The argument in one paragraph

The US drug-market apocalypse of the last twenty-five years — from Purdue's 1996 OxyContin launch through the fentanyl flood of the 2010s and into the current ~75,000 overdose deaths per year — is the largest preventable public-health disaster the developed world has produced since the AIDS crisis. It happened across a fifteen-year collapse in the visible quality of life in US inner cities: open-air drug scenes in San Francisco's Tenderloin, Los Angeles's Skid Row, Portland's downtown, Philadelphia's Kensington, Seattle, Baltimore. The pain-treatment pipeline manufactured the demand. The fentanyl supply met it. The treatment apparatus failed to absorb it. The criminal-justice apparatus produced more harm than it prevented. The harm-reduction apparatus, where it ran at all, ran on shoestring budgets and was politically unable to defend itself. America watched a slow-motion catastrophe in real time and lied to itself about the cause from both political tribes simultaneously. This is the page where the lie is named.

Open-air drug markets

Walk three blocks west from Powell Street BART in San Francisco. Walk five blocks east from Pershing Square in downtown LA. Walk from Frankford Avenue south into Kensington Avenue in Philadelphia. Or — for the historical case — walk through the Tenderloin in the late 2010s, or down East Hastings in Vancouver's Downtown Eastside, or through Camden, Newark, parts of Baltimore. You are inside an open-air drug market. People are using on the pavement. People are dying on the pavement. People are providing services to people who are using on the pavement. The political class is making statements about it. The police are arresting some of them. The courts are processing some of them. The hospitals are receiving some of them. The morgues are receiving more of them.

This is not subtle, contested, or footnote-able. The CDC's WONDER database has the count. US drug overdose deaths exceeded 100,000 per year for the first time in 2021, peaked at ~111,000 in 2022, and remain above 75,000 in 2024 ([CDC overdose data, ongoing](https://www.cdc.gov/drugoverdose/deaths/)). The bulk are now synthetic opioids — fentanyl and its analogues — that arrived in the US drug supply roughly 2013–14 and never left. The cohort effect alone is the largest sustained mass-casualty event in US peacetime since the 1918 flu, measured in life-years lost.

The visible street outcomes are the canary. They are what every other measure of the apocalypse looks like when you can see it.

The pain-treatment pipeline — manufactured demand, met by murderous supply

The American opioid crisis has a date of birth. 1996. Purdue Pharma launched OxyContin with a marketing campaign explicitly designed to expand the prescribable-pain market, predicated on the false claim that the drug's slow-release formulation made it minimally addictive. The claim was unsupported. The marketing was aggressive, GP-targeted, and resourced. Sales reps were paid bonuses. Doctors were given junkets, plaques, free dinners. Pain became "the fifth vital sign", a clinical reframe Purdue's marketing helped engineer. Prescriptions rose. Dependence rose. By the early 2000s the prescribed-opioid wave was visible in mortality data and the company knew it.

What followed is documented in the Sackler family deposition records, the Patrick Radden Keefe book Empire of Pain, the multi-state attorneys general settlements, the eventual 2021 Purdue dissolution. Purdue made approximately $35 billion from OxyContin between 1996 and 2017. The Sacklers extracted approximately $10 billion in distributions while the company was in active litigation. The US opioid death count between the 1996 launch and the present sits north of one million Americans, depending on how you count synthetic-opioid deaths attributable to the pipeline the prescribed-opioid wave opened.

The pipeline went in four phases. Phase 1 (1996–2010): prescribed OxyContin. Iatrogenic dependence at population scale. Phase 2 (2010–13): reformulation and heroin substitution. Purdue reformulated OxyContin to be harder to crush; the dependent population transitioned to heroin, which was cheaper and more available. Phase 3 (2013–17): fentanyl arrival. Mexican cartels and Chinese precursor chemists discovered fentanyl was cheaper to manufacture and traffic than heroin; fentanyl entered the US drug supply at scale. Death rates accelerated. Phase 4 (2018–present): poly-substance fentanyl-everywhere. Fentanyl now contaminates cocaine, methamphetamine, counterfeit pills. The dependent population that began with a Purdue prescription in 2001 is now mostly dead, in OST, in prison, or buying fentanyl-laced street drugs.

The honest summary: this was an American policy failure compounded by an American regulatory failure compounded by an American criminal-justice failure compounded by an American harm-reduction failure. No other developed country produced a comparable death count from a comparable starting position. The UK, France, Germany, the Netherlands, Switzerland, Australia, Japan — none of them ran the four-phase pipeline. None of them have the body count. America is exceptional here in a way no political coalition wants to claim.

Fifteen years of inner-city collapse

The pipeline's visible expression has been the slow collapse of physical and social order in a specific set of US urban centres. The acceleration was uneven, geographically and chronologically — Baltimore had been collapsing since the 1980s crack era; San Francisco's Tenderloin became unrecognisable across 2017–2022; Portland's downtown emptied of office workers post-COVID and the encampments expanded into the vacancy; Kensington in Philadelphia became its own ecological niche by the mid-2010s; LA's Skid Row had been a containment zone since the Rampart era and grew dramatically through the 2010s.

The mechanism is well-documented in urban-policy literature. An open-air drug market is corrosive of the physical and social environment it sits inside. Property values fall. Retail closes. Working residents leave. The remaining residents are increasingly those who cannot leave: the elderly, the poor, the people who are themselves in the drug economy on one side of the transaction or the other. The street-level chaos drives out the working population that pays the tax base that funds the services that might absorb the people in the encampments. The fiscal feedback loop runs the wrong direction. Each year of the collapse makes the next year of the collapse worse.

The other contributor — and this is the part Shellenberger's framework underweights — is housing. The American inner-city collapse coincides almost exactly with the deepest housing-supply crisis in any developed country. San Francisco, LA, Portland, Seattle, Boston, New York have median single-family home prices that price out the working class within thirty miles of the city core. The encampments are partly addiction, partly housing, partly mental-health-deinstitutionalisation, partly criminal-justice churn, and partly the four interacting. Removing any one of those four reduces the encampments only modestly because the others sustain them.

Mental health — the Reagan-era closures, the load-bearing cause

Of those four contributors, mental-health-deinstitutionalisation is the load-bearing cause that no political tribe wants to claim. The American psychiatric inpatient bed count peaked in 1955 at ~558,000 patients in state hospitals — roughly 338 beds per 100,000 population, the largest such system any country had ever built. The dismantling happened in three distinct waves across four administrations and is best understood as bipartisan policy compounding, not a single party's error.

Wave 1 — Kennedy, 1963. The Community Mental Health Act promised federally-funded community mental health centres as the alternative to large state institutions. The promise was credible at the time: thorazine and the first generation of antipsychotics had arrived in 1954–55; outpatient treatment looked technically feasible. Kennedy signed the Act six weeks before his assassination. The state-hospital population was already falling; the federal funding for the community alternative began to flow.

Wave 2 — Reagan in California, 1967. Governor Reagan signed the Lanterman-Petris-Short Act, which sharply restricted involuntary civil commitment and made it nearly impossible to hospitalise an adult against their will unless they were imminently dangerous to themselves or others. California's state hospital population collapsed from ~37,000 patients in 1959 to ~7,000 by 1973 — a more than 80 per cent reduction in fifteen years. The community-care alternative the legislation assumed would catch the released patients was never built at anything like the required scale. The visible chronically-mentally-ill population on California streets dates from this period; what Shellenberger calls open-air drug scenes in the Tenderloin and Skid Row today is, in significant part, the third generation of the population released under LPS.

Wave 3 — Medicaid IMD exclusion + the Carter–Reagan inversion, 1965–1981. The 1965 Medicaid statute included the Institutions for Mental Diseases exclusion: federal Medicaid would not pay for psychiatric inpatient care at facilities with more than 16 beds. This created a structural disincentive for states to maintain large psychiatric hospitals — every long-stay patient discharged onto the street saved the state budget while shifting the cost to the federal government via downstream emergency-services, criminal-justice, and shelter spending. Carter's Mental Health Systems Act of 1980 was designed to repair the community-care side of the equation with substantial new federal funding. Reagan's 1981 Omnibus Budget Reconciliation Act repealed most of Carter's funding, converted what remained into block grants to states with roughly a 25 per cent cut, and the community mental health infrastructure that LPS and the IMD exclusion had presupposed never got built. The closure of inpatient capacity continued; the alternative did not arrive.

