Start with a distinction most headlines skip
When people say violence is "drug related," they usually mean one thing: someone used a drug and then did something violent. The research says that is the smallest part of the story, and often the wrong part. A more honest account has three separate relationships between substances and violence, and they point to very different solutions.
The framework comes from Paul Goldstein, who studied New York City homicides in the 1980s and sorted drug related violence into three kinds.1 First, psychopharmacological violence, where intoxication itself changes behavior. Second, economic compulsive violence, where someone commits a crime to get money to buy drugs. Third, systemic violence, where violence grows out of the illegal drug trade itself: turf disputes, debt collection, robberies of dealers, and punishment for selling diluted or fake product. That third kind is the one most people never hear about, and it is the largest.
What the numbers actually say
Start with alcohol, because it is the substance most tied to violence and also the one that is legal. In Bureau of Justice Statistics surveys, about 37 percent of people convicted of violent crimes reported being under the influence of alcohol when they committed them.2 A meta-analysis of homicide studies found that 48 percent of homicide offenders had been drinking beforehand.3 The Centers for Disease Control and Prevention estimates roughly 7,000 alcohol attributable homicides per year in the United States.4 If the conversation about substances and violence leaves out alcohol, it is leaving out the biggest contributor.
Now look at illegal drugs through Goldstein's lens. In his analysis of 414 homicides in selected New York City precincts in 1988, 52.7 percent were classified as drug related. But of those drug related homicides, 74 percent were systemic, 14 percent were psychopharmacological, and only about 4 percent were economic compulsive.1 In other words, nearly three quarters of drug related killing in that study was not caused by drugs changing someone's brain, and it was not caused by someone stealing to feed a habit. It was caused by the structure of an illegal market: people fighting over territory, collecting debts, and settling disputes in a business that cannot call the police or sue for breach of contract.
That finding has held up in its broad shape. The US Sentencing Commission, reviewing the evidence on crack cocaine and violence, concluded that the primary association between crack and violence was systemic, the violence of the black market and distribution.5
The most faithful recent replication comes not from a US city but from England and Wales. In February 2024, the UK Office for National Statistics applied Goldstein's framework to the Home Office Homicide Index for April 2018 to March 2023. Of 3,148 homicides, 1,627 (52 percent) counted as drug related under the Home Office's broad definition, and 838 (27 percent of all homicides) met the criteria for at least one Goldstein category: 464 systemic (15 percent), 399 psychopharmacological (13 percent), and 159 economic compulsive (5 percent), with some overlap between categories.6 Two caveats matter. First, nearly half of the drug related cases, 789 of them, were classified as having an ambiguous link to drugs, meeting none of the three criteria, because administrative records, unlike Goldstein's case file coding, often lack the detail needed to classify a killing. Second, the ONS counted only suspects under the influence for the psychopharmacological category, where Goldstein also counted victims. Even with those caveats, the shape rhymes with Goldstein: systemic largest, economic compulsive smallest. It is also worth stating plainly what the data do not say. Most people who use drugs never commit a violent crime. Association is not causation, and intoxication is one factor among many, including poverty, trauma, access to firearms, and the social setting where drinking or drug use happens.
The second ledger: what the trade itself produces
Illegal drug markets are big businesses operating without courts, contracts, or regulators. RAND estimated that Americans spent $109 billion on cannabis, cocaine, heroin, and methamphetamine in 2010, and between $120 billion and $145 billion a year on those four drugs from 2006 to 2016.7 When a market that size is pushed outside the law, violence becomes its enforcement mechanism. That is not a metaphor. It is how disputes get resolved when no lawful alternative exists.
The United States has run this experiment before, with alcohol. During Prohibition, armed gangs fought to control the alcohol trade, and the violence was widely blamed on drinking. When Prohibition ended, the gang war over alcohol ended with it, because the product moved back into a legal, regulated market where disputes go to court instead of to a gunfight. Nobody needed to argue that alcohol was harmless to see that prohibition itself was manufacturing a particular kind of violence. The same logic applies now, and it is the reason the systemic category matters so much: a large share of drug related violence is violence we built through policy, not violence that drugs built through chemistry.
Systemic violence moved online, and split in two directions
Goldstein's market was a street corner. Part of today's market is a screen, and the move online has pushed systemic violence in opposite directions at once, which turns out to be clarifying.
