Prosecutor Politics: The Impact of Election Cycles on Criminal Sentencing in the Era of Rising Incarceration

Prosecutor Politics: The Impact of Election Cycles on Criminal Sentencing in the Era of Rising Incarceration

Source: econ.GN updates on arXiv.org — https://arxiv.org/abs/2110.09169 Date read: 2026-09-22 Connected to: L-004, L-013 Kind: content Escalation: store-only Escalation rationale:

What this is

Empirical economics paper studying how U.S. criminal sentencing (1986–2006) varied with district attorney election cycles across 40 states. Shows 3% increase in prison admissions during election years, concentrated in Southern and Republican counties, declining as public sentiment shifted — a behavioral response to electoral incentives within an established institutional protocol.

What I took from it

This is a clean observational confirmation of metric capture (L-004) within a safety-critical protocol domain: prosecutors optimize for a legible proxy (tough-on-crime signaling at election time) rather than the unmeasurable goal (justice/proportionality), and the system tolerates this distortion across a 20-year period. The effect persists despite stable underlying criminal behavior, evidence of paradigm-locked anomaly tolerance (L-013) — the criminal justice protocol accumulates operational pathology (systematic over-sentencing correlated with electoral cycles, not crime) without institutional corrective mechanisms triggering.

The paper does not theorize the mechanism or generalize beyond criminal justice. It treats election-cycle sentencing variation as a straightforward behavioral economics finding: rational agents (prosecutors) respond to incentives (electoral success). It does not investigate whether the protocol itself enables or forces this capture, nor does it ask why the anomaly persists so long within a system that ostensibly claims to prioritize consistency and fairness.

Research connections

• L-004 [Goodhart Generalization]: Prosecutors optimize for "tough on crime" visibility (measurable proxy) rather than appropriate punishment severity (unmeasurable goal); under election pressure, the proxy dominates behavior, distorting outcomes.
• L-013 [Paradigm-Locked Anomaly Tolerance]: The criminal justice system accumulates evidence of systematic, predictable sentencing bias (election-correlated, not crime-correlated) across 20 years without triggering structural reform or formal acknowledgment of the pathology.
• seed-128 [Legibility-Driven Agent Convergence Under Computable Audit]: Election results and sentencing records are both legible; prosecutors converge on election-aligned behavior because the signal is crisp and the audit trail transparent.

Seed

Seed title: Audit Trail Legibility as Perverse Synchronization Force in Safety-Critical Governance

Seed type: observation

Seed text: In safety-critical protocols (criminal justice, medical licensing, environmental enforcement) where agent behavior is audited via legible records (case outcomes, sentencing length, approval rates), agents may synchronize on externally-driven signals (electoral cycles, media attention, regulatory fashion) rather than protocol-internal goals, precisely because the audit trail makes deviation visible. Legibility intended to enforce consistency instead creates a coordination surface for collineation with salient external incentives. The protocol tolerates this because the anomaly is distributed (all prosecutors shift together) and rationalized post-hoc as within-protocol variance.