.edu links

37 comments
Source Title:
Links from .edu's
Story Text:

I can only offer a "gut feeling" on links from .edu domains, and it goes like this "bollocks, it just happens that .edu's tend to have more authority"...

But Andy does offer a different opinion, and i'd bow to his seo knowledge over mine any time (except maybe not this time...)

what do you think?

Comments

.edu domain availability resemble the .com

Suddenly all the good URLs are registered.

Google seems to favor .EDU links

But I say that based on the research I conducted earlier this year. It has been several months since I tested this hypthesis and for all I know they have changed the way they handle that top-level domain.

However, all they need to do to favor .EDU domains is to adjust their damping factors across the board. That is, given two damping factors X and Y, if X is better (gives more power to links) than Y, then Google can arbitrarily apply a damping factor of X to .EDU domains while applying Y to everything else.

In my opinion, Google started assigning better damping factors to .EDU, .GOV, and .MIL domain names late last year or early this year. I believe they may be using more than one damping factor.

Again, I have not looked at this issue in several months. A lot has changed in Google since then.

I have not noticed any sort of bias toward .EDU, .GOV, or .MIL links on either Yahoo! or MSN. Someone suggested to me that ASK seems to favor .EDU domains, too, but I don't pay much attention to ASK.

quality indicator - however weak

Let me pose this question:

Take a random .edu URL. Take a random .com URL. Do this 1000 times.

Out of those two sets, which set has a higher % of quality, authoritative sites?

The .edu set, every time. By a significant, but probably not huge, margin.

> "bollocks, it just happens that .edu's tend to have more authority"..

I disagree. If for no other reason than the fact that edu's are harder to spam. DaveN can register 500 .com's a day. He can't get 500 different .edu URL's on 500 different C-classes per day.

Thus edu = quality indicator, since the .com set would have a higher % of blackhat type sites. And Google appears to be using any quality indicator they can (e.g., # of years domain is registered).

Now, I ALSO believe that .edu's tend to have more authority naturally. But aside from that (and that would already be factored in via Hilltop, PageRank or other link algo's), there is a quality indicator simply by the virtue of the edu TLD.

well

>He can't get 500 different .edu URL's on 500 different C-classes per day.

Bet he can :)

I'm with you though, links from .edu's are "worth more".

The true hypocrite is the

The true hypocrite is the one who ceases to perceive his deception, the one who lies with sincerity.

odd, 2nd person to call me a hypocrite today

geez NFFC what's with the personal attacks!!??

;-)

hehe

Its my sig here, I keep it as a reminder to myself, must have hit a post in error. Honest.

worth more

I'd be willing to bet some candy and beer (skittlebrau?) that Google places extra 'clout' with .edu domains. My guess is that it may be more than a little, too.

Haven't really looked into

Haven't really looked into this, but I know .edu domains are restricted to educational institutions, so how would any of us even go about registering one of them?

Still, this thread did inspire me to go do some linking work on the personal web pages I have at both the Universities I attended. So I guess education really does pay off, after all.

.edu

You don't need to register/own them just get a link on them. Volunteer some FREE webdevelopment time for your childs/nieces/nieghbors/complete stranger's school in exchange for a link or two.

Good Karma and a Good One Way Link.

In my opinion, Google

In my opinion, Google started assigning better damping factors to .EDU, .GOV, and .MIL domain names late last year or early this year.

Heck, the immediate aftermath of Florida (Nov 2003) really exposed the extra weighting to .edu and .gov - this was actually a big pointer for people to re-read papers such as Hilltop.

Phil Craven wrote a peice at the time discussing Hilltop and Florida, and used an example of how a professor's personal page about his putting up new shelves - ranking for the single keyword "shelves" - as an overt example of the extra weighting that .edu domain pages had. (Phil actually went on to dismiss Hilltop and prefer LocalRank as a prime set of factors applied in the Florida update),

However, Mike Grehan's Search Engine Book 2nd edition from 2002 also has techs talking about different weighting factors for weighting different domains - was it a Google engineer who mentioned once giving a little extra scoring to .com's in the very early days??