The cumulative arithmetic. Between 1955 and 2010, the US state-hospital bed count fell from ~558,000 to ~50,000 — a 91 per cent reduction. The US population over the same period roughly doubled, so the per-capita reduction is closer to 95 per cent. The Treatment Advocacy Center's standing estimate is that the US is currently short roughly ~95,000 to 130,000 psychiatric beds against any reasonable target derived from peer-country ratios. The Substance Abuse and Mental Health Services Administration's National Survey on Drug Use and Health estimates that approximately one-third of unsheltered homeless adults in the US have a serious mental illness, and the proportion is higher in the named cities where this page's stats table concentrates. LA County's own data puts the figure at roughly 25–30 per cent of its homeless population with severe mental illness, plus a substantially larger fraction with substance-use disorders, with high co-occurrence between the two.

The mechanism is unambiguous. The hospitals closed; the community alternative never opened at scale; the population the hospitals would have held is now on the street, in the encampments, in the morgue intake, in the prison intake, or cycling between the three. Shellenberger's framework treats progressive permissiveness as the cause of the visible street picture. The pure-housing-supply frame treats NIMBY zoning as the cause. Both are partly right and both badly underweight the fifty-year deinstitutionalisation arc that put the floridly psychotic, untreated chronically-mentally-ill population on the pavement to begin with — most of them before they encountered fentanyl, most of them before housing supply tightened. They were left to fend for themselves by a sequence of bipartisan policy decisions across four administrations, and they are still fending for themselves, on the streets the rest of America is now arguing about.

This is the cause both political tribes refuse to claim because both share responsibility for it. The right cannot easily endorse rebuilding state-level inpatient psychiatric capacity because Reagan signed the law that gutted it and the libertarian tradition objects to involuntary commitment on principle. The left cannot easily endorse rebuilding it because the legal-rights movement that drove LPS in the 1960s and 1970s is still institutionally hostile to involuntary commitment as a violation of patient autonomy. Both positions are intelligible on their own terms; neither produces a policy that addresses the population the closures created. The European-synthesis answer — supervised, time-limited civil commitment when capacity is genuinely compromised, plus community step-down treatment, plus permanent supportive housing at the back end of the pathway — exists in published form, works in the Dutch and Belgian and German systems, and is politically unbuildable in either US coalition for the same vocabulary reasons that block the rest of the synthesis on this page.

The political class processes this through frames that suit it. The right calls it progressive permissiveness. The left calls it neoliberal precarity. Both diagnoses are partly right and both prescriptions are partly impossible. What both refuse to name is that the policy package that has demonstrably worked on the same problem in other countries — the European synthesis of harm-reduction plus accountability plus housing-first plus OST in primary care plus structured re-engagement — exists, is politically achievable in those countries, and is politically impossible in either of America's current major-party coalitions.

The numbers — top Democratic-led cities, scale + direction

The visible street outcomes have empirical shape. Two sets of numbers matter: overdose mortality and homelessness, particularly the unsheltered slice. Both have been measured at city level by official agencies; the trajectory is consistent enough across the named cities that the political-vocabulary fight on this page is fighting over the same underlying picture.

Five things worth naming honestly about this table:

