Former Atlanta police chief Erika Shields was talking about crime statistics with me a few years ago, as we were reviewing yet another year of small improvements in the homicide rate.

We had a sense that the fall in violent crime in Atlanta, like the rest of the country, had become a natural progression. Crime fell as the product of small iterative improvements in policing and underlying sociological changes — more people graduating from high school, lower unemployment rates, improvements in mental health care, and other effects that no one could put their finger on.

The city of half a million had fallen under 100 murders in a year. It was still too high of course — it’s always too high — but at 17 per 100,000 people in 2017, it was small enough to be sensitive to a wild swing with a cluster of crimes.

“All it takes is a couple of bad days to erase it,” she said. Or so we believed.

Atlanta is not looking at a couple of bad days, today. The number of homicides in Atlanta rose 89 percent year over year, as of last week. Put another way, we have 81 more dead people here than we should have expected. The aggravated assault rate has risen 24 percent, driven by a 50 percent increase in gun violence. It is one of the largest increases in violence in the country.

More people are trying to kill other people here. We should be asking ourselves why.

Right now, I think one reason might be because Atlantans no longer trust the police as much as they once did. I say this because many Atlantans stopped calling the cops last year, even as violent crime was rising.

Last year, I received two months of detailed 911 call data showing time and address, length of response time, type of disturbance and the names of callers, from May 11th to July 10th, and the same period from 2019 for comparison.

You might expect the number of calls to 911 to fall when no one can go outside. It’s hard to rob a house when everyone is at home. Fewer cars on the road means fewer car accidents. And the rest of Atlanta’s crime statistics bear that out: robberies, burglaries and other crimes fell off a cliff, down 29 percent year after year.

But look at the massive drop-off in calls on the first night of the George Floyd protests, and then again the day after Rayshard Brooks was shot by an Atlanta police officer.

From June 13th on, call volume fell 37 percent below the average of the preceding year.

This mirrors the results of academic studies that look at the public reaction to high-profile police misconduct cases. Essentially, there's a correlation between a big media profile police brutality case, a fall in 911 volume and a concomitant rise in violent crime. The theory is that people lose trust in the police and stop calling them to intervene in disputes ... which then develop into acts of violence that might have otherwise been prevented. 

Consider some of Atlanta’s unique characteristics. Atlanta has the highest inequality measured in America, according to a Bloomberg analysis. That plays out with people struggling to make the rest across the city, living with extra roommates in violation of their leases or squatting in abandoned properties. Atlanta is all but defined by in-migration: five out of six Atlanta residents were born more than 50 miles away.

Now layer the pandemic on top of this.

My understanding is that the CDC eviction moratorium did not prohibit evictions on the basis of lease violations other than nonpayment, if the tenants were not otherwise high risk for COVID-19. Roughly half of tenant households fell behind on rent last year and some landlords were looking for a legal predicate to remove them. The presence of extra residents not on a lease (which would be revealed by a police response to a domestic disturbance) would be one lease violation. Criminal activity (which would also be revealed by a call to police) would be another. 

Thus, there was a basis for a reduction in calls to 911 for service even before the mass protests began.

The data set I have raises additional questions that might be answered with access to additional data, a team of volunteer analysts chomping at the bit to get at this stuff and some Big Data techniques to visualize and parse it. 

Call volume fell ... but ... where? Was this a strike by Atlanta's black residents against police involvement, or "woke" white people who stopped calling the cops on Black jaywalkers? Did the fall in 911 volume occur across the city, or only in some neighborhoods? Did it persist through the winter to today? Is there a correlation between where the "missing" calls are and high numbers of delinquent tenants? Is there a correlation between those missing calls and the places where violent crimes occurred? How about other misconduct cases -- how do they plot against a map of loss of call volume? 

Yesterday, I finally managed to connect with the right person at APD. I’m getting the whole 911 data set from 2018 to today, and all of the initial incident reports for homicides in 2020. Bit by bit, we’ll get at this.