4 Prevalent Themesin the ATProto Discussion
| # | Theme | Key Take‑away | Illustrative Quote |
|---|---|---|---|
| 1 | “Instances” are a Mastodon‑centric notion – asking “where are the Bluesky instances?” misses the architecture. | ATProto separates hosting (PDS), apps (AppView), and relays; there is no single “instance” to point to. | 1dom: “But where are the Bluesky instances? … a category error.” |
| 2 | Centralization risk tied to Bluesky’s business model – the protocol’s promise is undercut by heavy VC reliance and a single dominant company. | The network can’t truly scale decentralized without a sustainable, non‑VC model. | timbray: “My concern isn’t technology or culture, it’s money. ATProto is existentially dependent on Bluesky PBC, a venture‑funded startup….” |
| 3 | Relays and appviews are cheap optimisations, not essential infrastructure – they can be run for ≈ $30 /mo or pooled, and many apps skip them entirely. | Decentralisation is achieved via cheap, interchangeable relays; they’re not a bottleneck. | hooverd: “Mostly because having a big centralized firehose relay is less decentralized than people want to admit.” |
| 4 | Data ownership & alternative services prove viable migration paths – multiple independent PDSes/AppViews (e.g., Blacksky, Eurosky) and a growing directory of PDSes show you can host, move, and diversify. | Users can self‑host PDSes, switch providers, and still participate across the ATmosphere. | EnglishMobster: “There’s just over 3,000 PDSes … you can self‑host your PDS to truly own your data up and down the stack.” |
All quotes are taken verbatim from the HN thread (HTML entities retained where originally present).