Using money as a gate has some issues. If you don't scale the amount to the region, you effectively exclude that region. But if you do scale to the region, it encourages the spammers to simply setup their operations (either physically or virtually) in the cheapest region (physically also has the benefit of cheap labour).

Requiring unique credit cards could mitigate that, but then there is PCI issues there, and its basically no different to requiring ID.

I assure you, paying for an account is a terrible idea simply because some of the moderators/admins in charge are awful at it and will cause an incredible amount of backlash with their unlimited ban hammers.

"I didn't pay to get banned by a dumbass! Moderators must be elected and administrators must be held accountable for their poor decisions! As a shareholder, I cannot condone the misuse of my investments! ... Etc"

Requiring a fee wont stop anything. Look at old school Rubescape, botters pay monthly for their bot farms on the member's worlds.

I think payment is a really good idea for deterring bots, and making people behave better too. It does leave a paper trail to your real identity though (unless instances are willing to accept cash or crypto), which many people would rather avoid.

A fixed fee is also relative to who pays it. To a company who only care about jacking up their share price it could be nothing, but for a person working in a very exploitative economy it could be several days wages. I would say that it's the poorest people whose voices are most suppressed, so I wouldn't want to be part of that problem.

Now, having said that, if there was a mechanism for distributing back the surplus from signups after running costs to good users of the instance I would be in favour of it. Per-user instance costs are very low on medium and large instances, so most users would get most of their money back, and perhaps even make money. Making it a cooperative would mean low costs for genuine users and incentivise participation in making it a good instance. If OP wants to do something like that let me know - I'd love to get involved!

Thanks, the third option is what I had in mind but I had completely missed the automated banning feature, reviewing manually each time probably wouldn't scale well.

I would recommend the invite tracking. Keep track of who invites bad actors, even higher order. If your friend's friends suck, that should reflect a little on you. If no one has vouched for you, then you're less trusted.