Wednesday, October 03, 2007

Debunking the FUD around "rel=nofollow"

I finally decided to put my thoughts in black and white and let people know why I think the Emperor G has no clothes which is why they invented "rel=nofollow" in the first place.

Think about how much we hear about relevance.

Relevant results are all the buzz in the relevance of Google's search results and targeting of AdSense ads so they must be the experts in relevance, right?

OK, if they have such a good lock on relevance then why couldn't Google simply determine that spam links from comments on blogs, forums and wikis had NO relevance to the topic of the post and simply discount those links automatically.

Once you grasp the implications of my last paragraph you'll realize that "rel=nofollow" is bogus.

If you're still not sure, think about this for just a moment with a simple scenario where Grandma posts on her crochet blog about a new crochet pattern and then a spammer spams her blog about Viagra or a bunch of other pharma and off topic crap.

Would we be led to believe that the world's greatest search engine with the best search relevance bar none can't tell that the link and comments about Viagra don't match the content of the blog post and can't automatically discount those links without a "rel=nofollow"?

Apparently not, and here comes "rel=nofollow" and all the FUD and fear mongering about who you link to, who you can sell links or legitimate ads to, and whether or not you can pass link juice or not without the risk being penalized if you don't bend to the will of the same company that ironically makes their billions selling paid links.

Does anyone besides me smell a hand job penalty for paid links?

If Google can't tell that the Viagra ad was off topic on Granny's Crochet blog then how are they detecting paid links?

With all that said I think "rel=nofollow" at a minimum is a good idea just to automatically discount random links from random people posting on blogs, forums and wikis just to take the possible SEO reward out spam. However, that won't stop the spam because the same stupid people that open those emails and go to those websites will still click the links in the spam posts as well so the direct traffic will still be a big enough incentive to continue spamming websites. The only upside is it thwarts their efforts to gain rank in the search engines.

However, if Google's relevance detection, especially with off topic links was really that good, the spam posts never would've been a problem in the first place.

Does anyone smell a rat?

Unfortunately the rat I smell is mixed content sites such as many news sites, forums and blogs with random topics per page that probably caused too many false positives for an algorithm to automatically discard what would appear to be off topic links.

Therefore, the algorithm probably failed and here we are scared into policing ourselves with "rel=nofollow" and every now and then someone caught selling a paid link or something is thrown on the sacrificial altar just to stir some high profile FUD and keep everyone in line.

That's my theory, what do you think?

3 comments:

Da Scritch said...

I use rel="nofollow" because some of my links should not be referenced. By example, in my blog, you have lots of rss feeds, mainly comments (by post, by category). I don't want them archived, because not every browser knows how to deal with rss.

But also because rss is absolutely not relevant in this case : comment are out of their context, and visitors are losing interactive experience with my site.

rel="nofollow" was an idea to stop spam-comments, it worked halfway... but it is also very useful in numerous ways, and, IMHO, this is a perfect testcase : links that should be seen out of their context.

IncrediBILL said...

Of course there are other valid uses for rel=nofollow but that's not what it was intended for initially.

Like you said, it was an idea to stop spam but if you understand my point Google should've already been able to discount those links because they weren't relevant.

Da Scritch said...

OMFG ! http://help.yahoo.com/l/us/yahoo/search/webcrawler/slurp-12.html Yahoo spiders (and probably other ones) are crawling those links, and parsing them. The rel="nofollow" is just taken for not modifing the ranking value of the target link for this specific link.
(and so IMHO, is useless)

So I'll have to write a special rule in my robots.txt.