It will be useless due to spam, unreasonably disgruntled employees posting lies, forced or fraudulent positive reviews, and even outright scams ("pay up or we will post a negative review every week"), etc.

I wish it weren't so, but I'm afraid that will be the reality of it.

All attempts to sell people on a truly anonymous system should begin with an explanation of the plan to not become 4chan.

I thought a little about this. While onchain you cannot censor it, the “view” created for convenience on web2 could just ignore sufficiently flagged responses.

This would still generally allow anonymity.

Of course you could also have anonymity on a traditional platform by not requiring people to sign in with usernames but use a private key instead. So just use that part of web3 but no blockchain storage.

How will you prevent a company flagging unflattering reviews in an effort to get them hidden (companies will pop up offering these kind of services, employing multiple computers, botnets, etc.)?

How does Twitter?

Answer: they don't.

If it's a big enough problem, it's noticed and addressed.

Sounds like you've never derived valuable information from 4chan.

This is a bit like saying sewage shouldn’t be kept out of the drinking water supply because it has valuable vitamins in it.