2c backlog (research, no priority): decentralized threshold CA #39

New issue

Open

opened 2026-07-06 23:36:44 +00:00 by vxfemboy · 0 comments

vxfemboy commented

2026-07-06 23:36:44 +00:00

(Migrated from github.com)

Context

Milestone 2c (decentralized discovery, spec docs/superpowers/specs/2026-07-06-decentralized-discovery-design.md) uses an offline single CA to sign membership certs + the root set. During brainstorming the user chose the offline-CA model for 2c and asked to backlog the stronger alternative for later research (no priority, after the rest of the roadmap).

Idea

Replace the single offline CA with a decentralized threshold CA: membership signing split across N roots via threshold signatures (t-of-n must co-sign to issue a member cert). No single root (or offline key) can forge membership, and no offline signing step is required — a node could enroll by collecting t partial signatures from the roots online.

Why deferred

Materially more complex than the offline CA: threshold Ed25519 (e.g. FROST) or a BLS threshold scheme, a multi-root enrollment protocol, key-resharing on root membership change. Over-scope for 2c's first cut; the offline CA already delivers "online-node compromise cannot forge membership."

When to revisit

After sub-project #2 (2a/2b/2c) and the higher-priority roadmap items (#9 rekey/PQ, #3 anti-DPI, #10 multi-core, #34 anti-replay) are done. Research-only for now.

## Context Milestone 2c (decentralized discovery, spec `docs/superpowers/specs/2026-07-06-decentralized-discovery-design.md`) uses an **offline single CA** to sign membership certs + the root set. During brainstorming the user chose the offline-CA model for 2c and asked to backlog the stronger alternative for later research (no priority, after the rest of the roadmap). ## Idea Replace the single offline CA with a **decentralized threshold CA**: membership signing split across N roots via threshold signatures (t-of-n must co-sign to issue a member cert). No single root (or offline key) can forge membership, and no offline signing step is required — a node could enroll by collecting t partial signatures from the roots online. ## Why deferred Materially more complex than the offline CA: threshold Ed25519 (e.g. FROST) or a BLS threshold scheme, a multi-root enrollment protocol, key-resharing on root membership change. Over-scope for 2c's first cut; the offline CA already delivers "online-node compromise cannot forge membership." ## When to revisit After sub-project #2 (2a/2b/2c) and the higher-priority roadmap items (#9 rekey/PQ, #3 anti-DPI, #10 multi-core, #34 anti-replay) are done. Research-only for now.