President Donald Trump is upset that anti-racism and anti-police brutality protesters remain on the streets of America’s cities. He blames local leaders, mainly Democrats, for not doing enough to quell the unrest, which he typically portrays as far more violent than it really is.

Local and state officials should “dominate” violent protesters, he says, by using overwhelming force to deter anyone from countering law enforcement again. “You have to dominate or you’ll look like a bunch of jerks, you have to arrest and try people,” Trump told US governors on a June phone call. “You don’t have to be too careful.”

Trump has put his theory into practice in Portland, Oregon, where he sent federal agents in early July nominally to protect federal property in the city, but really to put down demonstrations that have lasted for several weeks.

Officials from US Border Patrol and other federal agencies, most of whom are not well-trained for handling mass demonstrations, have shot protesters in the head with “less lethal” munitions, launched tear gas, and arrested citizens far from federal property after stalking them in unmarked vehicles.

The problem for Trump, though, is that the seductive logic of quelling demonstrations using immense force has proven faulty. More of Portland’s residents are on the streets now than before federal agents arrived, and violent incidents have ticked upward, not downward. Despite those results, the Trump administration shows no signs of changing course in Portland, and may even export the strategy to other major American cities like Chicago.

When reporting on the federal response to the Portland protests this week, I interviewed Arizona State University’s Edward Maguire. He co-authored a well-regarded guidebook on how police should handle protests, based on research from past events in the US and around the world. What he told me then was that the incidents in Portland weren’t a surprise — in fact, they were likely to happen precisely because Trump’s idea of indiscriminately employing violence to quash protests is backward.

I called Maguire two days later to dig deeper into why, exactly, that was the case. It turns out police actually need to connect with demonstrators, letting them know they are there to protect their First Amendment rights and keep them safe.

This task should be left to local authorities, not federal ones, since they should already have some relationships with the community. But if federal agents do come in, they shouldn’t aim to exacerbate the situation. “You need to win hearts and minds,” Maguire urges authorities.

Our conversation, edited for length and clarity, is below.

Alex Ward

Why has policing, and the federal government’s approach, gone so wrong in Portland?

Edward Maguire

It’s part of a much larger problem. Generally speaking, in the United States we prepare police officers for the more extreme end of civil disorders: riots. We’re typically not doing a good job of preparing them for the less extreme end of civil disorders, basically any crowd event that hasn’t devolved into a full-blown riot.

That affects officers’ thinking in these instances, and that’s a major issue. What we know from crowd psychology and just decades of studying this stuff is that when police prepare to treat these types of events as riots, they can actually instigate a riot.

Alex Ward

So planning and preparing to put down a riot, where one doesn’t exist, may actually ending up creating one.

Edward Maguire

That’s right. What we need to do in handling these protests, then, is really understand not only how crowds function but also how crowds react to what the police do. What we have right now is law enforcement agencies under the Department of Homeland Security in Portland just ignoring everything we know about how to do protest policing right.

Alex Ward

But one has to admit there’s a seductive logic to having law enforcement respond forcefully to a protest to deter further participation. Plus, in Portland there have been some cases of people setting fires, looting stores, and throwing rocks at federal officials. That’s violence, and it seems intuitive to meet that violence with overwhelming power so it stops.

Yet you’re saying that reasoning — as seemingly straightforward as it sounds — isn’t effective?

Edward Maguire

Portland is a fascinating case study because the intensity of the protests and the numbers out on the streets had decreased pretty substantially before federal law enforcement showed up. Object throwing, fire setting, graffiti — all that increased dramatically after federal authorities arrived. So I don’t think anybody can make the argument that the federal authorities came to town and made things better.

Just look at some of the violations Homeland Security cited in the July 16 statement. Before federal authorities arrived, a lot, but not all, of what they acknowledged was graffiti.

Then check out later sections of the statement — after law enforcement arrived — and you’ll see that the violations got more serious.

This is completely consistent with what we know about crowd policing, and in many ways what the federal authorities have just done for me is illustrated all of the theories that I talked about in my guidebook for police. They’ve done a fantastic job of illustrating everything that I say not to do.

Alex Ward

And what are those things they shouldn’t do?

