The question

Most arguments about addiction skip this. They start mid-stream, on tactics: minimum unit pricing, Mounjaro on the NHS, sin-tax escalators, whether drug possession should be a fine or a custodial sentence. None of those tactical fights can be settled until the prior question is answered, and the prior question is this:

Is addiction a real condition of the body and brain, or is it a moral label we attach to people who haven't yet learned to make better choices?

It matters because almost every public-health policy choice — from school PSHE lessons to the NHS Mounjaro rollout, from the war on drugs to the cost-of-vice papers from the IEA — is downstream of how you answer it. If addiction is a learning gap, the cure is more education and a firmer hand. If it is a real biological state with predictable mechanisms, education and willpower are necessary but not sufficient, and a policy built on them is a policy designed to fail.

The steelman — in her voice

Before the evidence, the strongest version of the other side. Not a strawman; a fair statement of the position held by a great many medically-trained, hard-working, decent people — including the lab-trained nurse who watches her colleagues prescribe Mounjaro and quietly thinks this is cheating.

I trained in pathology. I have seen what a liver looks like at autopsy after thirty years of drinking. I have measured the HbA1c of every diabetic on the ward. And what I have seen is this: the patients who get better are the ones who decide to. The ones who don't, don't. We are now medicalising every human weakness — appetite, sadness, restlessness, the inability to stop scrolling — and calling it disease so that nobody has to do the hard work. Mounjaro is the latest example. We could give a fat patient the same outcome with a calorie deficit and a pair of trainers, but that requires character, and character is exactly what we've stopped asking for. My grandmother lived through a war. She did not need an injection to control her appetite. She needed less food, more work, and the dignity of being trusted to manage herself. Addiction is real for a small handful of cases. For the rest, it is the medical name we give to people who have not been taught — or have not yet chosen — to behave well.

— composite voice, drawn from the position itself. The full steelman is on the next page.

Read this carefully. There is something in it that is true. Self-control is real. Education matters. The set of people who would describe themselves as "addicted" is now wider than the set of people whose lives are organised around a substance they cannot stop using. Some of what gets labelled addiction is, in fact, learnable behaviour change waiting for the right intervention. The lab tech is not stupid and she is not cruel. She is partly right.

She is also partly wrong, and the part she gets wrong has consequences. Read the full steelman →

The evidence at a glance

Five numbers, each defended in full on its own page:

The panel — three experts with twenty years

The thesis is defended by a three-person panel, each with two decades of field experience. They do not always agree — that is the point. They agree on the answer to the question.

How to read this site

The site is a defended thesis, not a survey. Read in order if you have an hour. Skip to the sections you most disagree with if you have ten minutes — those are the ones built to be argued with.

1. The Steelman — the willpower-and-education position, fully voiced. Read this in the voice of someone you respect, not someone you are arguing with.
2. The Pathway — what the neuroscience actually shows. Mesolimbic dopamine, Berridge's wanting/liking, Koob's allostasis, the contested status of food and porn, and the four competing theoretical frames (Volkow, Lewis, Szalavitz, Hart).
3. The Costs — the NHS and global numbers, the Snowdon counter-arithmetic, the inconvenient van Baal 2008 finding that healthy non-smokers cost the system more over a lifetime than smokers do, and the worked actuarial case for tirzepatide.
4. The Treatments — what works at the individual level, with effect sizes. Methadone, Managed Alcohol Programs, the Portuguese arc, contingency management, the surprise Cochrane finding on Alcoholics Anonymous. What doesn't: most US abstinence-only residential rehab.
5. Countries — what worked at country scale. Portugal, France's 1995 primary-care buprenorphine, Swiss heroin-assisted treatment, Australia plain packaging, UK smoking ban + sugar tax + Scottish MUP, Iceland (cross-ref), Estonia, Ukraine OST through the war, the Netherlands coffee-shop separation, New Zealand's smokefree-generation reversal, and Singapore as the harsh-penalty counter-case.
6. Structural Interventions — the meta-pattern across what worked. Change the environment, not the person's information. Iceland's Planet Youth (highest→lowest adolescent substance use in 15 years, zero classroom drug-ed). Medellín's social urbanism. Smoking bans, plain packaging, sugar tax, MUP, choice architecture.
7. The Untaken Lever — the harder thesis, steelmanned then dismantled. Five drugs already exist that, fully adopted, would avert ~80–140M DALYs a year (~6% of global disease burden). Current adoption is ~10%. The blocker is moralising, not science. Naltrexone is the simplest single move. Anthony's position on the page, taken apart and reassembled.
8. The Industry — the political economy of not-solving. Shellenberger's San Fransicko thesis steelmanned and taken apart, Narconon as the cleanest cautionary case (Scientology recruiting through schools), and the US 28-day rehab model as the dominant evidence-light commercial industry. When "solving" addiction is a salary, the incentives don't align with solving it.
9. America — the political-economy argument applied to the largest live case study on earth. US drug-market apocalypse, OxyContin-to-fentanyl pipeline, fifteen years of inner-city collapse, the Wire's Hamsterdam in real life, and a two-voices dialogue (Shellenberger versus a fictional Inner City Harm Reduction Advocate) that arrives at one inescapable conclusion both US political tribes refuse to admit. City-level stats — SF, LA, Portland, Seattle, Philadelphia — included for empirical teeth.
10. The Policy Frustration — the load-bearing argument of the whole site. When policy assumes the willpower-and-education model is sufficient, it produces Just Say No, the war on drugs, sin taxes that hit the poor twice, fat-shaming campaigns that drive weight up not down, and an abstinence-only treatment apparatus with the weakest evidence base in addiction medicine. Getting the answer to the question wrong is expensive.
11. The Panel — three experts, three reads of the same evidence, one shared conclusion.
12. The Coda — the personal closing essay. Not what the evidence is. What it was like to build the argument, and what changed for me in the process — from an unworthy disagreement to a stable conceptual foundation I can climb from. First-person, vulnerable, honest about my own failure.
13. The Sources — full bibliography, organised by page. Every numeric and propositional claim on the site is traceable from here.