Could we Build a User Governed Social Media Platform?
We'll never know the full implications of the decisions made by Nikita at twitter, just because they changed their algo.
That is bullshit.
A brief history
I had ChatGPT summarize the governance models of the major social media platforms.
| Platform / era | Governance in one line |
|---|---|
| Usenet | An anarchic federation: server operators chose what to carry, while individual newsgroups could appoint moderators, with no universal sovereign. |
| Slashdot | A technocratic jury system: editors selected stories, trusted users scored comments, and other users audited those scores through metamoderation. |
| Wikipedia | A consensus republic: communities write and enforce most rules collectively, while the Wikimedia Foundation retains the constitutional and legal backstop. |
| Classic Digg | Algorithm-mediated direct democracy: users voted content up or down, but the company secretly determined how votes translated into power. |
| Platform federalism: Reddit sets the constitution, volunteer moderators rule individual communities, and users vote on visibility—not governance itself. | |
| Facebook / Instagram / Threads | A corporate monarchy with a supreme court: Meta controls policy, product, and algorithms, while an independent board reviews a narrow set of moderation decisions. |
| X / Twitter | An owner-led monarchy with a citizen fact-checking jury: X centrally governs the platform, while Community Notes can add context but cannot enforce the rules. |
| YouTube | An algorithmic bureaucracy: Google writes global rules, machines and employees enforce them, creators may appeal, and users mainly vote with attention. |
| TikTok | An algorithmic benevolent dictatorship: the company centrally controls rules, enforcement, and distribution, with AI doing much of the governing at scale. |
| Discord | Private-city governance: Discord establishes universal law, while each server owner appoints a local government that can impose stricter rules. |
| Mastodon | A federation of city-states: each server makes and enforces its own rules and can cut diplomatic relations with other servers. |
| Lemmy | Federated Reddit: instance administrators govern the jurisdiction, community moderators govern forums, and instances decide whom to federate with. |
We've never had a social media platform governed by users!
What might that look like? How could we do it? Guaranteed, there has never been one bootstrapped by a bigger gov nerd than chainpy.eth and the folks at Agora.
Lets try. What would, should, or could it look like?
What if all revenue and expenses for the platform was onchain? What might be achieved, if alignment was found? Is it even possible? What mechanisms would be needed?
It needs to fund itself, it needs to throw off enough revenue and profit to sustain itself, but other than that - it wouldn't need to think about global domination, pandering to advertisers, or surveillance capitalism.
How would it work?
I have part of an idea here...
Users earn "Signal", for participating
This can be anything like posting, commenting, reading, or moderating.
Users earn "Voice", for voting "correctly", early
The earlier you vote, the more you earn, for being "correct".
Users can "bet" their signal, to earn more "Voice"
At the time of voting on content, you can hold down the vote button, and drain any accrued "Signal"
Voice balances decay
Users need to keep earning "Voice", so we don't get lop-sided kings and queens
Voice balances are used in decision making
...which is TBD, based on some constitution I'm working through.
Thoughts?
Hi all, I'm new here 👋
I'm working on a new decentralized decision-making method that might be an interesting component in the governance of social media. It's not yet published (but almost finished and even tested in a first experiment), but I can give a quick overview. Also you can read this older article (https://medium.com/@carlo.michaelis/evocracy-thinking-democracy-from-scratch-657a0293c070) or check out this presentation at Metagov (https://www.youtube.com/watch?v=DFOWceVToes&t=276s).
Please note that the full process is quite complex and flexible and the following is just a simplified example.
Let's assume there is a community guideline document. Let's assume it to be pre-defined by an interim council or similar. Now in a governance area of the platform each user in the social network can suggest an issue with the guideline. That might be a suggested extension, an update, or just describing a concrete problem which is in conflict with the existing rules.
All users in the network can now access a list of these issues and they can continuously vote on the relevance of these issues. Issues which reach a certain relevance threshold are opened for discussion. Every user can (but don't have to) suggest an individual proposal/opinion on this issue. The proposal collection is open for a certain time window. After that time window, the collection closes and all participants are assigned to small groups. In these small groups they are asked to meet once (self-organized for e.g. 30min or 1h) to discuss their opinions/proposals and write down a common solution in a collaborative document. They also elect a representative for their small group (i.e. the most engaged/skilled member of the group with respect to the particular issue).
The group representatives are again grouped to small groups and asked to integrate their previous groups' ideas again in a short meeting. Again they agree for a new representative. This process repeats until a single group is left. This group suggests a final change on the community guideline (e.g., a git commit). All original participants cast a vote if they agree with the solution from the final group. If approved, the community guideline is automatically updated.
Advantages are:
I'm working my way through this, but this is cool so far. I'll need a bit of time to write a full reply, but before I do that, I must know...
Did you write this, in 2023?
edit: Oh - ChatGPT launched in 2022. Wow, time flies.
I think the idea was even from before ChatGPT (just using some summarization algorithms from the "old" machine learning time), but with today's LLMs is of course much easier.
Perhaps important to mention: The Evocracy process does not necessarily require any AI (in fact, the first experiment was already successful without using any AI assistance). Evocracy can actually even be applied without digital tools at all, just with pen & paper. It's really "just" a collective intelligence method. However, digital tools (API, web apps, blockchain, AI) add of course significant accessibility, scalability, efficiency and trust. So usually I frame it in this setting.
Hi there! GM.
I love the question, and the “governance in one line” table is a great way to open this conversation. My only caution is that these labels can flatten very different layers of governance. Mastodon is federated, but an instance may still be controlled by one administrator; Reddit delegates moderation while retaining authority over the platform; and cases like Wikipedia or Social.coop complicate the claim that user-governed social media has never existed.
I might frame the gap more narrowly: we have never seen a genuinely user-owned and user-governed social network achieve mass-market network effects.
By the way, network effects may also be the hardest part of the challenge. It could be less about designing the perfect governance model from scratch and more about helping existing efforts such as Farcaster, Mastodon, Bluesky, and Nostr combine their strengths and reach enough critical mass for migration to become real (and from there, refine the governance model based on actual operations.).
User governance also should not require everyone to behave like a delegate: settings, surveys, usage choices, and feedback already provide valuable signals, while smaller groups with greater commitment and alignment can take on representative responsibility.
A better classification
Instead of assigning each platform a single political regime, I would examine six layers:
This produces clearer families:
It also reveals why putting revenue and expenses onchain would help with financial accountability, but would not by itself create user governance. The deeper questions are who appoints operators, controls ranking, changes policies, resolves disputes, and can amend the governance system.
Thanks for the thoughtful reply!
You raise a very valid point.
I think I'm most interested in figuring out how to release control in terms of:
I'm very open to innovating on all those fronts. If I could find the magic mechanism that felt credible and defendable, without adding cost, or creating a security, then I'd run with it!
Hi there. I’d be very interested in helping think through that co-design. :)
My instinct is that the credible path is progressive rather than trying to do everything at once:
So the transfer takes place by domain and through milestones
The key questions seem to be:
Who should we bring into the conversation to answer these questions?
Awesome! Lets do it!
All of these are great questions that need answers. Lets figure it out together. I can write a draft of my thoughts on the answers, but would love to hear how you'd answer them.