So, the reviews never happen without the marketing push.
The interplay between the marketing and the reviews are to help consumers decide if the vendor fulfilled their promises to the public. Critics with prestige will only review films that they anticipate will have an audience. The audience will go to the critics to see if they verify the sizzly promise from the trailer.
There is no demand for critical evaluation of a product without trying to generate that demand first through marketing and advertising. Try launching a product without doing any marketing at all and see what happens (absolutely nobody will buy your thing).
Different consumers care differing degrees on how much they want to research before they make a purchase.
Ask an indie film director how unimportant promotion and hustling are to actually getting people to see a movie.
Yelp and all crowdsourced reviews that don't even try to verify purchasing are easily scammed and gamed at scale.
Similarly, reviewing on Youtube -- especially for things like games and children's toys -- is rife with concealed bribery for influence -- illegal in the US. This stuff is harder to 'disrupt' than it seems.
>But this doesn't scale to huge cities with tens, hundreds or thousands of choices. We can't know the reputation of such a wide array, and we can easily be fooled by advertising. Internet-based recommendation systems make the grapevine and reputation scalable. I hope we see more innovation. Yelp is a start, but it fails miserably because if I rate a place 5 because I love very authentic Thai food, and another person rates it a 1 because they are used to Americanized Thai, the restaurant gets a 3 (I'm simplifying for illustrative purposes). In other words, Yelp's rating is useless to both me and the other person.
What happens is that people in cities tend to pick a small number of vendors that they trust based on repeated positive trading interactions. They might decide to test out different vendors if they receive a sufficiently enticing offer.
Yelp is not terribly good and doesn't even make money. This is also kind of a non-problem in search of a solution: people will evaluate restaurants based on peer evaluations, their own experience, professional critics, and crowd reviews. The kind of people who make judgments based on crowd reviews might just want to be with the popular crowd rather than quality. What crowd reviews tell you is what's popular -- since you don't have a sense of the personality and knowledge level of each amateur critic, each individual review carries little weight.
So for example, most Yelpers would freak out at the 'slow service' of a super-high-end restaurant because they're tasteless scenesters. If you want to know what the herd of mindless scenesters think is hot right now, Yelp'll tell you. Or in your Thai example.