Community Response Teams As Alternatives To Policing

Roughly 240 million 911 calls are made in the United States each year. Credible analyses — including ones by police departments themselves — consistently find that somewhere between 50% and 80% of calls involve no actual crime. They involve noise, welfare checks, mental health crises, substance use, homelessness, neighbor disputes, traffic issues, minor medical events, and confusion. Of the calls that do involve a crime, the majority are property crimes or minor disputes, not violence in progress.

This is well-known inside policing. It has been well-known for fifty years. The 1967 President's Commission on Law Enforcement noted it. So did the Kerner Commission in 1968. So did every major policing study since. The profession has had a long internal conversation about how much of what officers do is not really police work. The problem is that the dispatch infrastructure — one number, one kind of responder — never changed to match that knowledge.

What CAHOOTS did, starting in 1989, is just change the dispatch.

The program was born out of the White Bird Clinic, a community health clinic founded in the counterculture moment of 1969 in Eugene. In the late 1980s, White Bird staff and Eugene police sat down together — this is worth emphasizing, because it is usually missed — and agreed on a protocol where 911 dispatchers could route certain calls to a White Bird van instead of to an officer. The agreement was practical, not ideological. The police were being pulled away from serious calls to handle things officers were not trained for and did not want. The clinic knew how to do those calls. Dispatch would sort.

A CAHOOTS team is two people: a medic (often an EMT or nurse) and a crisis worker (usually with experience in mental health, addiction, or social work, and often with lived experience of the systems clients are moving through). They work 24/7 in vans, unarmed, unarmored, dispatched through the same 911 and non-emergency lines as police.

Typical calls: - Welfare checks on an unhoused person - Mental health crises — someone in psychosis, suicidal, or in a severe panic - Intoxication — someone drunk in public, overdosing, coming down hard - Minor medical — a cut, a diabetic episode, a fall - Family disputes without violence - Grief calls — a family member has died and someone needs support - Dispositions — transport to shelter, detox, clinic, ER

CAHOOTS handles roughly 15,000 to 24,000 calls a year depending on the year. It operates on a budget of about $2.1 million (2020 figures, slightly higher since), funded partly by the city of Eugene and partly by the city of Springfield. The Eugene Police Department estimates the program saves the city roughly $8.5 million a year, plus an estimated $14 million in ER costs. Those numbers come from EPD and White Bird jointly, which is the kind of joint-reporting situation that usually means the numbers are honest because nobody wants to exaggerate in front of the other.

The safety record is the number people come back to. Over decades of operation, there have been zero known responder fatalities from violence on calls, a near-zero rate of civilian injuries during CAHOOTS response, and a backup-request rate to police of 1–2%. The reasons are not mystical. Unarmed responders slow down. They don't escalate. They don't trigger the instinct in a person in crisis to fight or flee. They sit, they talk, they wait. Most crises are time-limited if you don't inflame them.

Denver STAR (Support Team Assisted Response), launched June 2020 as a six-month pilot. A University of Denver / Stanford evaluation (Dholakia et al., 2022, published in Science Advances) found that in the pilot area, reports of targeted minor crimes fell by 34% during STAR's operating hours, with no increase in more serious crimes. Zero arrests on STAR calls in the pilot period. Denver expanded the program; by 2023 STAR was operating across most of the city and responding to tens of thousands of calls a year.

Portland Street Response, launched 2021, initially focused on mental health and homelessness calls in one precinct. A Portland State University evaluation in 2022 found 4.6% reductions in police responses to non-emergency calls and extremely low rates of use-of-force or arrest on the calls they took. The program has been politically contentious within Portland — police unions have pushed back on expansion — but the outcome data has held up.

Albuquerque Community Safety Department, launched 2021, explicitly as a new cabinet-level department, not a nonprofit contract. That matters structurally; more on this below.

NYC B-HEARD, launched 2021 in Harlem, mental-health-specific teams of a social worker plus two EMTs. Early data showed higher rates of voluntary hospital transport and lower rates of arrest compared to police-only responses, but the program has been criticized as under-resourced relative to need.