Florida is a red herring

You guys really need to stop naming the updates, as the names have no significance or relevance to anything.

Quote:
Heck, the immediate aftermath of Florida (Nov 2003) really exposed the extra weighting to .edu and .gov - this was actually a big pointer for people to re-read papers such as Hilltop.

No, the visibly significant increase in importance for these top-level domains is recent.

Quote:
Phil Craven wrote a peice at the time discussing Hilltop and Florida,...

I can now understand your confusion. Phil is not the best source of information on Google.

Quote:
However, Mike Grehan's Search Engine Book 2nd edition from 2002 also has techs talking about different weighting factors for weighting different domains

The original Google paper talks about applying different damping factors to various domains. It was generally assumed (at least by those of us who read the paper when it was first discussed on the Web years ago) that Yahoo! was given a favorable damping factor. When DMOZ became important to Google, it was assumed that DMOZ was also being given a favorable damping factor.

But the consistent improvement in results for .EDU, .GOV, and .MIL listings in numerous searches only showed up late last year and only solidified early this year.

.edu links

Michael, I have great respect for your hobby interests, but you seem to be "discovering" things that have actually been part of the commercial SEO conversation for a couple of years. And if "Florida" as a name has no significance or relevance for you in SEO terms then perhaps this is part of the problem.

I can now understand your confusion. Phil is not the best source of information on Google.

I remember the antagonism you both shared in SEW, but I doubt many SEO's will accept absolutes about how Google works now, from non-Google employees, discussing the finer details of very old papers.

No, the visibly significant increase in importance for these top-level domains is recent.

In the immediate aftermath of Florida, .edu and .gov pages were extremely visible even on commercial searches, and was one of the pointers towards an authority system having been implemented then.

An important point about an authority system like Hilltop (not necessarily applied, but illustrative) is that it can only work on a small set of terms, but can be expanded. The Austin update was fingered as evidence of an expanding set of terms impacted by an authority model.

Brian, your condescension doesn't change anything

You're still obviously confused about several issues here.

First, .EDU and .GOV pages have ALWAYS been highly visible. There has been an increase in publication of content on those domains over the past few years. The named Google updates have not had any impact on that visibilty.

What has changed, and only recently, is the way Google treats those sites. The Hilltop technology you keep referring to was actually incorporated into Google's algorithm PRIOR to the named update you so officiously cling to.

Instead of repeating the nonsense that people like Phil Craven have published on the Net, it would behoove you to actually pay attention to what Google is doing.

If .edu links carry extra weight ...

... then the equivalent from other countries ought to do so as well. Example: .ac.uk.

I'm not convinced that such links pack any extra punch, but nonetheless I'll certainly take them when I can get them!

Outvoted, fair enough :)

Outvoted, fair enough :)

I've done some checking on travel competitors

And there are 1 or 2 quite highly ranked outfits who are DEFINITELY of the opinion that .edu / .ac.uk / .gov etc links are worth it, considering the huge number they have striven to acquire. Fortunately, most of the rest of their SEO model sucks :)

It is noticeable that deep pages with a "non-com" backlink tend to be the pages that compete seriously for some big terms

... can register 500 .com's

Quote:
... can register 500 .com's a day. He can't get 500 different .edu URL's on 500 different C-classes per day.

And that's the reason why there is a perceived benefit in .edu IMHO not because there is a greater or lesser damping effect for the domain name but becuase there are less of the domains inthe 1st place and generally they are containers of higher quality, authoritive content, which becomes link laden.

This is where I agree with the thought process but not the implimentation of the experiment Notredamekid said.

Quote:
Take a random .edu URL. Take a random .com URL. Do this 1000 times.
Out of those two sets, which set has a higher % of quality, authoritative sites?