In one direction, dark web cryptomarkets have reduced some of the violence Goldstein described. Buyers and sellers never meet face to face, payments sit in escrow until delivery is confirmed, vendor ratings and reviews substitute for reputation enforced by threats, and marketplace administrators mediate disputes. Researchers studying these markets report that buyers experience fewer threats to personal safety and less violence than when sourcing through known dealers, strangers, or even friends, and that physical intimidation is largely unavailable in a digital environment.8 The nonviolent harms did not disappear: exit scams, fraud, blackmail, and doxxing, the exposure of someone's personal information, all still occur. And the benefits are uneven, with more resourced participants moving online while more marginalized people remain in physical markets where systemic violence concentrates. But the basic point supports the argument of this post. When a market gets even a rough substitute for courts, contracts, and dispute resolution, violence falls, because violence was doing the job those institutions normally do.
In the other direction, the beef moved online too, and got faster and more public. Researchers studying gang involved youth in Chicago describe a "digital street" where street culture is translated onto social media: bragging, insults, threats, and displays of affiliation, a practice called internet banging. Outreach workers interviewed in that research identified specific online behaviors that increased the likelihood of violent confrontation, including taunting rivals by posting video from their turf, disrespecting a rival group's symbols, and disrespecting fallen rivals.9 Where a conflict once needed a face to face encounter to escalate, a post can now do it instantly, in front of an audience, and the script of reciprocal violence gets written online in real time, anticipating, directing, and mourning neighborhood violence as it unfolds. This is still systemic violence in Goldstein's sense, violence arising from the structure and culture of an illegal market, but the territory being defended is now partly reputational and partly digital, and the audience changes the stakes. A 2025 review of the drugs and violence literature argues the framework needs exactly this kind of expansion for a globalized era, accounting for synthetic drugs, online transactions, and the different stages of the drug route.10
Put the two directions together and a pattern emerges. Anonymity plus governance reduces violence. Publicity plus reputation increases it. That is a useful guide for policy, and for platform design: anything that gives people a nonviolent way to resolve a dispute, or a reason not to perform a conflict for an audience, is working against systemic violence, whatever the technology involved.
The prison industrial complex, and who gets paid
If systemic violence is the product, mass incarceration is the business built around it. About 1.9 million people are incarcerated in the United States.12 Drug offenses account for 45 percent of people in federal prison (64,600 people as of September 30, 2023) and 13 percent of people in state prison, where violent offenses dominate at 62 percent.11 Those two facts together are important: the federal system is heavily a drug enforcement system, while the state system mostly holds people convicted of violent offenses, many of which, as the alcohol data show, involve intoxication rather than drug trafficking.
Follow the money and the picture gets wider than private prisons. Direct corrections spending runs about 81 billion dollars a year, and the full criminal justice system costs roughly 445 billion dollars a year across policing, courts, and corrections, according to a 2026 Prison Policy Initiative analysis.12 Private prisons account for only about 3.1 billion dollars of that total, less than one percent. The much larger business is public: staff salaries, construction, and contracts for food, health care, commissary goods, and phone calls. Families of incarcerated people pay an estimated 27.7 billion dollars a year in fines, fees, bail premiums, and commissary and communication costs.12 The system does not need private prison corporations to be profitable for the people who run it. It is already a large, self sustaining economy, and every drug arrest feeds it.
That economy is also distributed unequally. Black Americans are arrested for cannabis possession at nearly four times the rate of white Americans despite comparable use rates, according to the American Civil Liberties Union.13 Enforcement falls hardest on the same communities that bear the most systemic violence, which means the policy response deepens the harm it claims to address. And because a criminal record limits employment, housing, and education for years afterward, incarceration itself becomes a driver of the poverty and instability that make both substance use and violence more likely. The status quo does not just fail to solve the problem. It reproduces it, and a lot of people and institutions are paid along the way.
The demand side, where the evidence points
If the supply side approach is enforcement, interdiction, and incarceration, the demand side approach is reducing the number of people who need illegal markets in the first place, and reducing the harm for those who use drugs now. The evidence for this shift is old, strong, and mostly ignored.
In 1994, the RAND Corporation compared what it would cost to cut cocaine consumption by one percent through different strategies. Treatment of heavy users would cost about 34 million dollars. Domestic enforcement would cost about 246 million dollars, or 7.3 times as much, for the same reduction. Source country control would cost 23 times as much, and interdiction nearly 11 times as much. When RAND measured crime and lost productivity rather than consumption, treatment was about 15 times more cost effective than enforcement.14 A later RAND study of mandatory minimum sentences found the same pattern: for each additional million dollars spent, treatment of heavy users reduced serious crime roughly 10 times more than conventional enforcement and 15 times more than mandatory minimums, even though only about 13 percent of people treated stopped heavy use permanently.15 Treatment does not have to cure everyone to beat prison on the math. It just has to work for some people, some of the time, at a fraction of the price.
