Balaji Srinivasan’s Network School has moved quickly after acquiring the decentralized social platform Farcaster, rebranding it as the Network State Protocol and introducing a governance regime that almost immediately restricted several accounts to read-only. The changes mark the second ownership shake-up for Farcaster this year and a sharp new experiment in how a crypto-native social network balances open protocols with centralized rule enforcement.

The acquisition was announced on Thursday. Network School said it had acquired Farcaster from Neynar, the infrastructure provider that had taken over the platform earlier in 2026. The same day, multiple users said on X that their Farcaster accounts had been limited to read-only access because of previous posts that violated the new rules. A former Farcaster engineer was reportedly among those affected, according to screenshots circulating online.

Key facts at a glance

• Network School, founded by Balaji Srinivasan, acquired Farcaster from Neynar and renamed it the Network State Protocol.
• The new operators restricted several accounts to read-only on the day of the announcement, citing past posts that broke new rules.
• Srinivasan described the project as “the Singapore of social media,” a positive-sum community of techno-optimists governed by a shared code of conduct.
• He compared the moderation approach to El Salvador’s crackdown on repeat offenders, saying a small fraction of users had been “moderated out” to stop random attacks.
• Users must opt into a “social smart contract” that authorizes the core team to set rules; the accompanying “Network Code” classifies violations as digital felonies, misdemeanors or infractions.
• The rules apply retroactively to past posts. Penalties range from warnings and suspensions to bans.
• The underlying Farcaster protocol remains open. Users who decline the contract get read-only access to the main app but keep their keys, funds, followers and post history and can move to another client.
• Srinivasan said humans and AI reviewed millions of posts, and an appeals process exists.
• Farcaster co-founder Dan Romero said the platform had 250,000 monthly active users in December 2025.
• The acquisition follows Malaysian authorities revoking Network School’s Forest City business licence and ordering it to cease operations; Srinivasan later announced a campus agreement in Kazakhstan.

The Singapore metaphor and the law-and-order pitch

In a post on Farcaster, Srinivasan framed the bans as part of a broader plan to turn the reference app into “the Singapore of social media.” The metaphor is deliberate. Singapore is often cited as a model of high-trust, rules-based governance, public order and economic dynamism. It is also known for strict speech and public-order laws, heavy fines and a state that does not hesitate to intervene when it believes social stability is at stake. Srinivasan appears to want both sides of that image: a clean, optimistic, productive online community and a firm hand against behavior he considers abusive or corrosive.

“In the pre-Internet world, it was simply unheard of to run up to a stranger and start cursing at them,” he wrote. “We’re restoring that offline norm online.” He added that the team had “moderated out” a small fraction of repeat offenders who had been muted or blocked by many accounts, arguing that this would stop random attacks within the Farcaster reference client. He likened the approach to President Nayib Bukele’s security crackdown in El Salvador, where a small fraction of repeat offenders were jailed in an effort to eliminate crime. The comparison is politically charged, but it captures the underlying logic: identify a narrow group of persistent bad actors, remove them quickly, and let the rest of the community enjoy a safer environment.

The pitch is not only punitive. Network School says the Network Code rewards “pro-social behavior,” including teaching others, sharing projects and disagreeing civilly. Those rewards are not yet specified, but the ambition is clear: create a visible incentive structure that encourages the kind of high-signal, techno-optimist conversation that Srinivasan believes can make the network valuable. In this vision, moderation is not just a cost center or a censorship function. It is a design tool for shaping culture.

Retroactive enforcement triggers backlash

The immediate backlash focused less on the goal than on the method. Several users said their accounts were restricted because of posts they had made before the new rules existed. For critics, retroactive enforcement is a red flag. It suggests that the new operators are applying a fresh code of conduct to old conversations, potentially punishing people for behavior that was once accepted or at least tolerated. That is especially sensitive on a platform whose identity is tied to decentralization, censorship resistance and user ownership.

Kunal Gandhi, a former India ecosystem lead at Eigen Labs, criticized the move. “I know most of these folks. They shaped Farcaster with their quirks and wit,” he said. “Obliterating accounts from the get-go for past posts is pretty silly. Goes against the ethos of decentralized social.” His comment captures the core tension: a decentralized social network is supposed to be pluralistic, with many clients and communities setting their own norms. A reference client can set rules, but when the new owner of the main client begins restricting accounts for old posts, users reasonably ask whether the network is still decentralized in practice or only in architecture.

Srinivasan said both humans and AI were used to review millions of posts to determine which accounts to ban, and he noted that an appeals process is in place. That process may address individual cases, but it does not resolve the broader governance question. Who decides what counts as a digital felony? How much discretion sits with the core team? How are appeals judged? And what happens when users disagree with the code but do not want to leave the main app? Those questions are likely to follow the project as it scales.