1. The 2023→2024 overdose decline is real and substantial. US overdose mortality fell ~27% YoY in 2024, the largest single-year drop ever recorded. The drivers are contested — naloxone saturation, fentanyl market shifts post-Chinese-precursor crackdown, increased buprenorphine access after the 2023 DATA-Waiver removal. The decline does not undo the cumulative cohort effect (>1 million Americans dead since the Purdue launch in 1996) and does not yet bring deaths back to pre-fentanyl baseline.
2. Homelessness, unlike overdoses, is still climbing fast. +18% nationally in a single year, +23% in King County, +38% unsheltered in Philadelphia. The trajectories on the two variables are decoupling.
3. San Francisco is an outlier in the wrong direction on overdose mortality and the right direction on homelessness growth. 635 overdose deaths in 2024 in a city of 800,000 is ~80 per 100,000 — five times the US average. Yet SF homelessness grew only 4% over five years against the California state rate of 20%. The two metrics are not moving together within the same jurisdiction. Any single-cause explanation has to account for that.
4. Unsheltered is the politically distinctive variable. Most US homelessness is sheltered (497k of 771k). The visible street picture that drives the political fight is the unsheltered 274k — concentrated geographically in the cities named above. The shelter system in those cities has not collapsed at the same rate as the unsheltered count has risen, which tells you the new entrants are people who are refusing shelter or who shelter is refusing — both of which point at addiction and mental illness rather than economic homelessness.
5. The data does not cleanly support either Shellenberger's "progressive permissiveness" frame OR the pure-housing-supply frame. If progressive permissiveness were the dominant cause, the SF homelessness number would be growing faster than the national rate; it isn't. If housing supply were the dominant cause, the Texas/Florida/Tennessee cities with comparable housing-affordability pressures would have comparable unsheltered counts; they don't. The actual cause is the interaction of fentanyl arrival (2014+), Reagan-era mental-health deinstitutionalisation now in its second generation, decades of underinvestment in OST and supervised consumption, and post-incarceration churn that produces homelessness on release. No single lever explains the table. The integrated European-synthesis answer addresses all four. Neither US political vocabulary names all four together.

Housing-First versus Treatment-First — Anthony's challenge

The challenge that surfaces directly out of the table: does housing come first, or does treatment come first? The harm-reduction orthodoxy is Housing-First — put the chronically homeless drug-using person in a permanent supportive housing unit without sobriety preconditions, and the housing stability creates the platform on which treatment becomes possible. Anthony's challenge to this site is the opposite: treatment first, earn the housing. Homelessness, on his reading, does not lead to drug use — the chronically unsheltered population in the cities above is overwhelmingly addicted and mentally ill, and putting an active fentanyl user in an apartment without treatment produces relapse-in-the-apartment, overdose-in-the-apartment, eviction, repeat. Steelman first, then teardown.

The chronically homeless population in West Coast US cities is not "people who lost their job and need somewhere to live". It is people who are actively dependent on fentanyl, methamphetamine, or both, with high rates of untreated psychosis and untreated mood disorder, who became homeless because the addiction or the illness made their previous housing untenable. Drop them into an apartment without treatment and you have not solved the problem; you have moved it indoors. The apartment becomes a shooting gallery. The neighbours leave. The building gets sold to the next investor. The person you housed is dead within twenty-four months. The Finnish, Utah, and Canadian Housing-First evaluations are real but were conducted on different populations — fewer fentanyl users, more newly-homeless economic cases, and crucially before the fentanyl era arrived in North American supply.

The morally serious position is: treatment first, then earn the housing through engagement. Medical detox plus OST plus mental-health stabilisation as the foundation, then transitional housing with continued engagement requirements, then permanent supportive housing as the destination. Sobriety as a condition for housing is not cruel; it is the operational requirement for the housing to be sustainable. The person cannot keep the apartment if they are actively in fentanyl-active relapse, and pretending otherwise is not compassion. It is left-wing virtue signalling that produces visible street outcomes worse than the ones it is trying to prevent.

What survives the steelman, and what doesn't.

The empirical evidence base for Housing-First is stronger than this challenge concedes. Sam Tsemberis's Pathways to Housing model in New York (1992 onward) specifically tested chronically homeless mentally-ill substance-users and found Housing-First produced 88 per cent housing retention at 24 months versus 47 per cent in treatment-first (Tsemberis et al., American Journal of Public Health, 2004). The US Veterans Affairs HUD-VASH programme, scaled nationally on Housing-First principles, drove chronic veteran homelessness down by about 50 per cent between 2010 and 2020. The Finnish national programme has reduced long-term homelessness by over 70 per cent since 2008. The Utah programme produced documented net savings on emergency services. These are not "different populations" — Pathways specifically targeted the population the challenge says Housing-First cannot work for, and produced the headline result on that exact cohort.

The challenge does land at a specific operational level, though. "Housing-First" in the published evidence base is not "give the person a key and walk away". It is housing plus intensive case management, plus on-site or rapidly-accessible mental-health services, plus OST availability through primary care, plus harm-reduction support inside the housing itself. The version Anthony is pushing back against — naked apartment placement without the wraparound layer — is not what worked in Finland or Utah or in the Pathways evaluation. It is the underfunded version of Housing-First that some US cities have implemented under the same name and that has, in those implementations, produced the relocation-of-drug-use outcome the challenge describes. The challenge is correct about the failure mode of underfunded Housing-First; it is wrong to generalise the failure to Housing-First properly implemented.