I'd suggest it becomes.

Quote:
Take a random .edu URL. Take a random .com URL. Do this 100,000 times for .com and 30 times for .edu
Out of those two sets, which set has a higher % of quality, authoritative sites?

My guess is that they will weigh out pretty evenly

.EDU "quality"

Quote:
And that's the reason why there is a perceived benefit in .edu IMHO not because there is a greater or lesser damping effect for the domain name but becuase there are less of the domains inthe 1st place and generally they are containers of higher quality, authoritive content, which becomes link laden.

While the perception of quality is very widespread, the reality is far from the perception. Many .EDU sites contain little more than fluff content (course descriptions, faculty vitae, class syllabi, students' notes, students' personal pages, faculty and staff personal pages, etc.). Some of the major institutions do actually publish substantive content in the form of peer-reviewed research (sometimes journals, or at least abstracts of journal articles) and a lot of un-reviewed research.

edu vs. com, in raw numbers

As of January, there were roughly 33 million registered .com domains, vs. 7480 .edu (a ratio of about 4460:1) - in case that helps anyone's calculations.

Thanks Erik, I'll update my

Thanks Erik,

I'll update my calc above :)

.EDU, .COM, .GOV, .MIL raw numbers

The number of registered domains doesn't tell you much. The number of pages that Google claims to have indexed from the various top-level domains will tell you a little more (but still not much).

NOTE: These numbers are accurate as of July 19, 2005. They will change over time.

550,000,000 .EDU pages
http://www.google.com/search?hl=en&q=site%3A.edu

1,790,000,000 .COM pages
http://www.google.com/search?hl=en&lr=&q=site%3A.com

370,000,000 .GOV pages
http://www.google.com/search?hl=en&lr=&q=site%3A.gov

9,770,000 .MIL pages
http://www.google.com/search?hl=en&lr=&q=site%3A.mil&btnG=Search

721,000,000 .ORG pages
http://www.google.com/search?hl=en&lr=&q=site%3A.org

80,700,000 .NET pages
http://www.google.com/search?hl=en&lr=&q=site%3A.net

61,200,000 .ac.uk pages
http://www.google.com/search?hl=en&lr=&q=site%3A.ac.uk

9,760,000 .gov.uk pages
http://www.google.com/search?hl=en&lr=&q=site%3A.gov.uk

I think the word "quality" is sometimes misunderstood

>> Many .EDU sites contain little more than fluff content (course descriptions, faculty vitae, class syllabi, students' notes, students' personal pages, faculty and staff personal pages, etc.)

Who cares? That's not the issue here. .edus are "quality" domains even if the content is little more than the undergraduate course notes, and students private pr0n collections.

It's all about the outbounds...

Fair point Michael but do

Fair point Michael but do you care to give the below a go which could answer the personal perception point of quality between .com and .edu ?

Quote:
Take a random .edu URL. Take a random .com URL. Do this 100,000 times for .com and 30 times for .edu
Out of those two sets, which set has a higher % of quality, authoritative sites?

The number of registered

Quote:
The number of registered domains doesn't tell you much. The number of pages that Google claims to have indexed from the various top-level domains will tell you a little more (but still not much).

With respect Michael you're changing the conversation from one thing to another. It was about .edu domains. You've now changed it to .edu pages. The 2 are not neccesarily related to each other.

When examining the value of links from one type of domain

Everything matters. So, no, I haven't changed the conversation from one thing to another. I have simply enlarged the scope of the existing conversation because looking at the number of registered domains won't tell us much of anything.

The fact that so many pages from .EDU domains are indexed actually weakens the favorable damping factor position that I and others have taken, because those pages are undoubtedly building up some internal link popularity for their domains.

Everything matters and I

Quote:
Everything matters

and

Quote:
I have simply enlarged the scope of the existing conversation

I agree 100% with #1 and you're right about #2, You have enlarged the scope rather than changed the conversation.

