links are riciprocal ]
They also link to each other.
Both provide different useful content.
Wht should either site or either link be discounted or disregarded, or devalued, whatever you want to call it?
My point is that Google can determine which site should show up where in its results pages. Cancelling the reciprocal links between the two does (relatively!) nothing to the sites' PR in Google.
And when Google cancels these types of links Google's results become more relevant.
Peter
Sure I know that people can misuse this. But how to say what's misuse and what's not (if you are no human being but an algo)?
The whole thrust of this topic is to improve the algo :)
As I said in my previous posting to Ken: there's no penalty for the reciprocal links--they simply do not count.
But then you ask: "Well without these links between common subjects how on earth would anyone find similar pages?"...
Um, Google has a 'Crawler' :)
Peter
