Poll — Article 11: Evidence Has a Route

[safal207]

This issue is the live reader poll for RESONANCE Issue 001, Article 11.

Thesis under vote

The strongest authorization / verification path is not automatically the best path. A runtime should select the lowest-cost admissible evidence route that still satisfies the proof obligations of the current action, state, authority, time and risk.

The article frames sync vs async authorization as a routing choice produced by policy, not the policy itself.

How to vote

React with 👍 to exactly one of the three option comments below:

• Agree — the thesis is directionally and operationally correct.
• Partially agree — the routing model is useful, but important constraints or counterexamples are missing.
• Disagree — authorization / verification should not be optimized this way, or the abstraction is unsafe / misleading.

If you choose Partially or Disagree, please add a short reason. Counterexamples are especially valuable.

This poll is intentionally non-binding. It is evidence about reader judgment, not proof of correctness.

[safal207]

✅ Agree

I agree with the core thesis: evidence routing should optimize for the lowest-cost admissible path that still satisfies the action's proof obligations at the current state/time/risk point.

React 👍 to this comment to vote Agree.

[safal207]

🟡 Partially agree

I agree with the routing idea, but I think the model is missing an important constraint, counterexample, cost dimension, or safety boundary.

React 👍 to this comment to vote Partially agree — and please add a short reason if you can.

[safal207]

❌ Disagree

I do not think authorization / verification should be optimized as a contextual evidence-routing problem, or I think this abstraction creates a material safety or governance risk.

React 👍 to this comment to vote Disagree — and please explain the failure mode or counterexample.

[safal207]

🧪 Executable update — ELR-001 falsification incorporated

The Article 11 thesis now has an executable reference implementation, and the first concrete external counterexample has already changed the contract.

babyblueviper1 independently inspected the ELR-001 PR, exact head, CI run and test count, then found a real temporal API gap: the first verify_receipt() could prove issuance-time correctness at tick T, but there was no distinct API for asking whether the same selected route remained admissible at use time T+n.

That counterexample is now preserved as a new invariant:

Historical verification != current applicability.

The implementation keeps strict historical receipt verification and adds a separate:

revalidate_receipt_for_use(
    receipt,
    original_request,
    original_graph,
    current_context,
)

The use-time gate re-checks the selected edges against current valid_until_tick, evidence age, authority epoch, policy/state/action bindings, risk and reversibility. A stale or drifted route fails closed.

The suite is now 30 tests: the original 24 plus six use-time falsification cases.

Canonical exact-head evidence:

head:   99cbc155ee3a0e3bced5734af4b1766b396cda25
run:    31875961467
event:  pull_request
result: SUCCESS

Updated proof addendum:
https://github.com/safal207/RESONANCE/blob/main/issues/001-age-of-agents/articles/11-evidence-has-a-route.elr-001-proof.md

Implementation PR:
safal207/pythiaLabs#261

CI:
https://github.com/safal207/pythiaLabs/actions/runs/31875961467

This is exactly what the poll is for: a reproducible Partially / counterexample is more valuable than an Agree vote because it can become a new executable boundary.