TruthNostr: A concept of "Truth Network"
Centralized platforms for news dissemination suffer from gatekeeping, bias, and censorship, enabling propaganda and misinformation to thrive. TruthNostr proposes a decentralized solution built on Nostr, a simple, open protocol for signed events relayed by independent servers. By combining anonymous posting, open-source AI validation, and Bitcoin-based zap incentives, TruthNostr creates a self-sustaining ecosystem where truthful information is rewarded and false narratives are penalized. Relays voluntarily share compute power, forming a distributed “validation pool” to ensure scalable, resilient AI analysis. This system empowers users to share and consume credible news without reliance on trusted third parties.
1. Introduction
Information networks often centralize control, allowing manipulation by authorities or bad actors. Nostr, a protocol for cryptographically signed events, offers a decentralized alternative for global communication. TruthNostr extends Nostr to address misinformation by enabling anonymous news posts, scored for credibility by a consensus of open-source AI models running on relays. Economic incentives via zaps align participants’ interests with truth, while a novel compute-sharing mechanism ensures the system’s resilience and scalability. This paper outlines the core concepts and mechanisms of TruthNostr.
2. Core Concepts
2.1 Anonymous News Posting
• Mechanism: Users create Nostr events (kind 1 for short posts, kind 30023 for articles) using pseudonymous public/private key pairs. Posts are tagged (e.g., #truth) for discoverability.
• Anonymity: No real-world identity is required, protecting free speech and whistleblowers. Posts are signed to prevent tampering.
• Spam Prevention: A minimal zap (e.g., 1 satoshi) is required to post, deterring low-effort propaganda.
2.2 Open-Source AI Consensus
• Validation: Lightweight, open-source AI models (e.g., fine-tuned LLMs) run on participating Nostr relays. They analyze posts for factual accuracy, source quality, and neutrality, assigning a credibility score (0–100%).
• Consensus: Multiple relays independently score each post. The final score is a weighted average, published as a signed event (kind 30024: “Truth Score”). Weights reflect relay reputation (based on zap earnings or uptime).
• Transparency: AI models, training data, and scoring criteria are open-source, hosted publicly (e.g., GitHub), and auditable by the community.
2.3 Zap Incentives
• Rewards: High-credibility posts (>80%) earn zaps from readers or automated bots, sent to the poster’s public key via Lightning Network (NIP-57). Relays contributing accurate scores receive micro-zaps from a community pool.
• Penalties: Low-scoring posts (<20%) forfeit submission zaps to a fund supporting relays or governance. Repeat offenders face client-side filters.
• Economic Loop: Readers zap to access premium high-credibility content, funding rewards and sustaining the ecosystem.
2.4 Relay Compute Sharing (“Validation Pool”)
• Function: Relays opt-in to share spare CPU/GPU cycles, forming a distributed compute pool for AI inference, akin to a Bitcoin mining pool but for validation (proposed NIP-89: “Compute Sharing”).
• Operation: Relays advertise compute capacity via signed events (kind 30025). A coordinator (client-side script or bot) shards validation tasks across relays, aggregating results as consensus scores.
• Incentives: Contributing relays earn zaps proportional to compute provided and accuracy. Non-contributors remain standard relays, ensuring network inclusivity.
• Resilience: Distributed inference across 3–5+ relays prevents single points of failure. Federated learning updates AI models collaboratively, adapting to new misinformation tactics.
2.5 Decentralized Governance
• Community Control: Rules for AI criteria, scoring thresholds, and dispute resolution are set via zap-weighted voting (e.g., NIP-20 lists) or a DAO-like event kind.
• Dispute Resolution: Users flag questionable scores, triggering re-validation by additional relays or human bounties funded by zaps.
• Adaptability: Community contributions evolve AI models, ensuring cultural and contextual relevance.
3. System Flow
1. A user posts a news event anonymously on Nostr, paying a 1-satoshi zap.
2. Relays in the validation pool analyze the post using open-source AI, producing independent credibility scores (e.g., 90%, 82%, 88%).
3. A consensus score (e.g., 85%) is published as a signed event, visible in clients.
4. High-scoring posters and relays earn zaps from readers or the community pool.
5. Low-scoring posts lose zaps, discouraging propaganda.
6. Readers access verified content via micro-zaps, funding the system.
4. Benefits
• Censorship Resistance: Relays cannot alter signed events; users switch relays if bias is detected.
• Incentive Alignment: Zaps make truth economically viable, reducing propaganda’s appeal.
• Scalability: Compute sharing leverages existing relay infrastructure, scaling with Nostr’s growth.
• Transparency and Trust: Open-source AI and public scores ensure accountability.
5. Challenges and Mitigations
• AI Bias: Mitigated by diverse, open-source models and community audits.
• Sybil Attacks: Prevented by zap-based posting fees and relay reputation systems.
• Subjectivity: Handled by categorizing posts (fact vs. opinion) with tailored scoring.
• Adoption: Simplified via client integrations (e.g., Damus, Primal) and Lightning wallet on-ramps.
6. Conclusion
TruthNostr harnesses Nostr’s decentralized architecture, zap incentives, and a novel relay compute-sharing function to create a robust, anonymous news network. By rewarding credible information and penalizing falsehoods through open-source AI consensus, it eliminates reliance on centralized gatekeepers. This system empowers users to share and verify news globally, fostering a transparent, truth-driven information ecosystem.
