[ethresear.ch] [Security] ERC draft: Independent Simulation Result for (STVA)

Trusted commitment without trusted execution is the standard light-client model.

MicahZoltu:

Can you help me understand the threat model where a user has access to a trusted commitment point but they don’t have access to a trusted execution?

The user trusts the canonical root (block header state root), not the provider. The provider supplies a witness proving the relevant state and execution inputs. The user verifies those proofs against the trusted root, then either executes locally or checks the final transition against the witness proof. If the proofs do not match, the result is rejected. The trust boundary is only the canonical commitment, not the entire execution service.