The Future of Moderation: From Censorship to Curation

Decouples the social graph and content from the client. The problem is platform capture and arbitrary de-platforming. The solution is an open social protocol where clients compete on curation algorithms and users own their identity and social graph.

• Key Benefit: Client diversity (e.g., Yup, Karma3) prevents single-point censorship.
• Key Benefit: Onchain storage via Farcaster Frames enables native app integration.

Treats user actions (follows, collects, mirrors) as ownable, tradable NFTs. The problem is that reputation is siloed and non-portable. The solution is a graph of social capital that any app can permissionlessly read and write to, enabling reputation-based curation.

• Key Benefit: Monetization via collect modules directly rewards creators.
• Key Benefit: Developers can build custom feed algorithms atop a shared user base.

Building trust graphs for onchain discovery. The problem is that spam and low-quality content dominate due to lack of Sybil-resistant reputation. The solution is OpenRank, a decentralized scoring protocol for ranking anything (profiles, content, DAOs) based on peer attestations.

• Key Benefit: Enables Google PageRank for Web3 (used by Galxe, CyberConnect).
• Key Benefit: Credible neutrality; no single entity controls the ranking algorithm.

Current moderation relies on centralized teams or naive token-voting, both of which fail at web-scale. The solution is automated, incentive-driven curation markets.

• Key Insight: Projects like Agora and UMA's oSnap use optimistic oracles for subjective data disputes.
• Key Insight: ERC-7281 (xERC-20) illustrates a model for decentralized, competitive security councils.

Replaces 'one-token-one-vote' with stake-for-access models. The problem is governance attacks and low-quality participation. The solution is systems where curation power is derived from staked economic value that can be slashed for malice.

• Key Benefit: Aligns cost-of-attack with potential reward (see EigenLayer's cryptoeconomic security).
• Key Benefit: Enables specialized curators (e.g., security experts for code reviews) to monetize their judgment.

AI agents will be the primary users and content generators. The problem is LLM hallucination and agent spam. The solution is onchain reputation and proof-of-work for AI, where agents build verifiable track records.

• Key Insight: Projects like Ritual and Fetch.ai are building inference networks with staking mechanisms.
• Key Insight: Bittensor's subnet model creates competitive markets for machine intelligence, a form of curation.

Curation markets like Farcaster Frames or Lens Protocol rely on token-weighted voting, but low-cost identity creation enables Sybil attacks to game reputation. The result is a curation layer vulnerable to the same plutocracy and manipulation it aims to solve.

• Attack Cost: As low as the gas fee for creating a new wallet.
• Mitigation Failure: Proof-of-stake social graphs can be dominated by whales or VC funds.

Decentralized Autonomous Organizations (DAOs) managing curation, like Aragon or MolochDAO forks, become legal targets. Regulators (e.g., SEC, FCA) can enforce rules by targeting identifiable founders or service providers (e.g., Infura, Alchemy), forcing protocol-level censorship.

• Precedent: Tornado Cash sanctions demonstrate infrastructure-level attack vectors.
• Centralization Pressure: Leads to re-centralization around compliant, jurisdiction-friendly nodes.

Competing curation standards (e.g., Farcaster vs. Lens vs. DeSo) and client-specific filters create walled gardens of reputation. Users face fragmented social graphs and portability nightmares, killing network effects and reverting to centralized platform convenience.

• Interoperability Gap: Bridges for social data (e.g., Ceramic Network) add latency and complexity.
• Adoption Barrier: Mainstream users reject managing multiple profiles and reputations.

Automated curation via ML models (e.g., OpenAI-powered filters) is vulnerable to adversarial prompts and data poisoning. Attackers can subtly manipulate content to bypass filters or trigger false positives, making automated trust systems unreliable and expensive to maintain.

• Ongoing Cost: Requires constant model retraining and human oversight.
• Censorship Risk: Over-correction leads to false positives, stifling legitimate speech.

Token-incentivized moderation (e.g., Steemit-like models) rewards engagement, not truth or quality. This leads to outrage optimization, brigading, and financialized spam, degrading discourse quality. Curators become profit-maximizers, not community stewards.

• Metric Gaming: Rewards for upvotes/downvotes incentivize collusion via vote-rings.
• Quality Erosion: High-signal users are driven out by financialized noise.

Blockchain-based denylists (e.g., for NFTs or addresses) are permanent. A false positive or malicious listing on a protocol like OpenSea's former filter or a DAO-curated list can permanently deplatform an entity with no recourse, creating a more rigid censorship apparatus than centralized systems.

• Irreversible Damage: No 'right to be forgotten' or appeal process on-chain.
• Governance Attack: A single malicious proposal can blacklist legitimate actors.