I’ve been working on a small experimental system around one narrow failure case:
an external side effect may have happened, but the caller loses the response.
Instead of collapsing that into success/failure, Truth Engine can remain UNKNOWN, deny blind retry when authority is insufficient, and reconcile against evidence before changing what the system may safely claim.
I’m not positioning this as a replacement for durable execution systems. I’m specifically interested in the boundary between durable orchestration and justified claims about external occurrence.
Try the live Playground — no signup required:
Reproduce the response-lost case yourself: UNKNOWN → retry denied → reconcile → SATISFIED.I’d especially value counterexamples or cases where this model is unnecessary/wrong.
Don’t trust the claims. Break it.