The synthesis that survives. Housing-First works when paired with the treatment layer the challenge is asking for. Treatment-First works when the treatment is actually available and the person has somewhere to be after discharge. The two positions are fighting over which element is the foundation; the empirical answer is that both elements are load-bearing and neither alone is sufficient. Anthony's challenge is correct that "give the person a key" without the treatment scaffolding is a failure mode; the published evidence base is correct that adding a sobriety precondition to housing access reduces housing retention without improving treatment outcomes. The European model — and the part of the US model that has actually worked at scale (HUD-VASH) — is Housing-First plus full wraparound services, which is functionally indistinguishable from "Treatment-First with permanent housing as the platform". The two camps are arguing over sequencing on the same underlying package. The political fight obscures the operational agreement.

The honest version of the housing-vs-addiction question for the table above: the chronically unsheltered population in the named cities is mostly addiction-and-mental-illness driven, not economic; both Housing-First (properly funded) and Treatment-First (properly available) can work; neither works in the half-funded version the US runs at the moment; the political fight over which comes first is a proxy for the funding fight neither coalition wants to have.

Hamsterdam — when David Simon's fiction caught reality

The Wire's third season (HBO, 2004) ran an arc that has become unavoidable in any honest discussion of American drug policy. Bunny Colvin, a Baltimore Western District major nearing retirement, set up an unofficial de facto decriminalisation zone — three condemned blocks where his officers were instructed not to arrest drug dealers and users, in exchange for the corner-boys moving their operations there and clearing the rest of the district. The zone was named Hamsterdam by the corner-boys, after Amsterdam, in a homophonic act of street vernacular.

The arc's narrative point was complicated. Hamsterdam worked in Bunny Colvin's terms: violent crime in the surrounding district fell measurably; the residents of the cleared streets reclaimed their pavements; the dealers and users were contained, observable, reachable by outreach workers, providable with services. The arc's narrative point was also that Hamsterdam was a humanitarian disaster: the conditions inside the zone were degrading, the public-health outcomes were catastrophic without parallel investment in treatment and housing, and the political class — when it discovered the experiment — destroyed it for being politically unacceptable rather than because it had failed at its stated objective.

Twenty years later, San Francisco's Tenderloin is Hamsterdam without the deliberate policy choice that produced it. So is Skid Row. So is Kensington. So was East Hastings. The American urban landscape evolved, by accident and by political default, into the model The Wire used as its 2004 thought-experiment. The fictional version was supposed to be a provocation. The real version is the equilibrium. And the response to the real version is exactly the response Hamsterdam got in the show: episodic political destruction of the visible zone, displacement to a slightly less visible zone, no parallel investment in the treatment and housing that would have made the containment humane.

David Simon, asked about this in interviews across the 2010s and 2020s, has been clear: Hamsterdam was not an endorsement of decriminalisation. It was a portrait of what happens when policy abandons a population. The current US street picture is what he was warning about. The country watched the show, told each other it was great television, and then let the city version happen without the part where anyone tries to do anything about it.

The multi-decade failure to stem the flow

The supply-side war on drugs predates the demand-side opioid crisis by decades. It also predates and outlasts every administrative attempt to win it. The historical record is the same shape in every category.

• Heroin (1970s–present): US street purity at point of sale rose from ~6 per cent in 1980 to ~40+ per cent by 2005, with prices falling proportionately in real terms (ONDCP / DEA price-purity data). The supply-side enforcement push did not reduce supply. It selected for more potent product.
• Cocaine (1980s–present): Crack epidemic devastated US inner cities through the late 1980s and early 1990s; supply-side enforcement against Colombia, then Mexico, then through the Plan Colombia / Plan Mérida period, did not reduce cocaine supply at the US street level. Production migrated. Purity rose.
• Methamphetamine (1990s–present): Domestic production wave 1990s–2000s; shut down by precursor controls; replaced by industrial-scale Mexican meth that is now purer, cheaper, and more available than the home-cooked product ever was. The "shut down meth labs" success became the "meth is now everywhere" failure.
• Fentanyl (2014–present): Precursor controls on Chinese chemical suppliers shifted production to Mexican cartels. The substance is so potent and so cheap to traffic per dose that interdiction at the border is operationally impossible — the entire annual US supply fits inside a single shipping container.

