Great taxonomy, clarifies many issues and possibilities that are widely conflated and misunderstood!

For a complementary perspective that is also very relevant, consider the model of online conversation in From Freedom of Speech and Reach to Freedom of Expression and Impression (https://techpolicy.press/from-freedom-of-speech-and-reach-to-freedom-of-expression-and-impression/).

The forgotten (and now threatened) right of Freedom of Impression affects all four of the models in Ethan’s taxonomy with regard to their different paradigms for reach and control -- and how the three different levers of mediation and who controls them apply (censorship, friction, and selectivity, as shown in my second diagram).

In the spirit of that Though as a Social Process cycle, I will be thinking further about how Ethan’s taxonomy can add richness to my framing of Thought => Expression => Social Mediation => Impression => further thought…

Ethan’s taxonomy – and the “pluriverse,” “a complex world of interoperable social networks where one can choose a technical architecture and governance structure appropriate to a community’s needs” -- also ties in with to the ideas of interlinked webs of digital “hypersquares” (or “hyperspheres”/”hyperspaces”) suggested in Community and Content Moderation in the Digital Public Hypersquare (https://techpolicy.press/community-and-content-moderation-in-the-digital-public-hypersquare/, with Chris Riley).

Expand full comment

I feel that this taxonomy - and the major tech platforms themselves - would benefit from the classical legal distinction between 'malum in se' and 'malum in prohibitum', or 'bad in itself' and 'bad in prohibition'.

Certain things can be understood as bad in themselves because wherever these rules are discarded or not enforced sufficiently those people subject to the rules, assuming they wish to maintain the integrity of the group or community, will come back to those precise rules irregardless of local context. Murder, for example, is fairly universal for societies of any scale. Racism is also probably malum in se, since it is explicitly an act against integration and shared or aligned purposes.

But other rules are contextually sensitive, and would not automatically be the rules that any community would come to. For example, you might say the word 'idiot' is forbidden, given that it is racist (associating people whose native language is not English with stupidity), and racism would seem to be a fairly low level thing to forbid. But since it is a word its meaning may have evolved, and as such the people who use it may not have any knowledge of its roots. It is not clear that every community whose aim is to come into confluence would forbid it.

A "big room" social network can separate communities into 'spaces' based upon topics or interests, or value systems, and therefore set 'ground-rules' (malum in se), while those spaces layer their own local contextual rules on top.

By nesting these spaces such that all of the content in the niches also gets published into the shared spaces that contain them, where those who achieve the most amplification are the users who obtain reputation by peer endorsement from multiple orthogonal communities (bridge builders), it is plausible to build an architecture with all of the benefits of federated and 'big room' social networks, without the major drawbacks of either.

It is also important that the 'ground rules', central to which is the core algorithm, is open sourced and totally transparent about how content is amplified or suppressed wherever it gets published.

Expand full comment

I like the taxonomy and argument for diversity. However I do think that these new platforms need scale and simplicity where an average unmotivated person can ask simple questions like ‘what are people saying about x? What did famous person y do? How do I follow everything that z does?” To that end I think we should try to convince as many of these systems to adopt a compatible protocol (eg Activitypub extended) so that they can all prosper without solving the scale and end user simplicity problem individually.

Expand full comment