Yesterday was sunny and warm in New York, and Mayor Bill de Blasio was announcing his appointment of an impressive group of people to criminal justice posts. After weeks of tough headlines, “Today is a good news day,” the mayor declared, in part because of some cheery crime statistics. As the mayor explained:

During the first 10 weeks of 2014, the NYPD has driven down already historically low levels of crime. Overall, major crime has gone down 2 percent in the first 10 weeks of 2014 compared to where the levels stood a year ago. That’s overall—that’s all major crime categories—a 2 percent decline. But let’s talk about homicides. Homicides are down nearly 21 percent from this time a year ago. Shooting incidents down more than 14 percent from this time a year ago. This is extraordinary progress. It is not surprising to me, given that we have the finest police force in the world, that this progress is made. It’s not surprising to me, given that we have the finest police leader in the world, that this progress has been made. Some naysayers suggested that you couldn’t bring down crime while bringing police and community back together. I think these last 10 weeks show—yes, you can and yes, we will. And I just want to thank Commissioner Bratton and all the men and women of the NYPD for their extraordinary efforts. This is real evidence of what they can achieve and will continue to achieve.

Given the way de Blasio’s opponents in the fall predicted bloodshed and mayhem upon his taking office, the mayor deserves a little room to crow. And he was quick to give street cops the credit they deserve rather than claiming all of it for his polices.

But regardless of who gets credit or blame, it’s unwise to put too much weight on small changes in the crime rate, or on trends over a short period of time—because when the numbers turn in the other direction, the cheering will turn to panic. Time and again in recent years, the city’s tabloid press has reacted to mini-spurts of crime as if they augured a return to the “bad old days.” In the end, the trends never last, and the overall crime rate keeps falling. In fact, just about a month ago, the Post was screaming about the 33 percent increase in murders during the first month of the de Blasio administration.

The very lowness of the key crime statistic—the murder rate—only enhances the danger of playing the stats game. If I went out and killed fifteen people in 1990, it would have increased the murder count by about six-tenths of 1 percent. If I go out and do that today (which is extremely unlikely, but can never be totally discounted), the murder rate this year might go up 5 percent as a result. When the base number is low, small changes look very large.

That’s not to say something very cool isn’t happening: Over the past year, even before the federal court ruling, even before de Blasio took office, the number of stop-and-frisk encounters was collapsing but—contrary to those naysayers—crime kept falling. The ten weeks the mayor points to are part of that longer trend.

And that trend, in turn, is part of something even larger: For two decades, under four different mayors, through economic booms and busts, violence and lawlessness have been decreasing in New York and throughout most of the country.

Focusing on short-term peaks and valleys misses that broader landscape of policy success. Beyond that, the stats obsession feeds back unhelpfully into politics and, then, into policy. When the Giuliani administration began its intense focus on crime statistics, with Bill Bratton and Jack Maple taking the lead, the numbers were intended to serve as a tactical guide for police commanders. But they became a politically charged barometer for mayoral performance. It’s hard to see how that eventually led to situations like the scandal in the 81st Precinct, where a cop recorded his commanders ordering officers to take steps to discourage the reporting of crime.

This is the dark side of the ascendance in the past decade of the use of metrics in government. It’s great to hold government accountable, and the numbers can help with that—I use them ad nauseam in my reporting. But when they become the lone, nearly instantaneous indicator of whether a government is succeeding or failing, they’re dangerous. Numbers can be manipulated, and can be understood to say more than they really say. And they can go from being a tool to being a master.