Nofollow - The Great Debate

[Micallef]

I've noticed something recently on BHW. Many new members have recently had to watch in horror as their thread topic surrounding any kind of link-based issue is inevitably transformed into a battle to the death regarding the nofollow attribute. The intention of this thread is to provide a place for arguments and counter-arguments to be held on the subject of the rel="nofollow" attribute.

I have detailed my thoughts below - although clearly opinionated in one way, they are meant more as an essay on the subject. I hope that even those who disagree with the final conclusion will find the information provided useful in some way.

View attachment 4743
​

rel="nofollow"

Nobody outside Google fully understands the effects of applying ?nofollow? to external links, and there is active debate on the matter in SEO circles. In order to compose the most probable working model of ?nofollow? and its effects, a holistic view must be taken.

• Google wishes to retain and expand its market share. It can do this by being perceived as the best search engine available. This entails producing search results which provide a better user experience than its competitors can provide.

• Despite tremendous advances in the complexity and plasticity of its algorithm, Google?s search results still rest heavily on the ?backbone? of the PageRank system. This system uses links as ?votes? to determine node importance. Because links can be artificially created (link spam), Google?s results are vulnerable to external manipulation by spammers, who have an adversarial relationship with Google.

• When Google?s results are manipulated by an adversary, the user experience is badly degraded in most cases. Therefore, link spam is a practice which seriously threatens Google?s public perception as the best search engine.

• It is therefore beyond reasonable doubt that Google will use any and every means it has at its disposal to either prevent linkspam, or negate its effects.

Regarding the rel=?nofollow? html attribute, Google has made the following statements:


?Essentially, using nofollow causes us to drop the target links from our overall graph of the web? ? Google Webmaster Support


?Google does not transfer PageRank or anchor text across these links? ? Google Webmaster Support


Google are frustratingly vague in many ways regarding nofollow. While they often assert that nofollow will not assist in rank, the official statement on Google?s website only states the above ? that PageRank and anchor text are not passed. This has left the implementation of nofollow open to debate by many who view the choice of words here as critical. Not passing PageRank or anchor text does not necessarily equate to no ranking benefits. However, I believe that "no ranking benefits" is exactly what was meant here. Google just assumed, when making this statement, that everybody thinks PR = rank.


?In general, we don't follow them? - Google Webmaster Support

(I would like to add that the webmaster support page was "updated 4/29/2010", and so is a current statement from Google.com)


?The essential thing you need to know is that nofollow links don?t help sites rank higher in Google?s search results.? ? Matt Cutts


These statements are not necessarily true simply because they come from Google. Indeed, there is every reason to be suspicious of Google?s public statements. However, it would also be unusual for such a large and respected corporate entity to explicitly lie about a matter such as this, especially one which prides itself on being "good" and "honest". Each person will have to make up their own minds regarding these statements, though, as there is no definitive proof - precisely the reason why this issue must be debated.

A simple Google search will reveal that the overriding opinion is that links carrying the ?nofollow? attribute are indeed dropped from Google?s graph. The popularity of this opinion does not make it valid. Nevertheless, it makes more sense to assume that it is correct rather than incorrect, for the following reasons:

• Google?s search results, as stated in more detail above, are based - at least broadly -around link graphs. PR is only one of 200 factors, but after content, it is the back-bone of the system.

• Distortion of these link graphs by adversaries results in a serious decrease in the extent to which Google can achieve its primary goals. I know this because I regularly engage in this practice, often with nothing more than brute force irrelevant high PR links for poor content. It still works.

• The ?nofollow? tag has been implemented on most of the major social networking platforms, e.g. Facebook, for links which users can create. It is also present on the most well-known blog commenting platforms, such as Blogger and Wordpress. Thus, an adversarial novice will find that most of his attempts at manipulation (spam links) are accompanied automatically by the ?nofollow? tag.

In essence, Google has struck upon a way to identify links which cannot be vouched for by a trusted authority. These links are normally user generated, and in using the ?nofollow? tag automatically on these links, the webmaster is effectively saying to Google:

?I cannot vouch for these links. Please do not include them in your ranking calculations.? - This is my interpretation of Google's statements on their webmaster support page. I appreciate that others have different interpretations.

