Skip to main content

Threat Model

last updated: 2026-02-27

Scintilla Locate is designed for adversarial environments. The protocol must remain economically rational and dispute-capable under:

  • dishonest provers (spoofed inputs, replay attempts, falsified commitments)
  • dishonest challengers (griefing, false accusations, spam)
  • colluding verifiers / committee capture attempts
  • stake concentration and bribery attempts
  • watcher exploitation and “objective dispute” games
  • economic overload / exposure flooding attacks

Security posture

  1. Assume attacks happen. We build guardrails, not optimism.
  2. Prefer objective disputes. If a claim is wrong under the spec, it should be demonstrably wrong.
  3. Make outcomes replayable. Every dispute and resolution must be reproducible later.
  4. Capital-weighted safety. The protocol throttles exposure based on staked security.

Where this connects to engineering

Threat modeling is not a one-off doc. It is enforced through:

  • deterministic rules in normative specs
  • canonical byte encodings
  • conformance corpora and cross-implementation tests
  • watcher-as-adversary recomputation in CI
  • governance gates for any changes that affect incentives, disputes, or mappings

See also:

  • Economic security: docs/foundations/economic-security-model.md
  • Dispute handling workflow: docs/operations/dispute-handling.md