Regulating free speech on Twitter

Inspired by a recent whatsapp group chat, a few idle thoughts on how I’d approach free speech on Twitter / what I would do if I were Musk.

Very much a first draft with plenty of room for improvement and no doubt many instances of short-sighted idiocy. But a fun exercise for 20 minutes.

1. We will allow freedom of speech
2. Accounts with an ID check will get a blue tick
3. Anonymous accounts will be allowed to enable functions such as whistleblowing or reportage.
   1. Accounts will go through multiple stages and checks to determine trust.
   2. Those which fail will be removed and added to a publicly available list, together with a summary of the reasons for removal.
   3. Those which earn trust will given a green tick. Trust will be hard to earn, and based on merit alone. In certain cases (such as whistleblower inside the Kremlin) we may hide the names of individual followers of an account to reduce the potential risks to everyone’s safety.
   4. Interactions (RT / likes) from unverified accounts and low-trust accounts will be tracked as part of the ‘trusting’ algorithm, but will not count towards any publicly displayed totals.
4. Business accounts will need to be verified.
5. Bots will be allowed, but will cost at least $1 per month.
6. People will be able to filter to choose what they see from a menu of: verified personal accounts, verified business accounts, high-trust anonymous accounts, verified bots, high-trust bots, everything else.
7. The “free speech” laws of Twitter will be these: [something based on a blend of pretty libertarian country’s laws]
8. Each account will have a ‘legal breaches’ score. This will be publicly visible. Unless there is abundant evidence, “I got hacked” will not be an excuse.
9. We will work closely with law enforcement agencies across the world if accounts breach our free speech rules. 10. High-follower accounts with a legal breaches score > 0 will have to pay twitter $250,000 per month for each breach for the duration of the existence of the account. If the law enforcement of your country agrees there has been a breach, your account will be banned for 3 years. If you do not close the account, the $250,000 will still be payable. The fees will go towards the cost of fact checkers, free speech support around the world, and other R&D to increase trust.

The ones I’m least happy with:

• the misaligned incentives in 10, and maybe 8. Both are intended as hypothecated disincentives, not profit centres.
• probably more to be done around spam accounts / preventing automation but I don’t know enough about the troll farm ops to regulate with any confidence.
• quite how the high-trust / peer-review process happens in 3.3