That's not to say I agree with the hypothesis but I've been wrong more times than I've been right in life :)

Jason D's percentage results

Quote:
Fair point Michael but do you care to give the below a go which could answer the personal perception point of quality between .com and .edu ?

Take a random .edu URL. Take a random .com URL. Do this 100,000 times for .com and 30 times for .edu
Out of those two sets, which set has a higher % of quality, authoritative sites?

This is, in part, why I brought up the number of pages that are indexed for the various top-level domains. The disparity between .EDU and .COM is not nearly as large as the domain registrations implies.

The .EDU top-level domain has plenty of content to serve up (even though quite often that content is not relevant to a query -- and I say that based on numerous queries of my own, where I have had to wade through the previously mentioned curricula, personal pages, notes, and etc.).

The reason I believe that .EDU sites are given favorable damping factors is that they seem to come up a great deal more often for certain types of searches than commercial sites which, I feel, provide more substantive content.

These are not necessarily commercial searches (keep in mind that 80% of all searches are NOT commercial in nature, according to the search engines).

It's not about the outbound links. It's about the assessment Google makes of the content. For reasons that none of us can explain, Google sometimes decides nonsense is more relevant to a query than substantive content. That's the price you pay when you rely on automation.

it just happens that .edu's tend to have more authority

imho, if this happens to be true it will be because the average of all "edu" sites will be higher than the average of all "com" sites.

More (of the right) people link to them and hence their status becomes higher. In the "com" world, OTOH you have a very large number of very inferior sites.

So, it's got nothing to do with the TLD, it's only because you have more crap on the average ".com" domain.

There is plenty of crap in .EDU content

The chief difference between .EDU, .GOV, .MIL and .COM, .NET, .ORG domains is that registration to the first three types of domains is restricted, whereas anyone can register domains in the last three categories.

The restriction itself tends to qualify the type of content that the sites will include, but .EDU sites usually include a great deal of fluff because they do allow students, faculty, and staff to set up personal pages (which can, in some cases, be quite extensive).

It's difficult to devise a reasonable test to determine whether .EDU domains are being favored over .COM domains. One means would be to search for content that is about evenly distributed between the two types of domains.

Many commercial sites, for example, make reference to studies which are also referred to on .EDU sites. But that is not a perfect test.

Some outside factors also tend to cloud the waters. For example, certain types of words (like "scholars", "scholarship", "peer review", "class", "reading", etc.) which are occasionally found on commercial sites (mostly news and media, in my experience) will be found more often on .EDU sites. So, I would expect those kinds of sites to rank better overall in classes of searches for those kinds of words.

Where I find myself seeing faroritism is when I look for surveys and studies that are meaningful to the business sector (economic studies, business performance, growth projections, etc.).

crap content perhaps, not necessarily sites

I agree that there's possibly plenty of crap content on "edu" sites, but the "authority" issue is not really about the quality of the content. And, the perception of "quality content" is individual - to each his own. Rather, it's about the linking power.

All I'm saying is:

Pages with lots of link power coming in tend to be able to give lots of link power away, regardless of TLD.

On the average, there's a better probability that any given "edu" page/domain will have good linking power, than any given "com" page/domain - because there is an extremely long tail of very inferior ("low link power") dot-com pages. The edu domains have a "low power" tail as well, only it's by all means much shorter than the "com" equivalent.

Claus, I see your point now

You're correct. There isn't likely to be any spam coming out of the .EDU top-level domain. Of course, that's another reason to assign that class of Web sites a favorable damping factor :).

Great Discussion!

Thanks for making an effort to keep the tone of the thread civil guys, it's not always easy when opposing views clash and there's little way of proving right/wrong on either side.

It's a good discussion, thanks!

Michael, I think you're

Michael, I think you're actually confused about the discussion - I'm not referencing content issues in terms of publication volume, but ranking visibility, and especially how .edu pages having extra weighting.

