Federated airspace isn't a technical decision. It's a trust decision.

In 2019 I pivoted Vizzbee from a centralized airspace control plane to a federated one — region-partitioned USS, each service supplier owning a geographic slice, handoff at boundaries via shared protocols. The model was the one ASTM was converging on with F3548 inter-USS. The Remote ID layer was F3411. The technical case was clean.

The technical case wasn't the reason.

Why centralized "obviously" wins on a whiteboard

If you stand at a whiteboard and design airspace coordination from scratch, centralization is the lazy default. One server holds all airspace state. Every drone reports there. Every regulator queries there. Conflicts get resolved in one place with global visibility. Easy.

It's also wrong for any airspace you'd actually want to operate in.

What centralized assumes

Centralization assumes a single party every operator trusts to read and act on their data. In an airspace shared between commercial delivery operators, defense, surveying contractors, foreign equipment vendors, and the local regulator, that party doesn't exist. There is no actor whose incentives align with everyone else's.

Bandwidth and latency arguments for federation are real, but they're downstream. The first-class question is: who reads the central state, and who can be coerced into changing it? In 2019, in India, that question had no clean answer. It still doesn't.

What federation actually buys

Region-partitioned USS lets each service supplier own its own data, write its own deconfliction logic for its slice, comply with its local jurisdiction's rules, and expose only what F3548 says it has to expose. Failure of one USS is local. Compromise of one USS is local. Disagreement with one USS is something you route around, not something that takes the airspace down.

The internet won. Air traffic control hasn't accepted it yet.

This is the same architecture that made the internet work and that the centralized PSTN didn't. Federation is what you build when no one party can be the trust anchor — which, for any airspace serving more than one stakeholder, is always.

The lesson

If you find yourself arguing for centralized UTM on technical grounds, you are arguing the second-order question. Ask the first one: who runs the center, and what stops them from selling, throttling, surveilling, or being subpoenaed in a way that breaks every operator who depends on them?

In 2019 we didn't have a good answer. So we didn't centralize.