Can we get more decentralised than the Fediverse?

Original Article ↗ Hacker News Discussion ↗

Efforts to build social networks more decentralized than the Fediverse raise hard trade‑offs between architecture, usability, and content discovery. Commenters debate pull‑based models (like RSS/WebSub) versus push‑based protocols (like ActivityPub), noting that truly distributed publishing often still depends on centralized aggregators for search, spam filtering, and finding replies. Many argue that while p2p and self‑hosting preserve autonomy and resist corporate control, most users will continue to gravitate toward centralized platforms that offer convenience, reach, and effective moderation.

Pull vs. push models (RSS, ActivityPub, WebSub)

  • Main critique of pure pull (e.g., RSS): you can’t discover reactions from unknown sources; you only poll what you already know.
  • This makes two-way interaction hard: replies from non-followers are invisible, and new users without followers are effectively muted.
  • Proponents argue WebSub-style hubs and aggregators can layer push-like behavior on top of pull, but that reintroduces central components.
  • ActivityPub is seen as better than RSS for synchronization (supports “since last time” queries, pagination), but still inherits federation complexity.

Content discovery and inevitable centralization

  • Many argue that search/aggregation is inherently centralizing (like Google vs. the web).
  • Even in “maximally decentralized” designs, powerful crawlers/aggregators would likely be few, expensive, and gatekey spam filtering and misinformation control.
  • Some see this as orthogonal to federation: both centralized and federated networks still rely on centralized discovery to scale.

Decentralization vs. usability and adoption

  • Several comments stress most people prioritize convenience and usability over decentralization; self-hosted blogs and personal instances are niche.
  • Running a Mastodon/Matrix/Lemmy server is non-trivial and costly even for technical users, so ecosystems tend to recentralize around big instances.
  • Critics of fediverse UX highlight poor search, discovery, and reliability, and argue fully centralized networks remain superior for most users.
  • Others counter that decentralization matters for resilience, censorship resistance, and reducing corporate control, even if only a minority cares initially.

Alternative protocols and architectures

  • Mentioned candidates: Nostr, Secure Scuttlebutt, Urbit, Aether, Peergos, IPFS, content-centric networking, p2p browsers, and LoRa/Bluetooth-style gossip.
  • Nostr praised for simplicity (JSON+WebSockets), bot/search friendliness, and Lightning payments; cultural downsides noted (e.g., monoculture topics).
  • P2P swarms are envisioned for highly personalized feeds, offline-friendly replication, and escaping ad-driven web platforms, but performance and complexity remain concerns.

Social effects, moderation, and propaganda

  • Mixed views on social media’s value: some see it as crucial for countering propaganda, others as a propaganda amplifier and echo-chamber machine.
  • Strong disagreement over whether friction and reduced virality (as in pull-based or low-discovery systems) are beneficial (limiting trolls/engagement hacking) or harmful (silencing new voices and fact-checkers).
  • Examples like Community Notes, instance-level moderation, and crowdsourced filtering are discussed as partial mitigations with their own trade-offs.