Edward Maguire

Well, the first thing to keep in mind is that these events are best handled by local authorities, not federal ones. The reason why is actually quite simple: If local police are playing the game correctly, they should already have a relationship with the leaders of the protest movement as well as other social justice groups.

The hope is that having those kinds of relationships prevents the situation from spiraling in the first place. But if it does start to go wrong, authorities can leverage those preexisting relationships to simmer things down pretty quickly.

Alex Ward

The implication of that is that outside authorities won’t have the ability to reach out to the aggrieved groups as easily, thus rendering them less effective. Right?

Edward Maguire

Exactly. There are instances when police are responding to events with big crowds and they feel like maybe they’re losing control a little bit. That happens. In those cases, they call for “mutual aid,” which allows them to bring in outside forces to support them — but often, mutual aid situations end up turning badly.

Just look at what happened during the 2014 protests in Ferguson, Missouri. We saw about 50 or 60 law enforcement agencies responding to the situation, all of which have completely different training, cultures, equipment, policies, and procedures. There was this mishmash of police coming in from multiple jurisdictions, and that’s often a recipe for disaster.

That’s where we typically see policing go badly: outside forces coming in, not knowing the area or the people, and failing to do the job correctly.

Alex Ward

Give me an example of when the dynamic you talk about — local police forming relationships closely with protest leaders and without outside help — worked well.

Edward Maguire

In 2011, Occupy Wall Street took over Pioneer Park in Salt Lake City. Police and protest leaders purposefully formed a relationship so the situation would remain more peaceful than it had in other cities. When I interviewed the then-deputy chief, I asked him how he was able to maintain the peace. His answer, and I’m paraphrasing, was “I bought gallons of coffee for those guys.”

Right across from the park was a coffee shop, and he was routinely going down to the protest and taking leaders of the movement there. He was always talking with them and asking, “What do you need? Anything going wrong? Is there anything I need to work on? Is there anything I can provide?” That kind of interpersonal dynamic is what stops violence.

Alex Ward

But say there is violence. Buying coffee for people isn’t necessarily going to do the trick, is it?

Edward Maguire

Let me use a personal analogy: I have a 9-year-old daughter. If I do anything wrong, she’s going to call me out on it. If she starts yelling at me, it would be an inappropriate response on my part to start yelling right back at her. You don’t increase tensions in response to the other party having ramped up tension. It just doesn’t work.

And if it doesn’t work in a one-on-one situation with my daughter, it’s obviously not going to work in a crowd-control situation, which is orders of magnitude more complicated.

Alex Ward

So what should federal officials in Portland or elsewhere be doing instead of ratcheting up tensions?

Edward Maguire

If you’re an authority figure dealing with subordinates right now — like federal and local officers in Portland dealing with protesters are — what you need to do is reduce tension. Calm everything down, and realize that what you’re doing is not working. That takes a kind of humility, but taking a step back will make you realize you’re making the problem worse by not ceasing and desisting from the status quo.

Meanwhile, authorities need to be engaging in tons and tons of communication. They need to reach out to the protest communities and say, “Hey, obviously this isn’t working. We need to talk about how to lower tensions here.” Engaging in a softened dialogue is key.

Alex Ward

Surely, though, there will be moments when arrests need to be made. How might that impact the hoped-for comity?

Edward Maguire

If there are people who are engaging in violence, of course they need to be arrested. But if the only response to the violence or other forms of illegal behavior is to rush in and make arrests, it will make things worse.

No credible authority figure on policing protests is suggesting police shouldn’t arrest people who are engaged in violence or property damage. In fact, it’s the opposite: When people are engaged in that type of behavior, they need to be arrested and preferably now, not later.

But here’s what virtually every police agency misses: There’s just so much room for rumors and misinformation in a crowd event — particularly one already upset with police — that authorities need to have a really high-quality public address system by which they can communicate with the crowd.

The communication that occurs through that PA system should not only be enforcement-related communication. What the crowd needs in that moment is validation, that those among them who are behaving peacefully and lawfully and expressing their First Amendment rights will still be allowed to do so.