How the social smart contract and Network Code work

Under Network School’s vision, users opt into a “social smart contract” that authorizes the core team to set the app’s rules. The accompanying Network Code classifies violations as digital felonies, misdemeanors or infractions, with penalties ranging from warnings and suspensions to bans. The rules apply to past posts as well as new ones, which is why the first wave of enforcement landed so quickly. The code also promises rewards for constructive behavior, though it does not specify what those rewards entail.

Network School says an open-weight AI model will apply the rules, with some manual judgment calls, and that it eventually wants to warn users of potential violations before they post. That pre-post warning system, if it works, could shift moderation from after-the-fact punishment toward real-time guidance. But it also raises familiar concerns about automated moderation: false positives, opaque model behavior, cultural bias and the difficulty of distinguishing spirited disagreement from harassment. The use of an open-weight model may make the system more transparent than a proprietary black box, but it does not eliminate the need for human oversight or clear appeals.

The contract model is an attempt to make governance explicit. Instead of pretending that a platform has no rules, Network School wants users to affirmatively accept a code and a process. In theory, that creates consent. In practice, consent is complicated when the main client is the most visible gateway to the network. Users who decline the contract will have read-only access to the main app but retain their keys, funds, followers and post history, allowing them to move to another client. Srinivasan described this ability to join, leave or build competing clients as “decentralized democracy.”

Open protocol, closed client?

The underlying Farcaster protocol will remain open, according to the announcement. That distinction matters. Farcaster is not a single app; it is a protocol with multiple clients and an ecosystem of developers. A user’s identity, social graph and posts can, in principle, travel across clients. If the main reference app becomes stricter, users can vote with their feet and use another interface. That is the promise of decentralized social media: no single operator should be able to erase a person’s presence entirely.

But the practical reality is messier. Network effects concentrate attention in the largest client. Followers, discovery algorithms, developer tools and cultural norms often cluster around the main app. A user who moves to a smaller client may keep their data but lose reach, community and convenience. So even if the protocol remains open, the governance of the dominant client carries outsized weight. The Network State Protocol may be technically decentralized while its most important front door is governed by a core team with a detailed code of conduct. That is not necessarily hypocrisy, but it is a tension that critics will keep probing.

The approach also reflects a broader shift in crypto social media. Early promises focused on censorship resistance and user sovereignty. As these networks matured, they ran into the same problems as conventional platforms: spam, harassment, moderation disputes, low retention and the challenge of building sustainable businesses. Purely open systems can become chaotic or unpleasant, while heavy-handed moderation can alienate the libertarian and cypherpunk users who were attracted in the first place. Network School is trying to thread that needle by keeping the protocol open and making the main client’s rules explicit, enforceable and opt-in.

From Forest City to Kazakhstan

The Farcaster acquisition is part of a wider effort by Srinivasan to turn Network School into more than an online community. Network School has described ambitions to take physical form through meetups and pop-up events, eventually developing into startup cities and network states. That vision aligns with Srinivasan’s longer-running advocacy for “network states”: cloud-first communities that begin online, build their own institutions and eventually acquire physical territory or recognition. The social network is meant to be a digital gathering place and governance laboratory for that project.

The timing is notable. The acquisition comes months after Malaysian authorities revoked Network School’s Forest City business licence and ordered it to cease operations, citing alleged licensing and premises-use violations. Srinivasan subsequently announced an agreement to establish a campus in Kazakhstan. Those developments suggest that Network School is learning how difficult it is to build new jurisdictions in practice. Operating a social platform may be easier than operating a physical campus, but it comes with its own regulatory, reputational and governance risks.

Farcaster’s own history adds context. Co-founder Dan Romero said the platform had 250,000 monthly active users in December 2025. That is a meaningful base for a crypto-native social network, but it is small compared with mainstream platforms. The platform has changed hands twice this year, first through Neynar’s takeover and now through Network School’s acquisition. Each transition has raised questions about direction, sustainability and the balance between decentralization and operational control. The reported plan to return $180 million to investors following the Neynar takeover underscores how much financial and strategic repositioning has surrounded the platform.

The bigger test for decentralized social media

The first wave of bans will be an early test of whether Srinivasan’s model can win legitimacy. If the appeals process is fast, transparent and consistent, Network School may persuade users that the rules are not arbitrary. If enforcement looks selective, retroactive or driven by personal animus, the backlash will likely grow. The use of AI to review millions of posts may help scale moderation, but it also makes the process harder to scrutinize. Open-weight models can be inspected, yet the policy choices behind them remain human decisions.

For the wider decentralized social ecosystem, the experiment raises a fundamental question: can a network be open at the protocol layer and governed firmly at the client layer without losing the culture that made it attractive? Srinivasan’s answer is that users can opt in, appeal, leave or build competing clients. That is a coherent theory of decentralized democracy, but it depends on users having real alternatives and on the main client not becoming a de facto monopoly. The next few months will show whether Farcaster’s users accept the trade-off or whether the “Singapore of social media” becomes a cautionary tale about the limits of top-down culture building in an open network.

Source:Cointelegraph News