# VWDC-05 Handoff — Reality-Tethered World Calibration, Transport Contracts, and External Validity

## Starting state

VWDC-03/04 use:

$$
D
=
L_T\varepsilon_W
+
\delta_{\mathrm{trans}}.
$$

VWDC-04 proves internal-only branching cannot reduce total claim debt below fixed:

$$
\delta_{\mathrm{trans}}.
$$

It treats external validation as an experiment competing with additional simulation.

## Objective

Make the world-to-reality transport layer itself a calibrated, versioned contract.

## Main questions

1. How should $T_{W\to R}$ be defined by target quantity?
2. How should transport discrepancy be estimated?
3. Which parts are marginally testable and which remain structural assumptions?
4. How should subtrace/conditional validation update transport confidence?
5. How should contracts expire when the world model or reality regime changes?
6. How should several external datasets/measurement processes be fused?
7. Which counterfactual claims remain assumption-indexed after validation?
8. How should transport debt be localized by task/state/action region?

## Desired form

$$
\boxed{
\text{world claim}
+
\text{external observations}
+
\text{validation design}
+
\text{assumption set}
\Longrightarrow
\text{versioned transport contract}
+
\text{residual debt}
+
\text{allowed claim scope}.
}
$$

## Prohibitions

- Do not equate simulation precision with external validity.
- Do not claim a counterfactual quantity is empirically identified if its dependence structure is unobserved.
- Do not reuse a transport contract across model/version/domain drift without revalidation.
- Keep world-internal and reality evidence labels distinct.