Just a quick glance says that is definitely not an earned page of links. I'd love to know if Overstock gave Mr. Arora just a discount, straight up cash to post, or what. And how did they reach out? Did they email every student at utdallas.edu and offer them an incentive in exchange for posting a page on their personal site?
It actually makes me feel a bit better to see this is actually pretty black hat, and not (at least in my reading) quite as debatably paid-links as the Journal's story made it sound.
It's not that bad, it's barely even optimized, too many keyword links and not enough unlinked mention of the keywords. This is amateur. Keyword density is way off.
If he did a page for each topic, then he'd be on the money. Heck, he isn't even keyword stuffing the description, the url, or H1 tags. Nor does he have any linked images w/ alt tags.
Mainly because if you're paying, you're paying too much. What you want to be doing is creating highly relevant content in PDF/word docs w/ links and then uploading them to forums. Can't nofollow a PDF/doc link :)
My opinion is that they do count. Intuitively, a URL is a URL and whether it's from an HTML document or a Word doc it's the same thing. You'd have to put in a bunch of extra code to make it not work like that. It's a $9 test, get a random domain fhg56784ei3dg.com with a random keyword, do absolutely nothing with the domain, throw a link in a PDF and see if your site gets indexed :)
You suppose Google has a "test detection" heuristic? For example, if we see a completely unique word suddenly appear on 2 domains that themselves have no possible meaning or prior reference, and everything is the same except 1 variable, then we'll just randomize the results? Seems like a test scenario itself might be algorithmically detected and f----ed with.
I intend to test this eventually, but I think the test has to be a little more complex since you want to measure if it shares google-juice, not just that Google reads the URLs.
Ideally I'll registering two random number domain names, both with random text around a unique word, then mention one in a keyword rich PDF and see which domain pops up first in the Google search.
It's not just about google juice. You can boost pages within your own site with internal links. Also, people search for an absolutely stupid amount of stuff on torrent sites. Try putting your PDFs on torrent sites for a little extra traffic. If you seed it on the right trackers all the torrent sites pick it up for you for free.
As well sometimes the torrents themselves rank really well. Keyword stuff your torrent files :) Remember to set the utm_source on your PDF links.
Is breaking Google's rules now a PR/link building tactic? WSJ didn't link to Overstock.com in their article, but if they did, then Overstock would have built a really strong, legitimate backlink as a result of their illicit practices.
I also think the article plays down what Overstock.com did. I'd love to hear what Matt Cutts has to say about it, but offering a discount to students and encouraging them and their universities to link to said discount page doesn't sound like it should violate Google's policies. Judging from some of the links posted in other comments, it looks like Overstock.com was being far more manipulative than that.
I'd also question whether this is actually a penalization? Or did Google just devalue the links that weren't kosher? I see those as two different things.
Google has said in the past that they defuse links and ignore any pagerank they would pass if they determine their are paid or for the sole purpose of passing pagerank. So, yes, there is a difference between penalties and devaluing links.
So why isn't Google better at search? Seriously. They're like tax legislators, constantly surprised that people game the system. Only they have less excuse, because there's no transparency in Google's 'laws,' and no coherent way to distinguish 'white hat' and 'black hat.'
