There was a forum that used to have as a requirement a non-free email account and seemed to have no issues with spam accounts with tens of thousands of members for more than 10 years. In that use case it seemed the non-free account aspect to sign-up was the threshold which seemed to keep spammers out vs the fact such an email account could be (with relevant authority) traced back to a real identity.

I'd be curious if there is a study that has looked into the thresholds for different use cases at which spam account creation drops to negligible amounts and how much price vs anonymity vs difficulty factors into it.

Which is great when databases leak. Absolutely brilliant.