Answers are grounded in this paper and cite their sources.
What does it cost to change the interface through which a system is described, and when may a cheaper partial route to a new interface replace a dearer one? This paper answers both questions exactly for finite systems. Part I shows that an interface change has no single price. Its structural quote is the set of Pareto-minimal repairs under an eight-axis cost, and the quote does not determine the route selected, the purchase history or the marginal value of budget. A four-state example has four pairwise incomparable exact repairs, none within a budget that the unrepaired interface meets. Width and horizon are separate costs, and noisy deterministic kernels have a sharp tolerance cliff. Part II works over an arbitrary partial cost order. One partial route may safely replace another when a local simulation of its next steps covers it; this certificate is computable, sufficient but not necessary, and a layered solver built on it returns a witness for every minimal value of the length-bounded goal paths. Budgets survive perturbation while exact fronts can jump, positive weighted sums miss some efficient points, and a family of singleton fronts need not be uniformly enumerable. Most results are formalized in Lean 4; the paper lists the exceptions.
No takes yet. Share an insight, caveat, or question.
Ioannis Tsiokos (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: