The idea that I would take some random stranger's advice because they earned some 'Yelp Stars' or whatever by wasting a bunch of their time on the site trying to feel important is ridiculous. I stopped looking at these pretty quick after I realized what a time-sink it was that did not seem to result in me patronizing any better places than I was before. (And the fact that Yelp seems to go out of there way to prevent me from surfacing details I might actually be curious about from reviews). It seems they are much more interested in optimizing the site to encourage people to write reviews rather than make those reviews useful to those who might be reading them. Why are they creating a situation where 500 people are writing 500-word reviews that no one will ever read? When the most sophisticated aggregation of this they seem capable of doing is "4 people mentioned the cashew chicken! [What they had to say about the cashew chicken, well good luck figuring that out]."
The only model of crowd-sourced reviews that interests me at this point would be one that normalizes ratings against an individual's biases (so someone who gives every place 2 stars is not skewing the average) and models preferences to give personalized recommendations (since you like establishments X and Y, you might like establishment Z that other similar users have rated highly). AFAIK Yelp isn't doing any of this.
If a place has 700 reviews and a 4.5 star rating that's good signal that it's good, regardless of how many words are put into the 3 reviews you see on the front page.
If a place has 5 reviews and a 3 star rating that doesn't mean anything to me. Maybe it's great maybe it sucks, but I'm not turned off by it. It's just basic stats. If you're trusting one good 500-word yelp-elite review on a listing with 2000 2-star reviews, frankly you're doing it wrong.
Sometimes long reviews offer good things to search for -- for example someone might talk about a patio or happy hour that isn't identified on the site or yelp listing. I've used them for that purpose a few times.
Exactly. I'm under no illusion that Yelp is a particularly useful filter to pick out the ultimate dining experience. But I travel something over 100 days a year. If for at least some of those days when I'm in unfamiliar locations, Yelp can--with a little bit of skeptical evaluation on my part--help me pick restaurants that are mostly to my tastes and on the better side of average, it's done its job. I am somewhat critical/fussy about restaurants in general so I don't mind spending a few minutes to do some research.
Same. Few things irk me more than spending money on mediocre food, especially as I've gotten better at cooking. I may not eat at the best/hottest/hippest joint in town when I'm in a new place looking things up on yelp, but I can pretty consistently eat someplace that I'm happy with.
The only model of crowd-sourced reviews that interests me at this point would be one that normalizes ratings against an individual's biases (so someone who gives every place 2 stars is not skewing the average) and models preferences to give personalized recommendations (since you like establishments X and Y, you might like establishment Z that other similar users have rated highly). AFAIK Yelp isn't doing any of this.