Rick Wilking / Reuters

1974 was a tough year in American history: a savage recession, inflation, high gas prices, the resignation of a president, looming defeat in Vietnam. Yet when surveyed about that year, Americans listed none of these as the nation's top problem. That unhappy distinction was awarded instead to crime.

By the mid-1970s, crime rates had been rising for a decade and a half. Crime was reshaping the physical structure of American life, chasing middle-class people out of cities and into suburbs where they hoped urban crime would not follow. Crime poisoned race relations. Up until the mid-1960s, white Americans expressed broad support for measures to open opportunities to black Americans, both in opinion surveys and by voting for civil-rights-supporting politicians like President Lyndon Johnson and Governors Nelson Rockefeller of New York, Pat Brown of California, George Romney of Michigan, and Otto Kerner of Illinois. As the industrial cities of the North and California were rent by riots and criminality between 1965 and 1970, white opinion progressively hardened. Richard Nixon's Attorney General John Mitchell predicted in 1970: "This country is going so far right you won't recognize it." His prediction proved prophetic.

The surging crime wave of the 1960s tempted big-city mayors and police chiefs overwhelmed by local record-keeping. Citizens ceased reporting minor thefts to police forces that never caught the thieves. It was widely suspected too that mayors and chiefs massaged their numbers to improve the image of their cities. In 1973, the federal government introduced national surveys of criminal victimization. The first report, delivered in 1974, found that one out of every three American households had suffered a crime of violence or major property crime in the previous year. (I take all these lurid details from my own history of the period, How We Got Here. )

That was bad, and worse was to come. Murder, rape, robbery: Year after year, Americans suffered more and more in almost every category of crime. A new generation of "tough on crime" politicians and prosecutors tried everything they could think of, and yet nothing worked. They sent more criminals to prison for longer. They reintroduced the death penalty. They hired more cops, equipped them more lavishly, and trained them better. They loosened gun laws, created the national 911 system, and on and on. Still the crime rates rose, reaching an all-time peak in 1991.

And then ... something happened. Between 1991 and 2015, rates of criminal victimization dropped by half, with the most violent offenses—murder and rape—declining the most precipitously. An American citizen is less likely to be a victim of crime today than at any time since good records began in the 1970s. New York City is almost certainly safer today than at any time in its history. And the weird thing is that nobody quite knows why it all happened. Ever tougher criminal sentencing helped. A single criminal will commit many crimes over his career. The more of that career he spends behind bars, the fewer crimes he'll have opportunity to commit.

More sophisticated policing surely helped. "Broken windows" and community policing got cops out of their cars and onto their beats. CompStat systems directed police resources where they'd do the most good.

Many other technological and social factors played their part too. Some have nominated the aging of the population, or the phasing out of lead paint, or new immigration flows, or improvements in educational levels, or the decline in use of crack cocaine. Some hardcore proponents credit more abortion of unwanted children or more concealed carry of private firearms. Yet the uncomfortable truth is that there's remarkably little clarity about the causes of this hugely important social improvement. Almost nobody predicted it beforehand. Even after it happened, social scientists cannot agree on why.

Stephen Levitt of the University of Chicago in 2004 credits more prisons and more cops for the largest part of the drop in crime. In the opposite corner, the Brennan Center at New York University argues that prisons deserve little credit for the decline in crime in the 1990s and actually became counter-productive in the 2000s. If not prison, then what? The Brennan Center seems baffled. It attributes nearly half the crime drop of the 1990s—and more than half of the crime drop of the 2000s—to mysterious “other factors."

Settling the dispute may ultimately prove unresolvable, for reasons explained by Friedrich Hayek in his 1974 Nobel Prize lecture, “The Pretense of Knowledge”:

Unlike the position that exists in the physical sciences, in economics and other disciplines that deal with essentially complex phenomena, the aspects of the events to be accounted for about which we can get quantitative data are necessarily limited and may not include the important ones. While in the physical sciences it is generally assumed, probably with good reason, that any important factor which determines the observed events will itself be directly observable and measurable, in the study of such complex phenomena as the market, which depend on the actions of many individuals, all the circumstances which will determine the outcome of a process ... will hardly ever be fully known or measurable. And while in the physical sciences the investigator will be able to measure what ... he thinks important, in the social sciences often that is treated as important which happens to be accessible to measurement.