That’s what protesters need to have confidence in: that if we’re out here exercising our rights, these knuckleheads aren’t going to shoot us with rubber bullets and chemical agents. The most potent form of communication the police can engage in, then, is to let protesters know they will facilitate a protester’s right to be there and behave peacefully and lawfully. However, the message should get across that those among the crowd who choose to engage in violence and property damage will be arrested.

That just needs to be a routine and repeated message. And if an arrest is made, authorities need to tell the crowd why they detained someone, because otherwise misinformation will spread.

Alex Ward

It seems these days that many police bureaus don’t understand the kind of playbook you just described, or the research behind what actually animates the crowds they face. Why aren’t they learning these pretty straightforward lessons?

Edward Maguire

Some surely do. I happened to be having lunch with an official from the Tucson Police Department, and I mentioned I study crowds, protests, and policing. All of a sudden he started talking about the “social identity model” and its implications for protest policing in the United States. I wanted to jump out of my seat and cheer. It was the first time that’s ever happened to me.

I mean, I know some officers know this stuff because I’ve taught it to them, but this is the first time I heard of an officer just finding his way to this literature. Clearly some see its utility to be better officers.

Alex Ward

What is the social identity model?

Edward Maguire

Based on the validated research, the social identity model suggests people in crowds have different social identities. The idea, basically, is that we all have multiple, different social identities. You and I share one, actually: We’re Red Sox fans. But being in a crowd can change that momentarily; we have fleeting identities in crowd events.

Here’s a great anecdote on just this point: I was in Washington, DC, watching a Nationals [baseball] game. I’m not a Nationals fan at all, obviously, but I went with friends. Anyway, a Nationals player hit a home run, and I temporarily lost my mind. I stood up. I was yelling. I was screaming.

When I sat down, I realized what had happened. Being in that stadium among Nationals fans, I was immersed in the crowd around me. At that moment, I was a Nationals fan and I shared that identity with those around me.

So anyone can show up to a protest with an infinite number of social identities. But in a crowd event, you can be heavily influenced by the social identities of the people around you. That’s not to say it’s a contagious effect or predetermined, but it can happen.

Alex Ward

I’d assume, then, that if law enforcement attacks a crowd that shares a sense of identity, the crowd will feel attacked as a unit.

Edward Maguire

Yes, and it’s counterproductive. Demonstrations usually have moderate protesters in them, people who are generally peaceful, law-abiding, and aren’t prone to engaging in violence or property damage. Of course, there are also typically some anarchists or people ready to cause trouble. The moderates and troublemakers may not agree on tactics, but they at least share a cause.

If law enforcement attacks the group as a whole for the actions of a few troublemakers, the moderates can start to embrace — or at least understand or appreciate — the strategies of those who are more radical. At that point, the crowd psychology shifts.

This is a key point, and where my recommended approach really comes into play: Law enforcement should want to avoid that shift. You want the moderates to stay moderate. You don’t want the fleeting social identities of the moderates to start to drift toward the social identities of the radicals.

The way I put this for police officers to understand is: “You need to win hearts and minds.” You’re not going to succeed with the radicals who are engaging in property damage, of course, but you just might win over the majority of the crowd.

Alex Ward

For the police forces and federal agents not familiar with this approach, or who haven’t had a chance to peruse the literature, what are your three main recommendations for how law enforcement should handle protests like the ones we’re seeing across America?

Edward Maguire

First, communicate with protesters before, during, and after protest events. Form relationships with these people. Work with them, talk to them, negotiate with them. That’s not to say give away the farm: Set the priorities that you need to set, but communicate with these people. [These are] people who are intent on First Amendment expression, so law enforcement should be seen as people who facilitate that expression, not block it.

Second, don’t take enforcement action against an entire crowd in response to the misbehavior of a small number of people in the crowd. If people are engaging in illegal behavior, by all means make arrests, but continue to facilitate the First Amendment expression of the people in the crowd.

Which brings me to my third point: Law enforcement should be seen as a guardian of the First Amendment. Not the guardian of a building, not the guardian of a skirmish line, but of the First Amendment. Embrace the constitutional freedoms these people enjoy, and only take enforcement action against those people who are behaving in a lawless manner.