This sounds very altruistic, but the webmaster has two motivations of his own for championing the use of the nofollow tag:

• As a deterrent - warning link-spammers that if they attempt to leave spam links on the website, they will not be credited by Google, resulting in wasted time and effort.

• As on-site SEO - preventing the traditionally understood outflow of PageRank from the webmaster?s domain. As mentioned earlier, Google has publically stated that PageRank is not passed over links which have had the ?nofollow? attribute applied to them. (Edit: this is now only a perceived benefit, as it was recently revealed that PR "evaporates" when hitting a nofollow link, equating to a loss for the webmaster similar to that of a d0follow link)

Google has therefore created a situation in which everyone except link-spammers seem to win: Webmasters benefit from decreased spam and not linking out to bad neighbourhoods, and Google users benefit from an enhanced search experience. Ultimately, Google benefits by retaining and expanding its market share and advertising profits.


A few notes - The problem of linkspam detection is technically an "NP hard" problem:

"A term used to describe problems or subproblems in artificial intelligence to indicate that the solution presupposes a solution to the "strong AI problem" - that is, the synthesis of a human-level intelligence. A problem that is AI-complete is, in other words, just too hard."

- http://foldoc.org/AI-complete

It must however be said that Google have made some remarkable leaps, but the nofollow tag is still extremely useful to Google in tagging and blocking potential spam links.

As mentioned earlier, Google's spam problem is severe. The potential benefits of recognizing more social signals are dwarfed by the problems that would be created if suddenly, every novice blackhat SEO was able to get full credit for links they crudely placed on twitter, facebook, wordpress comments and more. Remember that only a very small percentage of the links on the internet are "nofollow" links.

Further, the suitability of links generated by social networking users to determine a web page?s relevance must also be questioned. Despite being real people with valuable opinions and recommendations for web resources, social networking users may not be the most considered and objective judges of the internet?s content. Links can be posted on a whim, in poor taste, in fact for any reason whatsoever. In other words, there is a huge amount of noise, and it would be extremely complicated and intensive to filter through it all.

Contrast this with a webmaster?s link to a web resource. The webmaster has presumably had to go to the effort of creating or editing a blog post specifically to cast his ?vote? for said resource. He also has the reputation of his website to consider, and he will be judged by the accuracy of his recommendations. Google has every reason to trust this type of link ? typically free of the ?nofollow? tag ? over links dropped on SN sites.

Broadly, these are my reasons for stating that rel = "nofollow" links provide insignificant Google ranking benefits, if any, in the overwhelming majority of cases (98% or more).

---

I ask that replies to this thread do not consist of circumstantial or anecdotal evidence concerning how "my site got ranked with a nofollow link". Tracking of this kind is difficult and can only be trusted under large scale controlled trials.

Read the post, challenge what I have said if you want.

Also, I would like to stress that this topic is about nofollow's direct effect on the target site's rank in Google. We all know there are other traffic sources out there, and that nofollow links can be useful for human clicks.

:swordfigh

 

[gorang]

It's my strong belief that no follow links are explained as simple as "They are crawled by Google for the index, but they do not pass any value to your website including "link juice" and anchor text".

Getting crawled faster is probably the only reason why some people believe there is an insignificant increase from using them.

 

[pk_69]

Great post, Micallef

With details supporting the idea on nofollow it's still a controversial topic and I wish I have more academic knowledge to support my arguments so this is all just thoughts..

G's intent to limit number of links from the web graph is make sense and if webmaster instructed to ignore links G just ignore them. Even more, webmaster is actually encouraged to do so, because nofollow links do not pass (and thus retain) page rank.

It's true that this is stopping link spam, but it also pushes webmasters to break web graph. At the beginning everything G said was in form of encouraging nofollow outlinks, but it looks like they realized that there will be nothing to spider soon, because all links are nofollow. Take as an example Trust Rank concept - all hight trusted web sites are almost 100% nofollow, you can't even spider the web now.

Recently G repeatedly mentioned that linking (do follow) to relevant sites is a positive factor and it obviously should outfactor negative value of passing page rank out to provide a value to ones web site.