Olympia Crisis Response Unit, Durham HEART, San Francisco Street Crisis Response Team — all operating, all reporting similar patterns: calls that go to unarmed responders end more peacefully than the same calls would have ended with police.

The cluster of results is consistent enough to matter. The mechanism is consistent enough to name: when the responder is not a threat, the situation is not a threat.

A police officer in the US is trained for roughly 700 academy hours on average. Of that, a small fraction is on de-escalation, mental health, or non-coercive conflict resolution. The bulk is on use of force, firearms, tactics, and law. Officers carry a gun at all times. Their legal framework — qualified immunity, stop-and-frisk doctrine, the Graham v. Connor reasonableness standard — is built around the assumption that they will at some point need to use force, and the law protects them when they do.

Contrast this with a CAHOOTS worker: hundreds of hours in mental health first aid, motivational interviewing, trauma-informed care, harm reduction, de-escalation, with ongoing supervision and reflective practice. No weapon. No authority to arrest. No legal incentive to dominate a situation. The worker's job is to make the person on the curb safer, not to impose order on them.

These are different jobs. Nobody sensible thinks a crisis worker should respond to an active shooter. And nobody should think an armed officer is the right call for a panic attack. The problem has been that until very recently, American cities routed both to the same van.

There is also a structural point. When the only tool you have is a person with arrest power, every problem gets solved by arrest. A person sleeping in a park becomes a trespass case. A person in psychosis becomes a resisting-arrest case. A child acting out becomes a juvenile case. You end up with a population of people cycled through jails and courts for behaviors that were never crimes — just expensive, traumatizing, life-derailing contact with a system that wasn't designed for them.

Across the programs with published evaluations, the recurring findings:

- Use of force on calls handled by community response: essentially zero. - Arrests on calls handled by community response: low single digits by percentage, often zero for entire reporting periods. - Responder injuries: low, and when they occur, usually minor. - Request for police backup: typically 1–3%. - ER transports: lower than comparable police responses, because crisis workers can deliver people to the actually-right destination (detox, shelter, clinic) rather than defaulting to the ER. - Cost per call: lower than police response, usually substantially — estimates range from roughly 1/3 to 1/2 the fully-loaded cost of a police response. - Total city savings: where measured, in the single-digit to low-double-digit millions per year.

The two most cited academic evaluations are the Science Advances paper on Denver STAR (Dholakia, Sukthankar, and Yakusheva, 2022) and the Portland State University evaluation of Portland Street Response (Zapata et al., 2022). Both are worth reading in full if you are trying to argue for this in your city.

A useful mental model: 911 is not a policing service. It's an emergency dispatch service. Dispatch is a routing function. The question is not "should police handle fewer calls?" but "what is the right responder for this call?"

Once you put the question that way, the answer emerges naturally:

- Violence in progress, weapons, active threats → police (or specialized tactical response). - Medical emergency → EMS. - Fire/structural → fire. - Mental health crisis without weapons → clinical crisis team. - Homelessness-related welfare check → street outreach team. - Substance use or overdose → harm reduction team (with Narcan, which cops usually don't carry anyway). - Neighbor dispute, noise, civil disputes → mediators or, frankly, nobody — some of these don't need a responder at all. - Traffic → traffic enforcement (which many cities are also separating from policing, since armed traffic stops are a major source of deaths and the job is largely clerical).

Most cities already have the pieces. What they don't have is the dispatch logic and the political will to route calls to something other than police.

Here's where it gets hard.

Police departments in the US are among the most politically organized institutions in civic life. Police unions, in particular, are disciplined political actors with substantial influence over mayors, city councils, prosecutors, and state legislatures. When cities propose community response teams, the public response from unions is usually one of three patterns:

1. "Fine, but we should run it." Turn the program into an internal police unit with co-response models (a clinician paired with an officer). This preserves the union's jurisdiction and payroll but reintroduces the armed responder to calls that don't need one. Co-response models have mixed outcomes; they are better than police-only but clearly worse than unarmed-only. 2. "Fine, but it can't take our calls." Limit the program to a narrow window of hours, a narrow geographic area, or a narrow call type, so that the program never grows large enough to threaten staffing levels. 3. "This will get someone killed." Predictions of responder deaths that have not materialized in Eugene over forty years or anywhere else in the data.

