My thoughts as well. I can envision a platform where you can review anything from specific make and model of a product to a restaurant to anything else. I think it could be immensely useful.

But it would quickly become a victim of its own success as people start using it and the database gets big enough to be useful. bots and spammers would swarm the platform to try to game it. That would require a constant cat and mouse game that would not be fun. Without significant ongoing attention, the site is going to become worthless. In order for a site like this that is labor intensive and not fun to exist, it would need to be monetized at least to the extent it can pay salaries. That's the point at which I don't see any realistic solution. In order to maintain neutrality, it would need to be supported by users, meaning users would need to pay. I don't see very many people paying for something like this. And even if some would, there would have to be some benefit or reason for paid users to pay, and that seems like it would open the door for more cat and mouse games and abuse.

Just a hard problem I think.

The Wikipedia model is close to what I had in mind. One way (there are others) identity can be solved for using Stripe Identity or something where you same for a gov credential to identity proof. It costs you $1.50 per request, but is no match for bots. Assuming you bind passkeys or other strong auth factors at account provisioning, and there are only so many high quality reviewers, you’ve got a reasonable limit to effort required.

High quality people want to be helpful, they just need a platform where curation and distribution takes place. See: here. I’m just tired of seeing startups and other companies build and kill these knowledge and community based third spaces, and the intangible value inherent to them.

(customer identity is a component of my work in fintech as an infosec lead, I get paid to squash bots and fraud)

I have to imagine that a big part of any solution to review gaming and bot spam is data diversity. eg: whether the submitter has been within X meters of the thing they're reviewing in the last Y days.

Google seems to have done a good job with review spam on Google Maps, better than their job with SEO spam and malicious search results, and deep+diverse datasets are effectively Alphabet's (not just Google's) entire business.

Yelp would have a very hard time getting to know as much about review submitters, an open data platform would have a hard time gathering even what Yelp already collects.

wikipedia depends on citations to reliable sources and has a policy prohibiting original research. this is why

Where "reliable sources" tends to mean traditional media. Which makes it both more reliable (for many purposes) and much more limited. For restaurants, I do tend to trust newspaper restaurant critics more but there's also far less of that especially at a small-scale non-urban local level.

i don't think newspaper restaurant critics count as 'reliable sources' in wikipedia with respect to the questions people are interested in here, such as whether the chicken is overcooked. i am pretty sure that if you try to put an assertion like that into a wikipedia article about a restaurant, it will be reverted as unverifiable pov pushing, regardless of how many newspaper restaurant critics you can cite in support of it. you can probably cite them for things like when the restaurant opened and why it moved to its current location, as well as use them for evidence of notability

I don't really expect something in Wikipedia to get into the nuances of particular dish quality. But, in general, I would expect that--to the degree Wikipedia gets into the quality of a particular restaurant--a New York Times review would carry a lot of weight. In general though I agree that Wikipedia is probably not the place for restaurant reviews even to the degree they're sources from well-known restaurant critics.

this is generally limited to things that wikipedia editors of vastly opposing viewpoints can agree on, like whether and when the restaurant was awarded a michelin star

like, imagine a vegan, a devout hindu, a muslim, a carnivore, a neighborhood kid, and a foodie editing an article about a texas steakhouse that is known for its barbecued pork. whatever text any of these people finds unfairly biased is going to get deleted from the article, and what's left is what they can agree on

this process produces astoundingly reliable articles on contentious topics like the armenian genocide, but the results are predictably of poor literary quality, and generally not very useful for things like guessing which restaurant you would prefer to eat at