The problem is that with G's help it stuck in every webmaster's head - the one of the important way to improve page rank is to turn ALL outlinks nofollow and until proper balance of follow/nofollow will be used by every site G just has to go though nofollow links just to access the web. I think they follow these links, but apply big negative weight to such links. If above is true (it's just an assumption) there might be some very small weight left after the math and this might explain why some people see a positive effect out of huge amount of nofollow links.

pk

 

[Micallef]

Thanks PK, this is the sort of discussion I was hoping for. Any more insights?

 

[RankWithShelly]

wouldn't it be nice if this took care of that problem because honestly, seeing the same freaking posts on the same subject over and over and over is like running on a treadmill. tiresome and boring. Now maybe we can just direct people here because frankly I don't think anyone who cares enough about the topic too debate it will change their mind any time soon.

 

[bo2kmm]

I'm sure we all agree that the majority of major authority sites that are easy to drop links on are all nofollow - for good reason.

Some people focus on only getting links that exclude nofollow, the technically incorrect "d0follow" term.

It is my understanding that if that the entire link profile is devoid of almost any nofollow links it will look unnatural. I have thought for some time that Google will algorithmically penalise any sites that gain a lot of links quickly with below a certain percentage being nofollow - as this would look unnatural.

Saying Google completely ignores them cannot be correct as if you look in Google Webmaster Tools it often has nofollow links listed - for example I have a number of links from yahoo answers listed in GWT and they are all listed as nofollow.

Does anyone else have any thoughts on this specific point about percentage of both?

Matt

 

[MisterGemini]

So this is where you been hiding.

Ok.. I don't have time right now to rewrite an essay like you did .. but I do have a little tidbit of UP TO DATE information. You see, this whole nofollow thing started 5 years ago. All that stuff we just got a heavy review by the OP of has been in spin for a while. It's intent was to stop link spam for the purposes of spamdexing and manipulating Googles then precious PageRanking system.

Now, Google has had a run at it for a number of years. The introduction of the microformat (nofollow) was really a test/attempt to do something about it.

HOWEVER, on October 14, 2009, something very important happened. Pay attention now.

We are (or should be) all familiar with the little tool from google called the Webmaster Tools section. Well on that day, we recieved confirmation from google employee Susan Moskwa that suddenly, google decided to REMOVE PAGERANK from the Webmaster Tools!!

But but but... it's SO IMPORTANT.. HOW COULD THEY DO THIS?! Remove the MOST IMPORTANT FACTOR that affects your SERPs?!

Well, here it is in her own words:

"We've been telling people for a long time that they shouldn't focus on PageRank so much; many site owners seem to think it's the most important metric for them to track, which is simply not true."

Ohhhhh... So today.. in 2010, PageRank just is not the most important metric anymore. Ok... but but but.. hold on... lets see what else google has to say about Pagerank today:

Q: My site's PageRank has gone up / gone down / not changed in months!
A: Don't worry. In fact, don't bother thinking about it. We only update the PageRank displayed in Google Toolbar a few times a year; this is our respectful hint for you to worry less about PageRank, which is just one of over 200 signals that can affect how your site is crawled, indexed and ranked. PageRank is an easy metric to focus on, but just because it's easy doesn't mean it's useful for you as a site owner. If you're looking for metrics, we'd encourage you to check out Analytics, think about conversion rates, ROI (return on investment), relevancy, or other metrics that actually correlate to meaningful gains for your website or business.

Ohhh ok. So there are 200 signals that are used, and Pagerank just isn't that important anymore. Got it now.

So alright.. Page rank is just 1 of 200 signals. Micallef has already supplied us with ample information and proof showing us that you just don't get any Pagerank passed with nofollow links. So lets do the math.

1 link with nofollow = (200 signals - 1 Pagerank signal) 199 signals
1 link without nofollow = 200 signals

So what have we learned?

1. Nofollow was an attempt by google to stop spamdexing back in 2005 because Pagerank was so easily manipulated by various link spamming methods.
2. In 2009, they finally took it out of Webmaster Tools entirely and said they have basically abandoned Pagerank all together, and that it was just 1 signal of 200 that they now look at.
3. 200 signals - 1 signal = 199 signals. Does this mean the nofollow link will not pass Pagerank. YES. Does this mean you will not gain 199 other signal benefits. NO.

