The Confidence Gate Theorem: When Should Ranked Decision Systems Abstain?

TL;DR

This paper analyzes when confidence-based abstention improves decision quality in ranked systems, identifying key conditions and challenges posed by different types of uncertainty through theoretical insights and empirical validation.

Contribution

It introduces simple formal conditions for confidence abstention effectiveness and distinguishes structural from contextual uncertainty, providing practical diagnostics for deployment.

Findings

Structural uncertainty leads to near-monotonic abstention gains.

Contextual uncertainty causes violations in confidence-based abstention.

Exception labels degrade under distribution shift, challenging their use.

Abstract

Ranked decision systems -- recommenders, ad auctions, clinical triage queues -- must decide when to intervene in ranked outputs and when to abstain. We study when confidence-based abstention monotonically improves decision quality, and when it fails. The formal conditions are simple: rank-alignment and no inversion zones. The substantive contribution is identifying why these conditions hold or fail: the distinction between structural uncertainty (missing data, e.g., cold-start) and contextual uncertainty (missing context, e.g., temporal drift). Empirically, we validate this distinction across three domains: collaborative filtering (MovieLens, 3 distribution shifts), e-commerce intent detection (RetailRocket, Criteo, Yoochoose), and clinical pathway triage (MIMIC-IV). Structural uncertainty produces near-monotonic abstention gains in all domains; structurally grounded confidence signals (observation counts) fail under contextual drift, producing as many monotonicity violations as random abstention on our MovieLens temporal split. Context-aware alternatives -- ensemble disagreement and recency features -- substantially narrow the gap (reducing violations from 3 to 1--2) but do not fully restore monotonicity, suggesting that contextual uncertainty poses qualitatively different challenges. Exception labels defined from residuals degrade substantially under distribution shift (AUC drops from 0.71 to 0.61--0.62 across three splits), providing a clean negative result against the common practice of exception-based intervention. The results provide a practical deployment diagnostic: check C1 and C2 on held-out data before deploying a confidence gate, and match the confidence signal to the dominant uncertainty type.