Saturday, June 22, 2013

Google's "unnatural link" detector is broken

I recently had a comment from a webmaster of a site I had linked to (who appears to check out as a real person) asking me to remove the link because they'd been hit by an "unnatural link penalty" from Google due to their being linked from my blog.

I did some googling around, and discovered that the "unnatural link penalty" is intended to reduce the page rank of websites that have spam posts linking to them, to render that kind of spamming useless.

The problem is, my link wasn't spam.  My link was a natural link in the truest sense of the word.

I'm not going to link to the post in question because apparently it causes trouble for this person's business, but it was one of my posts from during the financial crisis, where I was trying to figure out money-related stuff.  One aspect of what I was talking my way through would vary greatly from person to person, so I provided a link to an online financial calculator so everyone could calculate their own number for themselves.

This was absolutely natural.  I'm just a regular person who writes about what's on my mind. I was writing out my train of thought, partway through my train of thought I realized that everyone would have a different number and it was complex to calculate, so I googled up an existing online calculator and provided a link.  This is the very purpose of hyperlinks dating back to the earliest days of hypertext.  I had no interest in the particular business I linked to, they were just the first googleable result providing that particular kind of calculator.

The thing is, this is the very basis of Google's PageRank system - that people will link to thinks that are useful to them.  Google Webmaster Tools support says:
The best way to get other sites to create relevant links to yours is to create unique, relevant content that can quickly gain popularity in the Internet community. The more useful content you have, the greater the chances someone else will find that content valuable to their readers and link to it.
The site I had linked to had created relevant content that was useful to me and valuable to my readers, so I linked to it.  And now they're apparently being penalized for it.  Why?  Because I'm a blog?  Because there are people who use blogs for spam?  I don't even know.

From a purely algorithmic point of view, Google should be able to tell that I'm a human being, not a spambot.  My blog has been around a lot longer than most spam blogs have.  I don't update on any particular schedule.  My posts vary greatly in length and nature.  Quite a few of my posts have no links in them whatsoever.  I change my template every so often, and there are posts with the word "template" in them around the time of these template changes.  There's a twitter feed in my right-hand column, and it's updated on no particular schedule with the majority of tweets not containing any links.

On top of all that, Google owns Blogger.   I'm sure they have access to information that will show them that I delete spam posts, I have drafts of posts in my drafts folder, and I hardly ever make scheduled posts.  They could probably also see that I use the account I blog with here to comment on other blogs.  That's not the behaviour of a spambot!

I know there has been a lot of blogger spam lately, but you have a system where a website creates useful content, a blogger thinks "That's just what I need!" and links to it, and then the website gets punished for this, your system is broken.  Google needs to fix this!

4 comments:

Lorraine said...

I will say that your blog attracts a lot more comment spam than most, based on the email comment subscriptions that I start whenever I comment to your blog (such as right now). I'm assuming you zap comment spam, but maybe googlebot beat you to one once.

Also, Google itself seems to be a recurring topic of discussion here, so maybe from some algorithmic angle it looks like an ŜÊÔ-oriented blog. Note the diacriticals. That is how one discusses that topic on Twitter without acquiring dozens of one-day followers from that industry.

Lorraine said...

I would unlink a page if someone asked me to, but I will not ask permission before linking, and I'm one of those people for whom "deep linking" is a political issue...

It's what makes the web a web, for heaven's sake.

laura k said...

Couldn't it be a momentary glitch on Google's part? Once in a while, I am denied access to my own account because Google doesn't recognize me (or something, who knows), and then it does. I assume it's just some imperfection on their part.

You're absolutely right that your link was natural, and Lorraine is absolutely right about not asking permission before linking. I always find it bizarre when someone asks me if it's ok to share a link of mine.

And lastly, I have never seen a blog get as much comment spam as this one.

impudent strumpet said...

I think Google has improved their spam filters in the past week or so - I haven't had any spam comments make it through in days, knock wood.

Although, algorithmically, Google should be able to differentiate from links in posts and links in comments.

I agree about deep linking too - before we even get into a question of the politics of the issue, linking to the home page just isn't useful enough to be worth doing! In the case in the blog post, I wanted to say "Here's a calculator so you can figure out how the stuff I'm talking about applies to you," not "Here's a financial services company that exists."