So there you go. nofollow is a dated/outdated method which Google themselves have attested to it's lack of import in relation to the Pagerank factor when it comes to the various 200 signals which they CURRENTLY use today in their search mechanism.

For reference to my quotes you can find them at the following links:

http://www.google.com/support/forum/p/Webmasters/thread?tid=6a1d6250e26e9e48&hl=en

http://sites.google.com/site/webmasterhelpforum/en/faq--crawling--indexing---ranking#pagerank

Now we all know that when it comes to links, THEY ALL MATTER! With or without nofollow there are at least 199 other signals (that's 99% more signals than pagerank) that are being passed on and back to the site linking and being linked to.

:tee: Drink and be merry again, and go get your links!

 

[boiler]

So this is where you been hiding.

Ok.. I don't have time right now to rewrite an essay like you did .. but I do have a little tidbit of UP TO DATE information. You see, this whole nofollow thing started 5 years ago. All that stuff we just got a heavy review by the OP of has been in spin for a while. It's intent was to stop link spam for the purposes of spamdexing and manipulating Googles then precious PageRanking system.

Now, Google has had a run at it for a number of years. The introduction of the microformat (nofollow) was really a test/attempt to do something about it.

HOWEVER, on October 14, 2009, something very important happened. Pay attention now.

We are (or should be) all familiar with the little tool from google called the Webmaster Tools section. Well on that day, we recieved confirmation from google employee Susan Moskwa that suddenly, google decided to REMOVE PAGERANK from the Webmaster Tools!!

But but but... it's SO IMPORTANT.. HOW COULD THEY DO THIS?! Remove the MOST IMPORTANT FACTOR that affects your SERPs?!

Well, here it is in her own words:

"We?ve been telling people for a long time that they shouldn?t focus on PageRank so much; many site owners seem to think it's the most important metric for them to track, which is simply not true."

Ohhhhh... So today.. in 2010, PageRank just is not the most important metric anymore. Ok... but but but.. hold on... lets see what else google has to say about Pagerank today:

Q: My site's PageRank has gone up / gone down / not changed in months!
A: Don't worry. In fact, don't bother thinking about it. We only update the PageRank displayed in Google Toolbar a few times a year; this is our respectful hint for you to worry less about PageRank, which is just one of over 200 signals that can affect how your site is crawled, indexed and ranked. PageRank is an easy metric to focus on, but just because it's easy doesn't mean it's useful for you as a site owner. If you're looking for metrics, we'd encourage you to check out Analytics, think about conversion rates, ROI (return on investment), relevancy, or other metrics that actually correlate to meaningful gains for your website or business.

Ohhh ok. So there are 200 signals that are used, and Pagerank just isn't that important anymore. Got it now.

So alright.. Page rank is just 1 of 200 signals. Micallef has already supplied us with ample information and proof showing us that you just don't get any Pagerank passed with nofollow links. So lets do the math.

1 link with nofollow = (200 signals - 1 Pagerank signal) 199 signals
1 link without nofollow = 200 signals

So what have we learned?

1. Nofollow was an attempt by google to stop spamdexing back in 2005 because Pagerank was so easily manipulated by various link spamming methods.
2. In 2009, they finally took it out of Webmaster Tools entirely and said they have basically abandoned Pagerank all together, and that it was just 1 signal of 200 that they now look at.
3. 200 signals - 1 signal = 199 signals. Does this mean the nofollow link will not pass Pagerank. YES. Does this mean you will not gain 199 other signal benefits. NO.

So there you go. nofollow is a dated/outdated method which Google themselves have attested to it's lack of import in relation to the Pagerank factor when it comes to the various 200 signals which they CURRENTLY use today in their search mechanism.

For reference to my quotes you can find them at the following links:

http://www.google.com/support/forum/p/Webmasters/thread?tid=6a1d6250e26e9e48&hl=en

http://sites.google.com/site/webmasterhelpforum/en/faq--crawling--indexing---ranking#pagerank

Now we all know that when it comes to links, THEY ALL MATTER! With or without nofollow there are at least 199 other signals (that's 99% more signals than pagerank) that are being passed on and back to the site linking and being linked to.

