Answers are grounded in this paper and cite their sources.
Demonstrates a positive theory of structural gaps in finite environments, indicating the need for adequate support in causal reasoning.
The hidden-support account of confounding gives a negative result: the same baseline public law, even augmented by the true average treatment effect, may be compatible with either zero or arbitrarily large update-relevant structural gap. This paper develops the corresponding positive theory, with one correction: arbitrary finite environment classes determine a nonempty poset of adequate supports that need not have a least element. A finite three-state example shows why the existence hypothesis is substantive. In the public-fiber-saturated case, however, the poset has a least element: the quotient generated by public observation and the fiberwise level sets of the estimand contrast. Failed public descent then has an operational measure: the largest disagreement in the target estimand between admissible environments that share a fixed public law, with a global worst-case version obtained by ranging over public laws. This is the diameter of the corresponding partial-identification set, and in saturated finite cases it has closed-form fiber-range formulae. The ordinary log-cardinality support gap remains a useful qualitative obstruction, but it controls transport risk only after calibration by effect separation and reweighting richness. The epistemological payoff is that causal warrant is not regularity plus truth; it is regularity plus the right support for update. Hume's missing necessity becomes missing descent.
No takes yet. Share an insight, caveat, or question.
Lorand Bruhacs (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: