A deep dive into X's Community Notes, including how the algorithm works, whether the feature actually fights polarization, and its use in high-stakes situations
Special thanks to Dennis Pourteaux and Jay Baxter for feedback and review. — The last two years of Twitter X have been tumultuous, to say the least.
Vitalik Buterin's websiteVitalik Buterin
Context & Ripple Effects
This analysis establishes a baseline for judging Community Notes as a platform-governance system: not merely whether notes appear, but how its ranking design handles disagreement and consequential claims. That baseline matters as later coverage tests the mechanism’s limits, including findings that many election-falsehood notes did not become visible to users in a review of note visibility on election misinformation.
The piece makes the trade-off in X’s system more legible: a note’s usefulness depends on reaching sufficient agreement across differing viewpoints, rather than simply attracting a large volume of raters.
For users evaluating high-stakes posts, it shifts attention from the existence of a contributor response to whether that response can clear the system’s publication threshold.
Second-order effects
Efforts to expand Community Notes toward agreement signals will face the same core constraint: consensus-oriented ranking can improve cross-group legitimacy but may leave disputed claims without a displayed correction.
Platforms adopting or adapting the model must treat contributor supply, rating behavior, and publication thresholds as product-design choices, not as a plug-in substitute for moderation.
Third-order effects
If peer-rated context becomes a standard layer of platform governance, moderation will increasingly be shaped by transparent ranking rules and contributor incentives alongside centrally enforced policies.
The next structural test is whether human validation remains viable as contribution becomes more automated; X’s plan to have AI agents propose notes while humans rate them makes that division of labor explicit in its proposed AI-assisted notes workflow.
The trend: Community Notes is one data point in the shift from centralized content enforcement toward algorithmically ranked, contributor-produced context—with effectiveness determined as much by distribution rules as by the notes themselves.
in the “what's the usecase for web3” conversations, a recurring misunderstanding I've found is around the definition of web3 for some its speculative tokens, for others its credibly neutral infrastructure. Community Notes are a great example of the latter [image]
Great analysis on @CommunityNotes (previously Bird Watch) by Vitalik. The Community Notes project is an excellent example of decentralized fact-checking. Except for Incentivization and breadth beyond the Twitter/X platform, Community Notes carry the majority of the crypto...
.@VitalikButerin: “The goal is to remind people that multiple perspectives exist, that certain kinds of posts that look convincing and engaging in isolation are actually quite incorrect, and you, yes you, can often go do a basic internet search to verify that it's incorrect.”
I called Vitalik an alien to his face. He wasn't human. So far ahead of me. Invented Ethereum. Reading this I am taken back to that interview. His brain just has so much information capacity mine doesn't have. Reading his work is like listening to Stevie Wonder perform.
Reminder that the Community Notes algorithm code and data are totally open source, so you— yes, you— can run an analysis just like @VitalikButerin did here to understand why any particular Community Note is showing or not. [image]
good technical overview of how Community Notes works on @X which imo is by far the best thing which has happened / been scaled up on this site under the Musk ownership regime
What a stunning writer : “an economist's algorithm and an engineer's algorithm. An economist's algorithm, at its best, values being simple, being reasonably easy to analyze, and having clear mathematical properties that show why it's optimal (or least-bad) for the task that it's.…
@VitalikButerin It's an interesting mechanism and research topic. It's not a truth oracle. My view is that it shouldn't play any official role in a town square, and in particular not a role that is dubiously equated with fact checking and granted the privilege of defacing posts m…
Very good write-up on #CommunityNotes, but you forgot the part where Elon spreads disinformation then deletes the very fact checking system HE put in place so HIS tweet can't be fact checked.
@JohnSmi37529503 @VitalikButerin Neither society nor a social platform should have any specially anointed “fact checker”, because the mere existence of such status will corrupt it. Everyone must decide what's true themselves, and decide which among many sources claiming fact chec…
Excellent analysis of Community Notes and its role in supporting consensus-based fact checking, rather than the “normal” authoritative fact checking approach.
I wrote a series of rating systems articles with @Appelcline starting in 2015, where we tested our ideas with real code & community through 2017. So I appreciate this quote from @VitalikButerin's deep dive into Twitter's latest rating system: “Engineer's algorithms are pragmatic.…
As one who used to fact-check, Vitalik did excellent work here. Some technical jargon aside, particularly as it relates to the instantiation mechanism of crypto, Community Notes—despite its criticisms—is largely beneficial and can be tweaked for improvement.
A good reminder that @CommunityNotes code and data is open source, making it possible to do public analyses of its performance and attributes, like this one today from @VitalikButerin 👇 [image]