This actually doesn't seem to address the rumors floating around about Yelp, which were that they would use (misleading) negative reviews as part of high-pressure sales tactics for small businesses. The "favorite review" feature is what Yelp themselves said they were selling. It wasn't the nexus of the accusations against them.
I would love to know how they intend on filtering reviews successfully while exposing the process behind the filter. It would seem any transparent filter system would invite people to find reliable ways of bypassing and subverting the things that get you flagged.
There's a reason Google won't give you a full look at a page's spamminess rating, and a reason reddit is open source except for the spam filtering bits. With any blacklist filtering routine, transparency puts you at an inherent disadvantage. Instead of a million trials to see what passes and what doesn't, the answers are already there.
...or they aren't really, and this is just a PR scheme by Yelp to claim more transparency with the illusion of fairness for review visibility. To me this is the only real possibility here.
If a spammer wanted to see what a site took down, all they'd have to do would be to keep track
of all the comments (including ones not by their accounts). Then they'd have some idea how the filter worked.
If you make the list open then honest people can see too.
The only reason you'd keep the list of what's spam private would be to maintain (perceived) infallibility. Every filter makes mistakes, only some filters let you see them.
They don't tell you why things get flagged, just that they were. So, it's not that transparent. And I would be willing to bet that if you spent some time trying to learn the system through brute force, they have a system to watch for that level of activity and ban your account for tampering.
Still no way to request a manual review though. I already see several good detailed reviews by people that I know personally that have been flagged. As well as reviews filtered for businesses that I know personally and know for a fact that they don't pay anyone for reviews. From initial appearance, looks like a lot of innocent reviews get caught in the filter.
I looked around at some of the filtered reviews for restaurants I frequent. For the vast majority I couldn't find anything wrong with them other than that they were penned by users with less than 5-10 reviews.
I can understand why they don't offer any sort of manual review -- how is a staffer who has never patronized that business capable of determining if a review is spam or legit? I wonder if it would be any better if questionable reviews for a given business would be put up to a vote by "more established" users (by whatever metric they use to determine that) who have also reviewed the same business. Let people who have actually been there determine whether it's spam from someone who never walked in the door.
Well, the other weird thing that I see is that the reviews that have been filtered, are overwhelmingly highly positive. Which leads me to think that while their spam filter is agressive in filtering positive reviews, it is still vulnerable to businesses posting negative reviews about each other. Maybe that's just my limited sample size though.
I did come across a number of negative reviews by users who had limited activity on the site. I guess this is an effort to block "hit-and-run" negative reviews.
The downside to filtering out reviews by "less-established" users is that you run a risk of angering new users who take the time to contribute a review to the site only to find it doesn't appear. Are they really going to bother posting a second, third, or fourth review that will also get filtered out, or will they just walk away?
I hate to be cynical right out of the gate, but we'll see how it goes down. Admittedly, its a better move than doing nothing.
Could be interesting to see what kind of reviews are being flagged anyways. I've always thought that review stuffing is way more common than people would think...