The cumulative empirical record, decade after decade, substance after substance, is that supply-side enforcement cannot beat market economics. When demand exists and the per-unit profit margin is high, supply finds the user. The supply-side push has cost trillions of dollars, incarcerated millions of people disproportionately of colour, and not moved the dependent variable. This is not a partisan claim; it is the consensus position of every serious drug-policy analyst in the developed world. American politics processes it anyway as if it were contested.

America is lying to itself — both sides, identically

The right's lie is that addiction is a moral failure and the solution is enforcement plus treatment of a specific abstinence-only kind. The empirical record on this is the four-decade history of the war on drugs and the 28-day Minnesota Model. Both have been evaluated. Both have failed. The right continues to advocate for them because the framework requires the failure: if the addicted population is morally responsible for its addiction, more enforcement is always the answer regardless of what the evaluation shows.

The left's lie is that harm reduction without accountability is sufficient and that the visible street outcomes are primarily a housing problem rather than a treatment problem. The empirical record on this is the post-2015 trajectory of San Francisco, LA, Portland, Seattle. The harm-reduction layer is doing exactly the work it was designed to do — keeping users alive long enough to be reachable. The accountability layer, the housing layer, the OST-in-primary-care layer, the structured-re-engagement layer — the things that, in the European synthesis, make harm reduction part of an actual treatment apparatus — have not been built. The left continues to advocate for harm reduction alone because the framework requires the addicted person to be a rights-bearing autonomous adult whose treatment choices must not be coerced, which makes the structured re-engagement layer politically unbuildable.

Both lies share a structural feature: each tribe's framework requires the other tribe's policy preference to be the cause of the visible failure. The right blames progressive permissiveness; the left blames neoliberal underinvestment. Neither tribe will name the package that has demonstrably worked elsewhere because that package contains elements both tribes find ideologically unacceptable. The left cannot endorse compelled treatment as an option; the right cannot endorse decriminalisation as the architecture. Both refuse the integrated European model and continue to advocate for the half they each like.

The result, twenty-eight years after Purdue launched OxyContin and a decade after fentanyl arrived in the supply, is that America has not adopted any of the policy levers that worked in France, Switzerland, Portugal, the Netherlands, Australia, or the UK. Methadone access is still federally restricted relative to those countries. Buprenorphine was DATA-Waiver-restricted until 2023. Supervised consumption sites are still legally precarious. Drug-court routing is patchwork. Housing-first programmes are oversubscribed and under-funded. The result of refusing the European synthesis is the body count the CDC reports every quarter.

Two voices, in dialogue

The page now hands the microphone to two characters who would disagree on essentially every political question except the load-bearing one this page is about.

Michael Shellenberger is the named voice on the centre-right. His critique of the US harm-reduction-without-accountability model is steelmanned at full strength on the Industry page. The position he holds here is the same.

The opposing voice is fictional but composite, drawn from twenty years of testimony by named harm-reduction practitioners who have worked the actual streets of the Tenderloin, Skid Row, Kensington, the Bronx, East Hastings. Renée Wilson is not a real person; she is the harm-reduction worker the Shellenberger framework systematically refuses to interview at full strength. Twenty years of needle exchange, naloxone distribution, post-overdose outreach, and the operational work that has kept hundreds of users alive long enough to talk to her about treatment. She knows Shellenberger personally and considers his framework politically convenient and operationally ignorant. She is also unsentimental about the failures he names.

Shellenberger speaks

I have walked the Tenderloin with you, Renée. We agree on what we both saw. We disagree on what produced it and what to do about it.

The visible outcomes are what they are. Open drug scenes. People using on the pavement, dying on the pavement, deteriorating on the pavement in slow motion across years. Encampments that the people inside them cannot leave because there is no exit ramp to anywhere better. Tens of thousands of dead from fentanyl every year in California alone. We agree on the diagnosis at the level of what we both observe.

