That's a bummer. I've found that TripAdvisor reviews tend to have a higher signal to noise ratio than some other sites (including Google's own reviews, where you can tell a lot of the reviews are just being cranked out so that power users can get points and "level up")
They are no angels either, or at least their daughter company. Two weeks ago TheFork featured in Dutch TV program Rambam where they proved that they boost restaurant ratings. They listed their own fake restaurant (possible, no problem) and gave their own reviews, trying to become lowest rated in the ranking.
Giving rebates on couverts may boost your ranking too.
Restaurant owners interviewed are not too happy as well. They feel forced to subscribe because of popularity of the site, but have to pay 2 euro per person per reservation. Also it happens that without their knowledge TheFork arranges the Reservation button on Google search, so they flow through their platform.
edit: To clarify the boosting. If according to ratings the average should be a 5.2 then TheFork would up that and make it e.g. a 6.5
I used to work on the search team at TheFork years ago (we were already a TripAdvisor subsidiary) and don't know about the current practices but back then there was no foul play about the ratings, which is not even the default sorting on the site.
Also, not specifically related to TheFork (because I don't remember...), do all ratings should always have the same weight? Are old ratings as relevant as newer?
If you want to book a romantic dinner are ratings from groups as relevant as the ones from other couples?
Is a new restaurant with two reviews at 7/10 average the same as another one with 200 reviews at 7/10 average ?
I used to work at TripAdvisor, and I can corroborate—while I was there the focus was really heavily on fighting back fake reviews and presenting the ones we got as effectively as possible. You can ding them for a lot (the UI has always been kinda painful) but, unlike Yelp, I don’t think they are or have been dirty.
I would expect rating should be an exponentially weighted moving average, first and foremost, with ratings also weighted by some form of user reputation. The user reputation is the hard part (at least for Amazon I guess)
I'm sure all services have similar problems so it's just a matter of picking your poison. The problem is that Google is so synonymous with search that people have forgotten they're a business. They expect the "best/most relevant" search results. Instead they get biased ones that tend to favor Google. And most will just take that as the best result.
And Google does its best to obscure the fact that the search is anything but fair (see also the recent move to blur the lines between paid ad results and organic search results). To most it just looks like the super smart algorithm found exactly what they needed. In reality it just found more Google.
So maybe the problem is not necessarily Google but people's expectations. Nobody asks a McDonald's clerk what the best burger is thinking they'll get a fair comparison with Burger King’s offer. They do expect this from a Google search.
> Restaurant owners interviewed are not too happy as well. They feel forced to subscribe because of popularity of the site, but have to pay 2 euro per person per reservation. Also it happens that without their knowledge TheFork arranges the Reservation button on Google search, so they flow through their platform.
I've always wondered how this works. Especially when most travellers these days rely on Maps / Tripadvisor to plan their trips. Restaurant owners are almost forced to be on those platforms.