Yet even if we can’t know which factor contributed exactly how much to the drop in crime, what we can say is that a complex of causes taken together has produced a hugely positive result—probably the single most-successful policy outcome in the United States in the past quarter-century. Americans are doing something right, even if they don’t precisely know what that something is. And because they don’t know what that something is, they had better be very, very careful about any changes it’s proposed to make.

Change is in the air, however. For one thing, the present U.S. approach to crime-fighting is hugely expensive. A year in prison costs state taxpayers an average of $31,000. In California, the cost rises to $47,000, thanks to higher health costs and a powerful prison-guard union. Other democracies have achieved considerable crime reduction since the early 1990s without imprisoning anything like so many people. Look at the U.K. for example: Britain imprisons only 85,000 people, as compared to more than 2.2 million in the United States. Australia and Canada show similar trends, likewise with smaller prison populations.

Imprisoning very large numbers of people also carries heavy non-economic costs. Men who are sent to prison face tremendous difficulty returning to the job market. They lose contact with their children, if they have them, who in turn are much more at risk of going to prison. Every aspect of the criminal-justice system reveals racial disparity. 60 percent of those in prison are black or Latino. One in three black men will pass through prison at some point in his life. Black ex-convicts face permanently lower lifetime wages as compared to white ex-convicts.

So it’s understandable that politicians from Hillary Clinton to Rand Paul would call for a different approach. But what?

Imprisonment rates are already declining, down from a peak in 2007. The majority of people in state prison, more than 53 percent, have been convicted of a violent crime: murder, rape, or robbery. Only about 3.7 percent of the state prison population has been sent there for drug possession alone. (In the much smaller federal system, drug offenses loom larger—but federal drug prisoners are overwhelmingly professional drug dealers, not casual possessors.)

Putting such people in prison and keeping them there is a harsh, crude, and expensive way to protect society from them. But the suggestion that less prison would leave society no less safe is dangerously glib. The last time the political pendulum swung away from incarceration—in the liberal decade from 1960 to 1970, the total number of prisoners dropped outright, and much more in relation to population—the country got in return the most serious crime wave since Prohibition.

The advocates of de-incarceration in the 1960s believed they were acting generously. They recognized that individual criminals reflect larger social problems, racism most fundamentally. Young and poor black Americans, followed by the newer immigrant Latino Americans who are joining them at the bottom of the American social hierarchy, grow up in circumstances weighted against them in ways deeply rooted in American history. De-incarcerators argued then, and argue with renewed ardor now, that tossing malefactors into dungeons for decades does nothing to address the true causes of crime.

And surely they have a point. Yet while the history of discrimination contains powerful explanations of crime, it offers little guidance about what to do about crime. Even if racism shaped a murderer, anti-racism won’t undo the murder—or prevent the murderer from killing again. The first task of government is to protect the lives and property of citizens. Governments that cannot or will not perform that task forfeit their legitimacy—as will, by the way, any cause or movement that argues that citizens must accept a higher risk of violence or robbery as their contribution to somebody else’s ideal of justice.

Diverting people from prison where we can is the right thing to do. Testing shorter sentences and easier parole for less dangerous offenders is feasible in today’s successful crime-fighting context. Prisons must be made more humane, and prison rape should be addressed as the national shame and scandal it is. Easing the return of prisoners to society—and mentoring their families until the prisoner returns—should be a front-of-mind issue for charities and philanthropy. But given how poorly we understand why and how crime was reduced in the first place, the debate about de-incarceration should proceed with caution and humility. Mass incarceration is bad. Mass-criminal victimization is worse.

And for those who think politically, a reminder: It was crime more than anything else that sent the country hurtling to the right in the 1970s. It was the decline of crime that enabled the revival of the liberal left in the 2000s. The fate that Hillary Clinton is tempting with her about-face on crime and punishment—it may be her own.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.