If you've only ever seen .edu and .gov pages more recently gain extra weighting for ranking purposes, then perhaps what you actually mean is that you've seen this happening in a more exaggerated manner of late. That's one issue.

The idea that the ranking of .edu and .gov pages have not been impacted in any way by Google's updates is another claim altogether, but I don't think you're going to be able to justify this, especially as you don't seem to understand the reference points being used.

As for claiming that Hilltop was incorporated into the Google algorithm before Florida - if it was, it was never particularly expressed until Florida, unless are seriously suggesting that before Nov 2003 Google based major keyword rankings according to an expert system as per that paper, as well as extensively devaluing links from within the same broad IP range.

There is no condescension intended here - everybody has their own opinion in SEO, everybody has different experiences of different markets areas - but you seem to be making an awful lot of high-brow claims for yourself that you seem to reluctant to be held to account for.

arg

IMHO, you guys are both stating things as "facts" which are not "facts". When I said my piece above where I thought .edu links were worth more, I explained a common sense reason why a search algo engineer might think they SHOULD be worth more: thus my conclusion was that it is likely they ARE worth more.

What you fellas seem to be doing is saying "I am sure I saw XYZ in the SERPs a certain time ago. This algo change was due to ABC. I am sure of it."

If we are trying to reverse-engineer what happens in the SERPs by gut feeling, or even watching tons of sites, I guarantee most of our conclusions will be flawed. The algo's have so many factors these days, that even someone who collects TONS of statistically significant SERPs data could only find loose correlations, NOT certain causations. This is all a bit silly.

Arg, indeed

Quote:
IMHO, you guys are both stating things as "facts" which are not "facts". When I said my piece above where I thought .edu links were worth more, I explained a common sense reason why a search algo engineer might think they SHOULD be worth more: thus my conclusion was that it is likely they ARE worth more.

I can think of plenty of reasons for why I might want certain types of domains to rank better than others, were I operating a search engine. However, I don't run Google so I can only look at what they publish both on their Web site and in their technical papers to understand what they are trying to accomplish.

Their system allows for some manual adjustments which are both arbitrary and fair -- in that they can change their damping factors for whole classes of sites.

Quote:
If we are trying to reverse-engineer what happens in the SERPs by gut feeling, or even watching tons of sites, I guarantee most of our conclusions will be flawed.

Long before you get to the point of reverse engineering anything, you have to identify the changes occurring in the SERPs.

The best reverse-engineers are the spammers, in my opinion. Some of them almost get on top of the changes at Google as soon as the changes are rolled out.

Now, maybe they just take the shotgun approach and cover as many possibilities as they can, and then see in the aftermath what works well.

Even so, that is a very methodical approach to figuring out what is happening, but it's not reverse-engineering.

For my part, I am still documenting the changes. Google has changed a lot of stuff over the past year. The conventional wisdom of the SEO community is really outdated for the most part.

Brian, you're still

Quote:
If you've only ever seen .edu and .gov pages more recently gain extra weighting for ranking purposes, then perhaps what you actually mean is that you've seen this happening in a more exaggerated manner of late. That's one issue.

Brian, you're still missing the point. The behavior of Google with respect to these sites has changed. The changes only began happening late last year.

Quote:
The idea that the ranking of .edu and .gov pages have not been impacted in any way by Google's updates is another claim altogether

And not one that I have made. I will appreciate people NOT putting words in my mouth (BIG HINT).

Google has been handling .EDU and .GOV sites differently over the last 8-to-10 months. You missed that fact. No problem. That's understandable, given that you seem to get your information from the SEO forums, which are usually 6-to-12 months behind the actual changes in Google, and which usually repeat outdated information for 1-to-2 years after it has ceased to be useful.

I have to research dozens of companies every week, and I look at their Web sites, and I look at the search queries I have to use to find their Web sites, and I know what comes up first in the searches.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.