At a town-hall-style forum hosted by Sean Hannity and airing Wednesday night, Donald Trump was asked what he’d do about black-on-black crime. His answer, reported by NBC’s Alexandra Jaffe, is worth reading in full:

Right, well, one of the things I’d do, Ricardo, is I would do stop-and-frisk. I think you have to. We did it in New York, it worked incredibly well and you have to be proactive, you know, you really help people sort of change their mind automatically, you understand, you have to, in my opinion, I see what’s going on here, I see what’s going on in Chicago, I think stop-and-frisk. In New York City it was so incredible, the way it worked. Now, we had a very good mayor, but New York City was incredible, the way that worked, so I think that could be one step you could do.

Trump’s answer comes in the context of his latest series of events aimed at black voters, and nearly every sentence here offers something to think about.

First, stop-and-frisk is already in place in Chicago and other cities, making this idea in keeping with Trump’s habit of suggesting policies, such as “extreme vetting” of refugees, that closely resemble practices that are already in place. (Stop-and-frisk is not federal policy, but it is practiced by police departments across the country.) Second, the best studies suggest that stop-and-frisk does not effectively reduce crime where it is used. Third, court decisions and settlements have acknowledged that the methods used in both New York and Chicago were unconstitutionally discriminatory, setting aside their efficacy. Fourth, one of the two New York mayors who oversaw the implementation of stop-and-frisk, Michael Bloomberg, has blasted Trump, saying, “I'm a New Yorker, and New Yorkers know a con when we see one.”

Moreover, it’s hard to take the proposal seriously as outreach to the black community. National polling on stop-and-frisk is tough to come by, but both anecdotal and statistical data from New York suggest that black citizens view the practice as discriminatory and dehumanizing. In a 2012 Quinnipiac poll, seven in 10 black New Yorkers opposed stop-and-frisk. In 2013, Marist found an even higher proportion, 75 percent, wanted an overhaul.

Trump’s supposed black outreach has taken place to a great degree in white communities, before white audiences, while his appearances in African American communities have not always gone so well. His advocacy for stop-and-frisk offers more evidence for the view that Trump’s goal is not so much to court black voters but to convince white ones who are rattled about crime to back him.

This is one of the many contradictions of Trump’s recent moves. Even as he campaigns on the basis of “law and order,” he appeared at events in Ohio Wednesday with Don King, the boxing promoter who shot and killed a man (the death was excused as justifiable homicide) and served time for manslaughter after stomping a man to death in 1966. King’s speech in Cleveland was a strange, sporadically coherent stream of consciousness, in which he said, “I want you to understand, every white women should cast their vote for Donald Trump, not for Donald Trump the man but to knock out the system.”

One important question is what the Republican presidential nominee means by “stop-and-frisk.” As Donald Braman wrote in The Atlantic in 2014, “No one thinks a police officer with a reasonable suspicion that a suspect has a gun should be barred from frisking the suspect, but that is not what stop-and-frisk has come to mean,” i.e., stopping vast numbers of people for vaguely “suspicious behavior,” then searching them. Since Trump cited New York City, it seems reasonable to assume that’s what he meant.

Stop-and-frisk is not inherently unconstitutional, but the recent record shows that it’s hard to execute without racial discrimination. In New York, a federal judge deemed the city’s approach unconstitutional. Of city officials, Judge Shira Scheindlin wrote, “In their zeal to defend a policy that they believe to be effective, they have willfully ignored overwhelming proof that the policy of targeting ‘the right people’ is racially discriminatory and therefore violates the United States Constitution.”

Something similar happened in Chicago, where the American Civil Liberties Union of Illinois documented a pattern of racially disparate enforcement. The report came in March 2015. Among its findings: Blacks made up three-quarters of stops but less than a third of the city's population. Chicago’s police department decided to reach an agreement with the ACLU to install closer oversight.

Proponents of stop-and-frisk argue that disproportionate stops are a regrettable but forgivable and necessary side effect of the practice. Poor black communities are often plagued by high crime rates, they say, and so it stands to reason that police would be making the most stops in the places where crime is highest. This may or may not be true. A Columbia study found that racial disparities existed even when controlling for location.

But in any case, what happened in New York undercuts the theory that stop-and-frisk is a necessary tool for reducing crime. New York officials credited a huge drop in crime to aggressive use of stop-and-frisk, though hard data told a different story. In 2013, prior to New York effectively ending stop-and-frisk, David Greenberg of New York University found “no evidence that misdemeanor arrests reduced levels of homicide, robbery, or aggravated assaults.” Since the end of stop-and-frisk, crime has remained at historically low rates or even dropped further. In 2015, murder rebounded slightly from its all-time low in 2013, though the first quarter of 2016 was the lowest on record.

Chicago, of course, has not seen such a pacific stretch. The violent crime rate in the Windy City remains extremely high. But there’s little indication that stop-and-frisk plays much role in that. Murder rates were already high before the August 2015 agreement on stop-and-frisk, which produced a sharp drop in stops. There was no clear resulting uptick in crime, which perhaps should be no surprise: WBEZ found that data from the stop-and-frisk era “show negative trends as officers reported more stops: Gun seizures dropped, detectives solved fewer murders, and a decade-long decline in gun violence ended.”

But crime in Chicago did spike after the city released (under duress) video of an officer shooting and killing Laquan McDonald. One reason for that is that arrests declined, as police retrenched. Another logical reason is that the sight of a police officer brutally killing a man—McDonald was not approaching officers, was not threatening Officer Jason Van Dyke, and was shot several times as he lay on the pavement—undermines trust in the legitimacy of rule of law, as did the realization that many police officers had relayed misleading reports of McDonald’s shooting, validating Van Dyke’s version of the encounter.

A January 2014 paper by professors at Yale Law School, Columbia Law School, and Columbia University concluded that stop-and-frisk undermined young black men’s faith in the legitimacy of the police, too:

Whether the police were viewed as exercising their authority fairly and lawfully shaped the impact of stops on respondent’s general judgments about police legitimacy. Fairness and lawfulness judgments, in turn, were influenced by the number of stops and the degree of police intrusion during those stops. Similarly, judgments of justice and lawfulness shaped the estimated influence of judgments of the general character of police behavior in the community on general perceptions of police legitimacy.

There’s some grim irony to be found here. African Americans, as well as many white ones, are currently outraged, horrified, and despairing over the deaths of Terence Crutcher and Keith Lamont Scott, the types of shootings that provoke widespread fear and distrust of the police around the country. It is almost unbelievable that against such a backdrop, and under the guise of outreach to black Americans, Trump would propose the expansion of a policy that also undermines black trust in the police. Unless he isn’t reaching out to them at all.