I have the feeling you didn’t read any of the comments in this thread. I don’t think anyone could and come away with the sense that Yelp’s filtering practices and policies make the reviews better, or that the filtered reviews are not worth anyone’s time.
I agree that they’re looking for consistency and/or community. I’ve written 50 or so Yelp reviews with a fair amount of variance among the scores, and none are filtered. Presumably if there’s an extortion scheme involved, sooner or later ones of yours or mine would get filtered…which hasn’t happened so far.
Here is one example I noticed. (Disclaimer: I’ve never been to the restaurant, just looked it up because a friend was planning to go.) 16 reviews and 31 filtered. The filtered reviews are overwhelmingly negative, but most of them don’t appear to be hit pieces or incoherent. Most of them appear, on their face, to offer legitimate (and fairly consistent) complaints about the food and service. Because they were disallowed, a restaurant that would otherwise have received almost unanimously bad reviews has a rating of 3.5 out of 5 stars, which gives the impression that most reviewers liked it.
I find the filtering interesting. Obviously, it’s a pretty tough problem to solve, and it’s important that Yelp do something about it or the whole thing would just be stuffed with fake 1 and 5 star reviews. I actually have an RSS feed of yelp reviews from my neighborhood (via EveryBlock.com, which appears to require an account now, but they let you build a custom RSS feed for various occurrences in your neighborhood, like crimes and yelp reviews) that I read regularly, but I only ever read the one-star reviews, because they’re the most interesting. They generally fall into one of two buckets: 1) legitimate warnings about shitty businesses (you can tell because they’re coherent and tend to agree with each other), and 2) absolute crazy people who are on a crazy rant about something hilarious and stupid. I recommend it.
So there’s a very popular (and overpriced) restaurant in St. Paul, MN that I signed up for Yelp specifically to review. There was an endless tide of praise for the place, and when I went there, it was just so overwhelmingly average that I went home to write a review right away. I didn’t slam the place, but did review the dishes my friend and I had as spectacularly mediocre, and that it was just factually inaccurate to refer to the place as “fine dining,” as many of the reviewers had, considering the wait staff wears jeans. It stayed up for a while, got marked helpful a few times, and another reviewer wrote that he agreed with me and was majorly disappointed. Next thing I knew, both reviews were filtered.
But whatever. I don’t really care about that. What I care about (and I started a thread complaining about this) is how useless Yelp reviews are, apparently in part due to their absurd filtering policy. It boggles my mind that 2/3 of Yelp reviews are 4 and 5 stars! How can every place be so uniformly good? I completely ignore Yelp reviews these days, because I could get as good accuracy by rolling a die. I use it as a restaurant search engine, and ignore the star ratings and reviews entirely.
Yelp sucks.
As the OP, I could hardly have put it better. I find it hard to use Yelp as just a listing service, because it’s built around ratings and reviews and you have to restructure the presentation as you browse through it.
I wish there was some clear vision of what Yelp *thinks *they are doing with this bizarre, opaque process. As it stands, it’s baffling and somewhere between idiocy and madness.
Actually no. I like the filtering. I hate obvious shill reviews or short useless reviews. And, yes, of course they get hit with shills & spam, and they’d be silly to post their metrics for IDing such crap.
The evidence is that as many legitimate reviews get filtered as anything obviously a slam or shill. That’s my entire point: their filtering makes no sense, does not eliminate obvious shill posts, does not eliminate one line “this place sucks” posts, and seems actively biased against what seem to be thoughtful, detailed reviews by people who actually visited the business and came away with something good or bad to say about it.
I think their “metrics” were written by a programming intern they hired from Craigslist and could be written on a matchbook cover leaving room for two jazz progressions and a phone number.
I see no such evidence, I have indeed checked the hidden reviews, and never have I found anything of use or interest. I am sorry one of yours got filtered.
shrug I guess it’s a boon for condescending assholes who are sure the system has their best interests in mind, then.
Speaking from experience: no, you couldn’t.
It’s likely that the filtering is not done on the basis of content, but on metadata. For instance, they might be filtering out reviews posted through proxies, or from networks that don’t match the location.
I don’t by any means mean create fully implemented instances, but the list of meta-instructions, certainly. I’ve done it for other platforms and purposes.
Once once once again, what baffles me about Yelp’s purported filtering system is how wildly erratic and nonsensical the results are. If we, the poor end-stage lusers can see it, it would seem that their in-house people could see the same faults and do something to correct it. A review system that rejects a high proportion of quite evidently sincere and real reviews in favor of a generic center-model smacks of a simplistic, first-draft idea that’s never been re-evaluated. As a WordPress plug-in, fine, who cares. As the system managing - automatically, and autonomously - the reviews on one of the largest review systems in the e-world, it’s… just yet more baffling.
Compare it with, say, Amazon - a review system I trust completely because while the system rejects some reviews (this just happened to me; see other thread) whatever standards they use don’t arbitrarily push the results to some center. There’s no reason Yelp can’t do as well.
Take a look again, the vast majority of those are explained by review count, I think. There are 3 or 4 exceptions though.
If you can solve this problem easily, you should probably be making a lot of money doing that instead of whatever it is you’re wasting your talents on.
You mean aside from Amazon having about 100 times the resources and infrastructure Yelp does, and generally being a fairly cutting-edge technology company?
I wouldn’t say they’re being coy or obscure about its existence, or not clear about the apparent reasons for it - they’ve had this posted for over 3 years:
http://officialblog.yelp.com/2009/10/why-yelp-has-a-review-filter.html
Oh, and they also made a 3 and a half minute video to describe it: http://www.youtube.com/watch?feature=player_embedded&v=Dqi-jjbEKcs
Yes, but it doesn’t really say anything. “Yes, we filter. No, we won’t tell you how it works. It’s automated and prone to changing its mind. And we change it around from time to time.”
And it’s clearly not doing a very good job, as a dozen and more people here have pointed out, with specific examples of crappy and suspicious visible reviews and inexplicably filtered reviews. It’s my belief that we’ll eventually find out that it’s the product of one quirky (=pathologically idiosyncratic) programmer or SA who controls the entire process. It just has that hallmark; I’ve used too many programs written and maintained by one person with a peculiar, and feedback-resistant, view of how the app should work.
As for Yelp vs. Amazon, ALL Yelp is is a review and ratings service - so I’d expect them to do at least as well on Job #1, managing user-contributed reviews, as does Amazon, which does dozens of things, most of them well, where review management is probably Job #38.
A place I used to work at currently has a pile of 4-5 star reviews, when you jump through the hoops you see a pile of 1 star reviews from pissed off customers. interesting indeed.
If I had to guess, this is due to a sort of heuristic (has elements of the availability heuristic and peak-end). When one wishes to review a restaurant, the most salient examples will be the best and worst restaurants one attends. Restaurants which are significantly below average will tend to receive less patronage than the better joints. Even the very act of dining out could be favourable enough to an individual that they rate such an experience as above average when compared to their typical meal.
I get that. Not a lot of people are enthused about adequacy, but I do raise an eyebrow when reviewing restaurants and it seems like *everybody *likes everything. Then you skip to the filtered reviews and go “Oh, actually, a lot of people hated this place.”