:tee: Drink and be merry again, and go get your links!

agree with you and i already knowing that

 

[Numa68]

Great info here, and it's consistent with what I've seen.

It is very true that Page Rank matters much less in terms of rank placement. Almost any search you do will have PR0 and 1 sites sitting above PR6s and higher, unless it is specific to a large company and site, for example 'ipod'.

Although nofollow is not supposed to pass link juice, I can tell you that some of the best referrals to my sites are from nofollow links. Is this because of some kind of Big G magic? Likely not, as they are on highly trafficked sites where real live humans can click the links. However, I have seen GBot traverse these nofollow links to hit my sites.

As MisterGemini mentioned, the Big G has a list of quality factors they use to determine the value of sites. I think the energy spent debating the nofollow issue would be much better spent instead determining exactly what these quality factors are.

 

[pk_69]

wouldn't it be nice if this took care of that problem because honestly, seeing the same freaking posts on the same subject over and over and over is like running on a treadmill. tiresome and boring. Now maybe we can just direct people here because frankly I don't think anyone who cares enough about the topic too debate it will change their mind any time soon.

I admire professionalism and I'm so happy to have a chance to pick info from some of no doubt great profi in the field on this forum.

I definitely wouldn't come back to read yet another time that "link is a link" and "build as many as you can" "if it's true I would do it to my competitors" and so on - that is really boring and tiresome. Not because it's not true, but because it's trivial.

And we have an option to learn how to rank and monetize 2 pages site optimized for wedding speeches and bit the crap out of the idea until new comes alone with new WSO or learn how the best sites do it. I'm sure it takes a lot of reading, thinking and experimenting. And discussing...

pk

 

[MisterGemini]

As MisterGemini mentioned, the Big G has a list of quality factors they use to determine the value of sites. I think the energy spent debating the nofollow issue would be much better spent instead determining exactly what these quality factors are.

We got 200 factors to figure out. It's difficult to wrap your head around all the possibilities. I am sure they are not listed anywhere in googles patent archives either.

The earliest data available from seomoz is from last year. Since this year we have experienced a climeratic change in how google fundamentally ranks, much of that previous data is questionable now.

You can view the 2009 data from seomoz here: http://www.seomoz.org/article/search-ranking-factors#ranking-factors

Again I emphasize that the information they reveal in that link I just shared here is now somewhat questionable. However they do look at some of the various factors that affect linking that give you some idea of the various factors that are involved. Note though that even the 'experts' that were involved there, had made mention that they believe the nofollow thing to be pointless.

I am not entirely sure what the ratio might be, but signals are both negative and positive. I believe from an algo perspective this could be as simple as a scoring number for each signal as a + or - and then calculated all together to produce a weight number. That number is then calculated into the site being linked to, and into the site linking out.

 

[Numa68]

We got 200 factors to figure out. It's difficult to wrap your head around all the possibilities. I am sure they are not listed anywhere in googles patent archives either.

True, but I'll wager that 20% of those factors impact 80% of a site's ranking. That cuts the list down to a more manageable 40 signals which can be guessed a bit more easily.

If I had to guess what some of those signals were, based on what used to work versus what works better now, the list might look like this (in no particular order):

• Page load speed
• Relevant backlinks
• Ratio of internal versus external links
• Ratio of code to content
• Searches and clickthroughs from Big G's sites
• Prevalence of a site within other Big G services
• LSI and LSA scoring and weighting of content within each page and against other pages in the site
• Rate of content change
• Usage of certain HTML elements

 

[MisterGemini]

I don't think you've understood a single thing I've read and already you're preparing to hit the quote button and repeat yourself quoting the same nonsense over and over again, the same nonsense I've shown above is irrelevant.


I've been a bit rude here but I find your arrogant attitude offensive. The way you say "what did we learn" is kind of annoying, you know?

Sorry you found it annoying, but this thread was stated as a 'debate'. It's not just what you say but how you say it to get your point across. Actually attacking/insulting my intelligence was your way of discrediting my position. So you know exactly what I'm talking about. I approached it by undercutting PageRank, since that is the dominate reasoning OP was using to justify outing nofollows. I went after the concept, not the person.

