Note: This post is still a bit of a work in progress, as I clean up some wording, grammatical errors, etc. Still, wanted to get this out sooner rather than later.
Core Tenets:
- You get one account – that’s it.
- There is no “one” algorithm
- There are no community moderators, only sherpas
- It’s all attribute tags
- AI acts as a helper, NEVER as a creator
- A centralized platform
Tenant #2: There is no “one” algorithm
Social media algorithms are under intense scrutiny – as mentioned in part one of the “problems” post, people are becoming more skeptical that a sufficiently complicated algorithm that ranks content based on criteria it determines to be “good” vs “bad” differs from being a “publisher”, which would revoke the Section 230 protections that most social media platforms operate under today.
If that’s the case, then, why not build a platform that gets ahead of this? If we assume that platforms that implement one algorithm can be deemed publishers, why not have a platform that is algorithm-agnostic?
For example, if you signed up as a user, what if you could:
- BYOA – Bring your own algorithm, a way you could decide how you wanted to curate content by importing or creating an algorithm of your own that you wish to use to sort and filter content
- Get suggested platform-specific algorithms
- E.g. algorithms built for the platform by the platform that represent common implementations of ranking content (think how Reddit allows you to sort posts by “best” / “new” / “controversial”, only more in depth).
- Get suggested “popular” community algorithms
- More on “community algorithms” later on…
Assuming we could make sure that no one algorithm “dominates” the platform, we could gather from this a lot of benefits:
- We could (much more clearly) argue for Section 230 protections, as we are not in any way trying to act as a publisher, in other words we are not forcing just one (algorithmically) way of thinking.
- People who create content would be more free from Goodhart’s law – if they didn’t feel like they had to optimize to a single platforms algorithm, they’d post content because they wanted to, in the way they wanted to, vs knowing that there’s only one “correct” way to best game the algorithm for success.
- People who consume content could tailor it better to them – now they don’t have to hope the algorithm understands them, they could have a large ability to adapt the platform to fit them, vs the other way around.
- People who cried that they were being “censored” by the platform would hold less weight – if users choose which algorithms define their feed, it’s the users that decide not to show a certain person in their feed, vs the platform themselves.
- E.g. If everyone decides to block Alex Jones, he can’t blame the platform for censoring him.