I name the cause as ideological. The progressive harm-reduction-without-accountability package — the specifically American version of it, not the European version — created the conditions for the open drug scenes by removing the structured pressure that European cities use to channel users into treatment. The accountability layer is the part you removed. Drug-court routing as a default not an option. Compelled treatment for the floridly psychotic and the seriously addicted who cannot consent functionally. Housing with sobriety-progress conditions. Supervised consumption inside designated facilities rather than on the street. Refusing treatment must have consequences or the user will not engage with treatment.

I name the institutional sustaining mechanism as the addiction industrial complex. The bed-night billing model. The grant-funded NGO pipeline whose budgets require the problem to continue. The political class that signals progressive virtue and cannot endorse accountability without losing primaries. The progressive coalition's refusal to admit that the European model it claims to admire actually contains the accountability layer it operationally refuses to build.

The fix is to restore the accountability layer. Mandatory drug-court routing for street users. Compelled residential treatment where consent is compromised. Housing-first with sobriety-progress conditions for continued residency. The European synthesis, properly assembled, not the harm-reduction half on its own. Until that happens the West Coast cities will continue to be a humanitarian disaster operated under a humanitarian slogan, and the rest of the country will continue to draw the lesson that harm reduction itself is the problem.

The Inner City Advocate speaks

I agree with you, Michael, on the visible outcomes. I agree with you that what we have is catastrophic. I agree with you that the European synthesis is the answer. I disagree with you on every other piece of the diagnosis and that disagreement matters because the political coalition you represent is not going to fund the European synthesis either.

The cause is not progressive ideology. The cause is forty years of compounding failure that arrived from multiple directions at once and that no political coalition has tried to fix. Purdue's OxyContin marketing — a Republican-deregulated FDA approval cycle, a Republican-friendly pharma lobby, a class of corporate executives that paid no individual price. Fentanyl arrival 2013–14 — a supply-side failure under both administrations. Reagan-era mental-health deinstitutionalisation that put a generation of unmedicated psychotic people on the street thirty years before fentanyl arrived to find them. Bipartisan tough-on-crime sentencing that produced the largest prison population in human history and a post-release homelessness pipeline both parties wrote together. NIMBY housing supply restrictions that both parties' base voters lobby for at the local level and that have made the cost of moving inside the city impossible for the population that would otherwise have left the encampments by themselves. The encampments are not progressive permissiveness. They are forty years of cross-partisan policy compounding.

The harm-reduction layer is doing the work it was designed to do. We are keeping people alive long enough to be reachable. The naloxone distribution programme prevented hundreds of thousands of deaths. The needle exchange prevented the second HIV epidemic. Supervised consumption, where it has been allowed to exist, prevents overdoses inside the facility every week. The reason the visible outcomes are catastrophic is not that the harm-reduction layer is doing too much. It is that the rest of the system — housing, OST in primary care, drug-court routing, supervised consumption, the structured re-engagement layer you are correctly asking for — has not been built.

I have been asking for that layer for twenty years. Every harm-reduction practitioner I have ever worked with has been asking for it for twenty years. We have not got it because the political coalition you represent will not fund it. Compelled treatment, the way you describe it, the Cochrane review of compulsory drug treatment found no consistent evidence of effectiveness. Drug courts that work — and they do work for the population that is mandated into a treatment they would not otherwise have chosen — work because the treatment exists to be mandated into. The drug courts in the cities you criticise do not have treatment slots to refer the mandated population to. They have waiting lists. The accountability layer requires a treatment layer to be accountable to. America does not have the treatment layer.

I will agree with you on the synthesis. I will not agree with you on the political vocabulary because your vocabulary makes the synthesis unbuildable on the left. When you call this "progressive ideology" you make every Democratic mayor of a West Coast city unable to endorse the structured re-engagement layer you are correctly asking for, because endorsing it now reads as conceding to a Republican attack frame. The harm-reduction frame I represent has the same problem in reverse — when I describe the synthesis as "harm reduction with accountability" the Republican coalition reads it as a left-wing capture and refuses to fund the accountability layer either. Both of our political vocabularies prevent the policy package both of us want to see built. That is the meta-failure.