I was actually well aware of the total 200 factors and never thought it was all just in links (no, I am not that dumb, thanks for the flame... again). I didn't present it as such or dive into those details for simplicity, which as you know, in debate can go a long way to convey an idea/position.

I added the link to seomoz so that people could see for themselves what some of those factors might be and decide for themselves if they really wanted to dive into it. Doesn't cover all 200, but it also obviously doesn't cover just backlinks.

That's all I got to say about that, and I don't think I want to have to explain myself for my debating anymore.

Back to backlinks... you are right about the anchor text point. That is a major factor that impacts a number of elements relating to links. Rather than dishing out a whole lot of that here, there is a great posting here that summarizes a number of those points well:

http://searchengineland.com/71-tech...is-from-30-link-building-experts-part-2-37240

Essentially they cite 71 various factors to look for when considering backlinks. This was made up from contributions of 30 link building experts, and it was put together March this year.

I feel driven to setup my own nofollow experiment so I can say 'look here is your proof', but the many examples cited by others about the nofollow links impact on their rankings has always been discredited with an ambiguous 'you have no way of knowing' response (OP oftens says that ).

So, in order to avoid such and wasting my time, I am going to ask everyone/anyone to suggest to me what they would consider to be as definitive a test as possible in determining the effects of a nofollow link from the various popular sites that almost everyone here uses and gets nofollow links from. At the same time, I will test d0follow and see where it all goes.

Think this will help with the debate?

Suggestions?

 

[blackhatterguy]

I believe no follow links are still good

for example you go on a nofollow page rank 5 or page rank 4 site and comment there, we are not sure whether we will get pr juice or not, but at least we will get a few more visitors to our sites through those comments.

So, follow or no follow, you will be benefited.......

 

[RankWithShelly]

I think a test would be interesting. Create 2 sites that have the same spun content for a certain keyword. Use a wordpress blog and call it keyword1.wordpress and keyword2.wordpress. same template, plugins, etc. and backlink one with no and one with do and see what happens...

 

[MisterGemini]

I think a test would be interesting. Create 2 sites that have the same spun content for a certain keyword. Use a wordpress blog and call it keyword1.wordpress and keyword2.wordpress. same template, plugins, etc. and backlink one with no and one with do and see what happens...

Here are some initial thoughts to add to this.

1. Will register separate domain names for each one.
2. Will use wordpress basic install as the test bed with all external links removed.
3. All Domains will be on separate IP addresses in different C Classes
4. Content will just reflect the keyword withing the code without any additional content. Or us lorus epsum filler?
5. Nofollow control group will get it's links from 2-3 specified nofollow sources that have PR3
6. follow control group will get it's links from 2-3 PR3 sites as well.
7. Pages links are obtained from must carry the same link density
8. Sites links are obtained from must be similar in size/authority/traffic
9. Use the following labs:

NoFollow Control Group

Lab1 - Domain Keyword Match + Nofollow Links

Lab2 - Domain No Match + Page Keyword Match + Nofollow Links

Lab3 - Domain Keyword Match + Page Keyword Match + Nofollow Links

Lab4 - Domain Unique Word Match + Nofollow Links

Lab5 - Domain No Unique Word + Page Unique Word Match + Nofollow Links

Lab6 - Domain Unique Word Match + Page Unique Word Match + Nofollow Links

Follow Control Group

Lab1 - Domain Keyword Match + follow Links

Lab2 - Domain No Match + Page Keyword Match + follow Links

Lab3 - Domain Keyword Match + Page Keyword Match + follow Links

Lab4 - Domain Unique Word Match + follow Links

Lab5 - Domain No Unique Word + Page Unique Word Match + follow Links

Lab6 - Domain Unique Word Match + Page Unique Word Match + follow Links

This is a start I think. I want to here from others though to point out where something else needs to be considered. I have a feeling after this people will say something like 'well it depends on the niche' or something like that. I don't know. This seems to cover every feasible possibility that could affect the links impact. Or is this overkill? We are trying to keep this focused on just the link itself and it's impact without other factors affecting it.

Suggestions?