Traffic stops have long been one of the gravest dilemmas in police reform, and the pop reaction has been calls to remove armed police from performing them, whether replaced by unarmed traffic enforcement agents, cameras or, to the truly aspirational, robots. They constitute the foremost incidence of interactions between police and the public. Dangerous driving can kill. And they sometimes turn angry, violent and tragic.

At the same time, it is indisputable that black and Hispanic people are disproportionately stopped and, because why squander an opportunity, subject to search. If one focuses solely on this fact, then it’s understandable why removing police from the scene would be simple answer. But if one looks at all the competing factors, it’s clear that taking cops off traffic stops trades one set of untenable problems for another.

Unarmed traffic enforcement agents? They’ll be abused. Beaten. Shot. At best, ignored. Yes, it works elsewhere, with a more compliant populace, but America ain’t elsewhere. And sometimes, a bad dude gets stopped, with a warrant, a dead body in the trunk or a disinclination to get a reckless driving ticket that day. It will not end well.

Cameras? Red light cameras are a disaster. They burden the car owner rather than driver. They can’t be challenged or cross-examined, even if the driver remembered what happened three months earlier. And they get it wrong. A lot. But there’s nothing to be done about it. Plus, it doesn’t do much to help the people who died in the car crash that happened six minutes after the camera took a pic of the reckless driver.

Robots? Please.

What’s shocking about these faux fixes is how many people who should know better embrace them, whether because they’re popular among the unduly passionate activists who believe their fantasies or the people who should know better just lack the guts to tell their activist pals they’re lovely people with dumb ideas.

But Neil Gross, sociology prof at Colby College, gave this dilemma some serious thought, seeing both the real problems and searching for viable fixes.

Police officers in the United States pull over more than 19 million vehicles annually, making vehicle stops the No. 1 reason for contact between citizens and the police. Studies carried out over many years show that Black drivers are stopped disproportionately, a gap that cannot be accounted for by factors like differential driving behavior or greater poverty, which might translate into more cars on the road with equipment violations. Research also suggests that when Black drivers are pulled over, they tend to be treated less respectfully by the police and are given less leniency.

Frequent, intrusive vehicle stops aren’t just an inconvenience. Beyond being a source of legal and even physical peril, and something that can get in the way of economic opportunity (since many jobs require travel by car), such stops, according to other studies, are a potent reminder to Black Americans of all the ways in which the full rights of citizenship remain denied to them.

It’s bad for everybody. It’s worse for black people. But short of fantasy answers, what can be done that might really work? Gross offers three fixes.

One thing that would make a big difference would be to end “pretextual” traffic stops. These are stops where a police officer harbors some vague suspicion that a driver may be involved in criminal activity — a suspicion so vague that it wouldn’t hold up in court. The officer makes the stop anyway, using as a pretext that the driver has violated a minor rule of the road.

The Whren decision was doctrinally sound and functionally an obvious nightmare. There are so many ways that a cop can claim a traffic violation that’s absurd to entrust cops to stop only when an objective violation occurs. Crossed a fog line? If the cops says you did, that’s good enough, and everything falls into the gutter from there.

But would a cop lie about such things just to stop some Mercedes driver by a black guy because what else could he be but a drug dealer? It could happen.

A second strategy would be to require written consent when an officer asks permission to search a driver’s car. (If the officer has probable cause, no consent is needed.)

Between the pressure of the lie, “what do you have to hide,” and the fear that refusal tells the cop you are a bad dude, people consent to searches all the time. Just as people are given the Miranda warnings, and still talk. No matter how many times we tell them to invoke their right to counsel and then shut up, they talk. They just can’t not talk. And they just can’t not consent, even when the police have no basis to conduct a search and all the driver has to do is say, “Sorry, I do not consent.”

Written forms, expressly stating the right to refuse consent, will help. People will assuredly consent anyway, because people do stupid things. But it will help.

A third reform has even more potential. Police departments these days are under considerable pressure to track racial disparities in their operations. Yet little is done with this information.

This is a bit tricky, as it’s meant to create internal incentives against cops engaging in biased policing, whether explicit or implicit. Given public pressure for police to not target people by race, creating a mechanism where both the brass and the public can identify cops who disproportionately stop minority drivers might give officers an incentive not to be that cop.

By making the race of drivers stopped by a particular cop more transparent, officers engaging in stops based on race won’t be able to hide. No cop wants his puss on the front page of the local paper as the cops who stops more black guys than any other. No sarge wants to be named his supervisor.

Will these three reforms end racism, make policing great again and restore our trust and faith in American institutions? Of course not. But they have real potential to help, while not opening the door to a new slew of problems that will compel the next round of fantasy reforms because the last round was simplistic and untenable.