A demand side approach, taken seriously, has several parts. First, treatment on demand: medication for opioid use disorder, counseling, and residential care available when someone is ready, not after a months long waitlist. Second, harm reduction that keeps people alive and connected to services: naloxone distribution, fentanyl test strips, syringe services, and overdose prevention sites, which reduce death and disease without requiring abstinence first. Third, diversion from arrest and prosecution toward services: drug courts, deflection programs, and crisis response teams that treat a substance use crisis as a health event rather than a criminal one. Fourth, the unglamorous work of addressing why demand exists: stable housing, employment, trauma care, and community investment in the neighborhoods where both the drug trade and its enforcement concentrate. None of this requires pretending that substances carry no risk. It requires admitting that punishment has been the most expensive and least effective way to reduce that risk.
There is also a democratic question buried in this. Opioid settlement funds, now flowing to states and localities in the billions, are being allocated through processes that often exclude the people most harmed: people who use drugs, their families, and the communities where enforcement fell hardest. Participatory allocation, where affected communities have a real say in how remediation money is spent, is not a courtesy. It is how you avoid repeating the pattern in which decisions about drug policy are made by people who profit from the status quo.
The bottom line
Substances are genuinely connected to violence, but not in the way the slogan suggests. Alcohol, the legal drug, accounts for the largest share of intoxication related violence. For illegal drugs, the largest share of violence comes not from intoxication and not from theft to buy drugs, but from the illegal market itself, a market that prohibition created and enforcement sustains. Around that market, we have built a 445 billion dollar system that pays many people to keep it going, and that falls hardest on Black and low income communities.
Demand side approaches are not soft on violence. They are the strategy with the best evidence for reducing it: treat people who want help, keep people alive, divert people from cells to services, and invest in the communities where the trade thrives. The violence we blame on drugs will shrink fastest when we stop manufacturing it ourselves.
References
Citations follow AMA style, numbered in order of first appearance.
- Goldstein PJ, Brownstein HH, Ryan PJ, Bellucci PA. Crack and homicide in New York City, 1988: a conceptually based event analysis. Contemp Drug Probl. 1989;16(4):651-687.
- Greenfeld LA. Alcohol and Crime: An Analysis of National Data on the Prevalence of Alcohol Involvement in Crime. US Department of Justice, Bureau of Justice Statistics; 1998.
- Kuhns JB, Exum ML, Clodfelter TA, Bottia MC. The prevalence of alcohol-involved homicide offending: a meta-analytic review. Homicide Stud. 2014;18(3):251-270. doi:10.1177/1088767913493629
- Centers for Disease Control and Prevention. Alcohol-Related Disease Impact (ARDI) application: alcohol-attributable homicides.
- US Sentencing Commission. Cocaine and Federal Sentencing Policy. 1995.
- Office for National Statistics. Measuring drug-related homicide methodology: February 2024. Published February 2024.
- Midgette G, Davenport S, Caulkins JP, et al. What America's Users Spend on Illegal Drugs, 2006-2016. RAND Corporation; 2019.
- Barratt MJ, Ferris JA, Winstock AR. Use of Silk Road, the online drug marketplace, in the United Kingdom, Australia and the United States. Addiction. 2016;111(5):774-783.
- Patton DU, Lane J, Leonard P, Macbeth J, Smith-Lee JR. Gang violence on the digital street: case study of a South Side Chicago gang member's Twitter communication. New Media Soc. 2017;19(7):1000-1018.
- Liem M, et al. Revisiting Goldstein's drugs-violence nexus: expanding the framework for the globalized era. Int Criminol. 2025.
- Bureau of Justice Statistics. Prisoners in 2023: Statistical Tables. US Department of Justice; 2025.
- Prison Policy Initiative. Following the money of mass incarceration. 2026.
- American Civil Liberties Union. A Tale of Two Countries: Racially Targeted Arrests in the Era of Marijuana Reform. 2020.
- Rydell CP, Everingham SS. Controlling Cocaine: Supply Versus Demand Programs. RAND Corporation; 1994.
- Caulkins JP, Rydell CP, Schwabe WL, Chiesa J. Mandatory Minimum Drug Sentences: Throwing Away the Key or the Taxpayers' Money? RAND Corporation; 1997.