The fix you and I would both endorse — OST in primary care at France-1995 scale, supervised consumption at Swiss-2010 scale, housing-first with structured case management at Finnish scale, drug-court routing as option with treatment slots actually available, decriminalisation of personal-use possession paired with funded CDT-style commissions — is what every European country I have studied has done in some combination. It is also, in 2026 America, politically impossible because neither of our coalitions can endorse it without alienating its base. We are both correct about the destination. We are both correct that the other side is part of why we cannot get there.

Where they agree, named explicitly

The two positions converge on a substantial common ground that neither US political coalition will admit.

1. The visible street outcomes are catastrophic and getting worse. Denial is not on the table for either of them.
2. The European model — France 1995 buprenorphine, Swiss HAT, Portuguese funded decriminalisation, Dutch market separation, UK smoking-cessation infrastructure — is the policy package that has worked. Both Shellenberger and the Inner City Advocate point at the same continent for the answer.
3. Harm reduction works when it is operationally complete. Both agree the naloxone, the needle exchange, the supervised consumption rooms are doing the work they were designed to do. Both agree that harm reduction alone, without the rest of the synthesis, leaves the visible street outcomes the page describes.
4. The accountability layer is real, defensible, and necessary. Shellenberger names it as the missing piece. The Inner City Advocate agrees it is missing — and adds that it is missing because the treatment layer it would be accountable to has not been built either.
5. OST in primary care is the single highest-leverage move available. France 1995 is the case both of them cite. Both agree the US has not done it. Both blame the political coalition the other one represents for the failure to do it.
6. The institutional sustaining mechanism is real. The bed-night billing model, the unaccountable NGO pipeline, the political-virtue-signalling-without-action layer — both of them name versions of this. They disagree on which tribe sustains it the most; they agree it sustains the failure.

One inescapable conclusion

The conclusion this page is built to deliver is the one neither US political coalition has been willing to say out loud.

America has the answer in front of it, in published peer-reviewed form, with three decades of European empirical evidence, and is choosing not to adopt it because the political vocabulary on both sides forbids the synthesis. The Right's vocabulary forbids decriminalisation as a starting condition. The Left's vocabulary forbids structured accountability as an operational tool. The package that demonstrably works in France, Switzerland, Portugal, the Netherlands, and the UK contains both elements as inseparable parts of one architecture. America cannot adopt that architecture under either of its current political brands because each brand owns half of it and disowns the other half.

The result is the body count the CDC publishes every quarter. The result is the visible street picture that has accumulated across fifteen years of compounding institutional refusal. The result is that Shellenberger and the Inner City Advocate — who would disagree on essentially every other political question — sit on opposite ends of an enormous bipartisan agreement that no American politician can pick up because picking it up would mean conceding to the other side's vocabulary at the level of public legitimacy. The synthesis is politically homeless and the deaths continue.

The Wire's Hamsterdam ended when the political class discovered it and could not publicly endorse it. The real-world version continues because the political class also cannot publicly endorse fixing it. Both refusals are the same shape and both produce the same outcome. The deaths are not a fact about addiction. They are a fact about the political vocabulary America has agreed to use about addiction. The first country to break the vocabulary will get to adopt the European synthesis and watch the body count fall the way it fell in France between 1995 and 2005. America has not broken its vocabulary in twenty-eight years. The OxyContin pipeline opened in 1996. The math is now written in the morgues.

That is the inescapable conclusion. The two stories arrive at it from opposite directions and would tell it in opposite political vocabularies. The site has tried to present both stories at their strongest because the conclusion only survives if both voices are taken seriously. Neither is wrong about the destination. Both are correct that the other side is part of why the destination is not being reached. The Policy page documents the forty-year graveyard of interventions that designed themselves around one half of the synthesis or the other. The Untaken Lever documents what could be done if America were willing to break the vocabulary. The Coda is a personal reflection on what it takes to hold a stable conceptual foundation on a topic this politically loaded. This page is the one where the foundation gets pressed against the largest single instance of policy failure in the field, and the conclusion that survives that contact gets named in plain sentences.

The conclusion in one line. America is killing roughly 75,000 of its own citizens per year because its political vocabulary will not let it adopt the policy package that has worked in every other country that has tried it. Not because the policy is unknown. Not because the evidence is contested. Because the words for the policy cannot be spoken by either of America's political tribes without conceding ground neither tribe will concede.

That is the inescapable conclusion. Two stories arrived at it. The site rests its case there.