# Common Knowledge Machines

- Channel: [Devcon](https://streameth.org/devcon)
- Date: 2025-10-07
- Duration: 07:49
- Watch: https://streameth.org/watch/yt-2wRdNrWMcuc
- YouTube: https://www.youtube.com/watch?v=2wRdNrWMcuc

## Description

Common knowledge is a precondition for collective action. Yet, increasing polarization in information ecosystems risks undermining common knowledge formation. This talk introduces Community Posts, a mechanism that leverages diversification and zero-knowledge proofs to help people identify divides, bridge them and find common ground, fostering greater common knowledge in social networks.

## Transcript

[Music] all right so let me get this on the right slide here common knowledge so what is common knowledge why does it matter so common knowledge is not just that you know I'm standing here it's that you know that I know that you know at infinitum it's a very different thing than just knowing that I here and a classic example of this is the Romanian revolution in 1989 for everyone for years everyone knew that this dictator Chia chesu was a tyrant um but nothing ever happened and it wasn't until actually uh a simple act during one of his public speeches a live broadcast whistle that cascaded into many whistles that a revolution began and the key wasn't just that everybody saw that key was not just that everybody was unhappy but they saw that everybody else was unhappy so we've had similar moments in American history I'm American so that's my history um but my contention is actually common knowledge is getting harder to get off the ground and that's because of increasing Pol political polarization and while political polarization predates the internet it certainly has been accelerated in the digital age and why is this so bad well if we can't find Common Ground if we can't have shared beliefs then Collective action becomes really hard and Collective action is the stuff for political movements and also for government and legitimate governance and so on the international stage this is really important so unions stay unified and can differentiate their allies from their adversaries and Marshall support and not collapse into the chaos of competing narratives which we seem to be trending towards and also common knowledge lets us solve higher forms of cooperation problem for example like climate change that requires consensus and agreement about the so since this is Devcon and an ethereum conference it's worth acknowledging actually that ethereum itself is an example of a common knowledge machine it produces or it seeks to actually produce an immutable Ledger about shared Global State um another obvious example of common knowledge machines are things like blue sky forecast or lens or X these are the functional equivalent of public squares but actually online but of course here common knowledge rarely ever gets off the ground and instead we just have divisions and competing narratives about reality and which one is true so today I want to introduce a quick mechanism and outline on the principles um to sort of nudge your thoughts and you can read more um if you want in my writings um so the idea is called Community posts and everybody here who participates in some social media platform has some set of followers and the idea is instead of blasting a post to all of your followers you first stress test a post across different groups of follow followers who don't necessarily talk listen or agree with one another to gauge what resonates before it's widely shared and so here's how it works um if you look at your followers your set of followers they actually all have differences in Behavior measured by their likes their reposts who follows them and who they follow their connections and you can aggregate this information about observe Behavior using things like ZK snarks and put people in clusters as represented by circles here and the idea is that you pick a subset of uncorrelated followers on your social graph people who behave differently on the platform balance for reach with a minimum diversification threshold now these polarity subsets aren't an echo chamber what they are is a sampling or random sampling of a subset of your fans followers balanced across different perspectives and there'll be multiple subsets to choose from for example you can handpick key anchors or key thought leaders or experts and then diversify against them in different combinations to fill the remaining set now once you pick your subset you send that post to the subset with a zero knowledge proof confirming that recipients follow you without revealing your identity now some might disagree with the post and send ZK feedback to you but some might actually take it further and pass it on to subsets polarity subsets of their followers and these rounds repeat forming a tree of polarity subsets across the social graph which essentially pushes your post across very diverse parts of the social graph allowing you to grage cross tribal appeal at every step in every layer through repost rates and feedback and of course the people who receive this get a notification ZK posted by the people you follow or and people you follow boosted this post and so on now the final step is the post goes public once it gains enough support across it reaches a diversification threshold and it's passed its social stress test and the post goes public and you can merge into your open feed and anyone who has reposted it can post it to all their followers and so on and then you know any remaining divides as represented by Red here on these red uh RED parts of the social graph you can use Community notes which we're all familiar with to bridge um and add context and this is really lwh hanging fruit divides um so anyways this is the summary of the mechanism um I think it works because essentially what you're doing is you're reversing the cycle that happens on social media of context collapse controversy cancellation and capture and these are things where social media leverages divisions to create more divisions and you reverse that cycle with diversification discounts and bridging to restore Common Ground um okay I think that's it for five minutes I can take any of your [Applause] questions that was kind of a complex mechanism for five minutes wants ask a question who who wants to ask a question here please this is amazing what do we do next what's like a Next Step here yeah so there are a lot of um experiments with uh social graphs right now in web 3 um forecaster and lens are an example and so you can leverage like ZK snarks um zero knowledge proofs to actually calculate this information without revealing you know the identities and participants in a social graph and start to enrage you know create a more meaningful parallel attention feed to show okay like what are what are posts where there's wide agreement on and instead of focusing on like the attention and the rage bait that comes from a global attention auction you can focus your attention on okay what are the things where there's agreement and then build on that agreement yeah one um do you envision uh this being applicable to like more popular social media like maybe plugin for Twitter or whatever yeah it could apply to any social media it could apply it's a road map for um making x a common knowledge ma machine I think that's actually was always the aspiration around a Public Square um and the Chao chesu example um it's also applicable to Tik Tok it's t it's applicable to really any social network um and it's just a more uh intelligent uh attention mechanism right that gets us that sort of reverses the cycle of let me um you know hit my dopamine receptors with division towards let me see what I can find common ground with and and build on that okay thank you everybody thank you so much puya