[than3]

I'd like to add that looking at sybil attacks as just a network attack would be a mistake.

These type of attacks apply broadly at many different levels, for example the minimum thresholds for amplification/de-amplification can be significantly lower than the 51% that's normally attributed by definition to these type of attacks.

Any decentralized moderation system would run into issues with a sybil attack distorting reflected appraisal within a community. Some benefits of such an attack might be dropping the post out of view/delisting it, using generative ai to create harassing/malign responses attacking volunteer psychology (where they stop contributing/volunteering when it costs them).

Useless/mindless generative chatter that is akin to jamming a communications channel, but indistinguishable by most people which in the absence of a clear signal causes isolation (thanks AI).

Some of these challenges also don't have a good answer because they are fundamental problems where a solution would break automata theory/determinism. So you are left with only a bad, potential liability from an approximate solution.

Most propaganda and political warfare focuses on attacking various parts of communications with the intent to either isolate, or destroy/debase personal identity, which is largely determined by reflected appraisal.

Once isolated, with their belief system destabilized it makes people vulnerable and does dangerous/weird things to people's psychology. This has been known for quite some time though hasn't gotten much attention outside certain academic circles, and by quite some time I mean roughly generally known between the 1920s-50s.

The main driving component was formalized under the Sapir-Whorf Hypothesis (1929), though it was in common use by governments prior to that.

Since most social media already has these issues (but they don't communicate about them in public regularly if at all), internally they do recognize these issues and take a Signals and Systems approach, but the result is as you've seen.

If you cant exceed parity vs. competitive solutions you wont be able to increase marketshare/market penetration.