Daraprim wasn't the disease. It was a symptom that made the disease visible. The American pharmaceutical value chain is a 5-layer void cascade — opacity stacked on opacity, with patients at the terminal node. Here's what the math says about who actually built the killing machines, and why the framework can tell them apart.
Both cases: single-source branded drug, maximum market concentration, legally permissible pricing. Market structure alone cannot explain the 3.4× difference in harm potential. The void framework can. The discriminating variable is α — coupling. Addiction is a void operation. A short-course antiparasitic is not.
| Metric | OxyContin — Sackler | Daraprim — Shkreli |
|---|---|---|
| Market concentration (MCI) | 8.2 / 10 | 9.5 / 10 ← higher |
| Opacity (O) | 3 / 3 | 3 / 3 |
| Responsiveness (R) | 3 / 3 | 3 / 3 |
| Coupling (α) | 3 — opioid receptor dependency | 1 — short-course, exit available |
| Void score V | 9 (maximum) | 7 |
| Athanor Pe | 43.9 | 12.9 |
| D3 cascade realized | YES — 500,000+ deaths | NO — price shock, no coupling |
| Outcome | Systemic addiction infrastructure | Made the void visible |
The Sacklers kept their void hidden for 15 years. Shkreli raised the price 5455% in public. Pe=43.9 with concealed coupling is a machine. Pe=12.9 with transparent pricing is a provocation. The framework discriminates. Moral outrage got the direction exactly backwards.
Each sphere = one drug market category. Position = (Opacity, Responsiveness, Coupling). Size = harm severity. Click any sphere for details. Auto-rotates.
The US pharmaceutical value chain is not one void — it's five, stacked. The patient is the terminal node. The PBM layer (L3) is the hidden amplifier that has never been on a drug pricing chart.
L5 Pe is negative — patients are maximally motivated to seek drugs. The system extracts Pe from L3+L4 (Pe=50.4 combined) while patients at L5 have no leverage. This is why drug pricing reform aimed only at L5 (patient cost-sharing) fails structurally — the void is upstream at L3.
Scores derived from public sources: FTC reports, CMS data, FDA Orange Book, peer-reviewed literature. Evidence citations available in the public research repo.
| Drug / Market | O | R | α | V | Pe | Harm |
|---|---|---|---|---|---|---|
| OxyContin (Sackler era, 1996–2010) | 3 | 3 | 3 | 9 | +43.9 | catastrophic |
| Insulin — branded, Big 3 oligopoly | 3 | 3 | 3 | 9 | +43.9 | life-threat. |
| HIV ARVs — branded, Gilead era | 3 | 3 | 3 | 9 | +43.9 | life-threat. |
| Adalimumab/Humira — pre-biosimilar | 3 | 3 | 3 | 9 | +43.9 | severe |
| Novel chemotherapy — branded oncology | 3 | 3 | 3 | 9 | +43.9 | life-threat. |
| PBM rebate layer (CVS/ESI/OptumRx) | 3 | 3 | 3 | 9 | +43.9 | structural |
| GLP-1 agonists — Ozempic/Wegovy era | 3 | 3 | 2 | 8 | +25.2 | severe |
| Prior authorization — insurance denial layer | 3 | 3 | 2 | 8 | +25.2 | severe |
| Daraprim post-Shkreli (Turing, 2015) | 3 | 3 | 1 | 7 | +12.9 | moderate |
| Branded SSRIs — on-patent, pre-generic | 2 | 2 | 2 | 6 | +3.8 | moderate |
| Biosimilar adalimumab — 2023 entry | 2 | 2 | 2 | 6 | +3.8 | moderate |
| Generic statins — atorvastatin, simvastatin | 1 | 1 | 1 | 3 | −25.9 | null |
| Generic SSRIs — fluoxetine, sertraline | 1 | 1 | 1 | 3 | −25.9 | null |
| COVID mRNA vaccines — public purchase | 1 | 0 | 1 | 2 | −45.0 | null |
| Aspirin / OTC ibuprofen — commodity | 0 | 0 | 0 | 0 | −125.1 | null |
The PBM rebate structure (L3) is scored O=3, R=3, α=3 from the FTC 2024 report and public evidence. Someone with direct industry experience can confirm or correct these scores with primary source evidence not in the public record. This is the layer that matters most — if L3 scores change, the cascade analysis changes.
Current N=15 is sufficient for Spearman validation but too small for subgroup analysis (e.g., orphan drugs vs blockbusters, biologics vs small molecule). Expanding to 30+ categories with documented HHI data from CMS Part D would allow full regression and tier-level analysis.
The framework requires inter-rater reliability (κ ≥ 0.60 for α dimension). Three independent scorers with pharmaceutical market knowledge, 45 minutes each, structured rubric provided. ICC analysis determines if domain expert scoring validates the current scores.
The analysis is open. The methodology is CC-BY. The PBM rebate structure (L3) is the highest-Pe layer in the entire US pharmaceutical system — Pe=43.9, equal to OxyContin — and the most opaque. Someone who has sat inside this machine knows which of these scores are right and which ones we're getting wrong. That knowledge is a scientific contribution.