Date: 2026-08-03. Tests seed hypotheses H5.1–H5.3 (H5.3 is the mandated equal-effort enforcement steelman).
Hypothesis verdicts
H5.2 (mass incarceration is largely not a drug-war artifact for state prisons) — Supported, cleanly. Drug offenses account for 13% of state prisoners (131,300 of 1,039,500, yearend 2022, BJS, cross-checked independently by Prison Policy Initiative at ~139,000) — low-double-digit and down 46% from a 2007 peak. State systems hold **~7.2x** the federal sentenced population, so this dominates the national picture. Federal prisons are ~46% drug offenses, but federal holds only ~1/7th as many people as state systems — federal concentration cannot rescue the "mass incarceration = drug war" narrative at the national level, where state systems dominate the aggregate.
H5.1 (marginal supply-side spending has no detectable effect on price/availability) — Supported, with a data-vintage caveat. The DEA STRIDE-based price/purity record (ONDCP, RAND/Kilmer) shows real cocaine prices fell ~80% and heroin fell similarly from the early 1980s through 2007 — while purity rose — during the exact decades of maximal supply-side enforcement escalation. This is strong historical evidence against a simple "more enforcement → higher price" model, corroborated qualitatively by Caulkins & Reuter's economist consensus. Caveat: the last comprehensive public STRIDE-based price series ends around 2007 — no rigorous modern (2010s–2020s) marginal-effect econometric estimate could be found. Treat the historical pattern as strongly supportive and the current-margin claim as data-starved rather than confirmed.
H5.3 (the steelman: swift-certain-fair supervision and focused deterrence outperform both mass enforcement and non-enforcement) — Partially supported, narrower than pro-enforcement framing typically presents it. This is the workstream's most important corrective finding — see below.
Arrests: possession dominates, and enforcement volume has fallen sharply
Possession has made up 82–88% of all drug arrests consistently over the past decade; sale/manufacture is a stable ~12–13%. Total drug arrests fell 47% from 2013 to 2023 (1.50M → 871K), with marijuana alone still accounting for ~217,000 arrests in 2023, the vast majority possession.
The price/purity paradox (detail)
Cocaine's real price per pure gram fell from ~400–460(early1980s)to 170–230 by 2000, continuing to decline through 2007 — an ~80% real decline since 1981. Heroin shows the same pattern. This occurred throughout the period of rising federal supply-side spending and sentence severity (1986 Anti-Drug Abuse Act onward) — the opposite of what a working deterrent-supply-suppression model would predict.
The "iron law of prohibition" — genuinely contested, not settled either way
The theory (enforcement pressure structurally selects for higher-potency products — fentanyl as an endogenous response to interdiction pressure on bulkier drugs) has real academic proponents (Beletsky & Davis 2017) but is directly disputed by name-brand researchers: Keith Humphreys calls it "an opinion with no empirical foundation" (citing legal cannabis/tobacco potency increases as counterexamples that undermine prohibition as the sole driver); Jonathan Caulkins calls it "bogus"; Peter Reuter suggests demoting it to a "copper law." Even its own co-authors concede it is "not an ironclad law" and "a common-sense observation," not a scientific one. Report as contested, not established, in either direction — fentanyl's rise is very likely multi-causal (low production cost, synthetic/non-agricultural supply chain, and geographic flexibility are independent factors from any enforcement-selection mechanism).
The steelman (H5.3) — the workstream's most consequential finding
HOPE-style swift-certain-fair supervision has a much weaker record than commonly cited. The original Hawaii RCT (Hawken & Kleiman 2009) found dramatic effects — 55% less rearrest, 72% less drug use. But the rigorous four-site multi-site replication RCT (Arkansas, Massachusetts, Oregon, Texas) found no overall advantage over conventional probation. Citing HOPE as validated evidence for swift-certain-fair supervision, without flagging the replication failure, would misrepresent the actual trial record — this is worth stating plainly in the whitepaper rather than repeating the popular, pre-replication version of the story.
Focused deterrence has real, RCT-confirmed support — but for violence suppression, not drug-market disruption. A 2025 systematic review restricted to RCTs and quasi-experiments (50 controlled evaluations, 9 RCTs) found ~23% overall crime reduction, ~16% when restricted to the 9 RCTs alone. Programs targeting gang/group gun violence show the largest effects; the drug-market-specific variant (Drug Market Intervention, the High Point NC model) shows smaller, "marginally statistically significant" effects on drug/nuisance offenses specifically (~5.5%, vs. 39% in the single best-performing neighborhood — not the average).
Drug courts have solid meta-analytic support, concentrated among serious offenders. A Campbell Systematic Review of 154 evaluations found recidivism of ~38% for drug-court participants vs. ~50% for comparable non-participants, an effect persisting at least 3 years. But the highest-quality single study (NIJ's Multi-Site Adult Drug Court Evaluation) found the benefit concentrated among more serious offenders with diagnosable drug disorders — not uniformly generalizable across the full range of low-level offenders typically diverted into drug courts.
Net read: none of the three steelman interventions demonstrate meaningful reductions in drug price, availability, or aggregate use at the population/supply level — consistent with, not contradictory to, H5.1's finding that supply-side enforcement doesn't move price. Their genuine, evidence-backed value is narrower and more specific: violence suppression (focused deterrence) and individual-level recidivism reduction for serious, court-involved offenders (drug courts) — not the broad "enforcement works" claim advocacy on that side often makes, and not the flagship HOPE model specifically, whose own replication undercuts it.
Implications for §10/§11
- §10's "enforcement steelman package" architecture should be built from focused deterrence (gang/gun-violence-targeted) and drug courts (targeted at serious/dependent offenders), not from HOPE-style universal swift-certain-fair supervision, which the replication literature does not support as a general model.
- The price/purity paradox is a direct challenge to any §10 architecture relying on supply-side/interdiction spending increases to reduce availability — the historical record offers no support for that mechanism, though the data-vintage caveat means this can't be stated as a current-decade-specific finding.
- The iron-law dispute should be presented in the whitepaper as genuinely unresolved — not resolved toward either the popular "enforcement causes fentanyl" framing or a dismissal of the concern entirely.
Confidence: Moderate
Strong on arrests/incarceration shares (primary BJS/FBI data, two-source convergence) and the steelman's RCT evidence (peer-reviewed systematic reviews). The price/purity finding is strong historically but data-starved for the current decade — flagged rather than papered over. The iron-law item is honestly reported as contested rather than resolved.