He seems to be intending to take Twitter back to the type and scale of content moderation social media firms had pre-2015, which is perhaps not co-incidentally when they achieved their huge levels of growth.

Too many HN readers appear to be trying to convince themselves of this idea: that content moderation is some insanely hard problem. It's not. Social media grew to billions of users with relatively straightforward policies. It only seems that way now because the type of people who are obsessed with "moderation" (read: banning ideas they don't like) also struggle to crystallize a coherent set of beliefs about what speech should and should not be allowed. The absurd false positive and noise rates on Twitter/FB moderation is evidence of this. In place of coherent policies there is an inconsistent and ever-shifting set of unwritten rules that must be reverse engineered by outsiders to even begin to understand the underlying "logic".

Reverting to a classical policy of:

1. No spamming.

2. No engaging in illegal activity (e.g. financial fraud, CP, copyright infringement).

3. Tweets that violate local laws will be hidden in that jurisdiction.

along with the usual boilerplates you find in any 2010-era ToS, would be plenty sufficient and would not in fact pose problems harder than literal rocket science.