RFC 9518 – What can internet standards do about centralisation?
Centralized internet services dominate because they solve real user problems cheaply and conveniently, from easy content publishing to integrated moderation and addictive recommendation algorithms. Commenters debate whether technical standards, regulation, or new decentralized architectures (like IPv6-based end‑to‑end connectivity, federated protocols, or blockchain-style DNS) can realistically counter this trend when most users neither want to run infrastructure nor prioritize privacy or control. Many conclude that decentralization will only gain ground if it can match centralized platforms on usability and features while giving builders better incentives and tools, especially for interoperability and moderation.
Why centralization dominates
- Centralized services solve real problems: easy publishing, search, analytics, and “always-on” hosting that home connections and tools rarely provide.
- Early web tools for authoring (Composer, FrontPage) existed but were too generic and cumbersome; users preferred hosted services.
- Asymmetric consumer broadband and NAT made running servers at home impractical, pushing people to cloud platforms.
UX, discovery, and “worse is better” dynamics
- Decentralized alternatives can be “good enough” (e.g., fediverse-style services), but users often prefer algorithmic feeds optimized for stickiness over agency.
- Search, recommendation, and popularity-based algorithms amplify centralization via feedback loops and Pareto distributions.
- Some propose injecting random low-popularity results or opposing viewpoints; others warn this could harm quality or user satisfaction.
User attitudes and incentives
- Many argue 99% of users don’t care about (de)centralization; they want cheap, simple services and will cluster on local winners (e.g., a dominant messenger per country).
- Others counter that lock-in and account loss (e.g., big-account bans) reveal hidden costs and justify regulation or design changes.
- A theme: decentralization mainly empowers builders (easier competition, interoperability) while users benefit only indirectly.
Infrastructure and self‑hosting hurdles
- Dynamic IPs, CGNAT, lack of static addresses, and weak upstream bandwidth hinder self‑hosting; dynamic DNS and VPS reverse proxies are common workarounds.
- Running email is cited as effectively impossible at scale due to spam-fighting rules and big providers’ blacklists.
Standards, regulation, and interoperability
- Some blame standards bodies for encoding centralizing patterns; others argue interoperability standards are the only path to decentralization.
- Views split on regulation: suggested tools include mandated interoperability/data portability or strong privacy laws that undercut ad-funded giants; concerns remain that governments prefer large, controllable platforms.
DNS, identity, and blockchain proposals
- DNS is criticized as “rented” and controlled by few; defenders call it federated, not centralized.
- Blockchain-based “permanent domains” are proposed; counterarguments focus on cost, validator incentives, and proof-of-stake centralization risks.
- DNSSEC is mentioned as a cryptographic alternative, but also seen as highly centralized at the root.
Moderation as a centralizing force
- Several commenters see moderation as the driver of centralization: quality, safety, and CSAM filtering all favor large, well-resourced operators.
- Federated platforms show moderation can be decentralized but struggle with scale, fragmentation, and admin power over user migration.
- Large centralized platforms also “haven’t solved” moderation; automation at scale brings false positives, lack of appeal, and cultural misfits.
Security, trust, and tooling
- Many users implicitly equate central authority with security (cloud backups, gated browser APIs).
- Suggestions to improve decentralization include wider IPv6 deployment, confidential computing for trust-minimized servers, and native apps that decouple data, compute, and hosting.