These patterns are not secret. They appear, with variations, in every city that has tried to build a community response program. The cities that have succeeded in scaling tend to share a few features:

- A strong community organization (often with a long track record) ready to operate the program, so the city isn't starting from zero. - Political leadership willing to take the fight with the union in public. - A budget mechanism that doesn't require taking money from police line items in year one (though eventually, if the program grows, it does — because police do less work). - Data infrastructure from day one, so outcomes can be reported honestly and defended.

The cities where it has stalled have usually lacked one of these, most often the second.

The deepest frame: how you respond to a stranger in trouble reveals what you think a stranger is.

If you think your neighbor having a breakdown is primarily a threat to public order — someone to be controlled, moved, subdued, and if necessary shot — then you will staff 911 with officers. If you think your neighbor having a breakdown is primarily a human being in pain — someone who needs someone to sit with them, calm them, and help them to the next safe place — then you will staff 911 with medics and counselors.

The data says the second framing produces better outcomes on almost every metric. Fewer dead people. Fewer traumatized people. Less expensive. Calmer streets. The first framing persists not because it works but because it is institutionally entrenched.

Law 1 — We Are Human — is not a slogan. It is a dispatch principle. When the call comes in, who do you send? The person who sees a threat, or the person who sees a person?

1. Find out what happens in your city when someone calls 911 with a mental health crisis. Is there a clinician-led response? A co-response model? Only police? Search your city's website, call the non-emergency line, or ask your council member. Write down the answer. It is the baseline you are starting from.

2. If your city doesn't have a program: find out which community mental health organizations in your area are large enough and well-run enough to operate one. Find out whether anyone on your city council has raised the idea publicly. Email one of them. The ask is small: a pilot in one neighborhood.

3. Read the Denver STAR evaluation (Dholakia et al., 2022, Science Advances) and the White Bird CAHOOTS FAQ and budget documents. You need to be able to cite specific numbers when someone says "this is unproven."

4. Audit, honestly, your own dispatch reflex. Next time you witness something on the street — someone yelling, someone slumped over, someone who seems off — ask yourself what you would want to happen. Would a van with two people in hoodies and a clipboard be more useful than two officers with weapons drawn? For most situations, the honest answer is yes. Sit with that honesty.

- White Bird Clinic. CAHOOTS Program: FAQ, Media, and Budget documents. whitebirdclinic.org/cahoots. - Dholakia, Nazish, Aaron Sukthankar, and Olga Yakusheva. "Evaluation of Denver's STAR Program." Science Advances 8, no. 25 (2022). Also summarized in the Vera Institute briefings. - Zapata, Marisa, et al. Portland Street Response: Year One Evaluation. Portland State University Homelessness Research & Action Collaborative, 2022. - Center for American Progress. "The Community Responder Model." 2020. americanprogress.org. - Vera Institute of Justice. "Behavioral Health Crisis Alternatives." 2022. vera.org. - Law Enforcement Action Partnership (LEAP). Reports on alternative response. lawenforcementactionpartnership.org. - Lum, Cynthia, et al. "Can We Really Defund the Police? A Nine-Agency Study of Police Response to Calls for Service." Police Quarterly 25, no. 3 (2022). - 911 call data and content analyses — Vera Institute's "911 Call Processing: Data, Analysis, and Reform" is a good entry point into how the call volume actually breaks down by type. - Balko, Radley. Rise of the Warrior Cop. PublicAffairs, 2013 — on the historical expansion of police roles and militarization.

The United States has been running an experiment for fifty years on what happens when you send armed responders to unarmed situations. The results are on the evening news every few days. Eugene, Oregon, has been running a counter-experiment for forty years. The results are that a couple of people in a van, without weapons, handling one in five emergency calls, produce fewer deaths, fewer arrests, fewer injuries, fewer dollars spent.

The scandal is not that the counter-experiment is radical. The scandal is that it has been sitting in plain sight for four decades and most cities still haven't copied it.

Every city with a 911 system can stand up a version of this. The question is not whether it works. The question is whether your city is willing to admit that most of what we've been calling "policing" was never policing in the first place — and to build the response the situation actually calls for.