Performance anxiety

Anyone who's seen The Wire is intimately familiar with the process of playing with crime statistics to make them look better, rather than actually controlling crime. Radley Balko has a dreadful real world example:

Arthur Tesler was the only officer to take part in the Kathryn Johnston raid who didn’t take a plea bargain. Despite admitting that he lied, helped cover up Johnston’s murder, and stood watch outside while other officers handcuffed the bleeding 92-year-woman—allowing her to die while they planted marijuana in her basement—he was convicted today only on the charge of lying to investigators. He’ll face a maximum of five years in prison.

The one good thing to come out of the case is we got to see just how vast, deep, and pernicious the culture of corruption and disregard for civil rights ran in Atlanta’s police department. Tesler testified that narcotics officers were required to serve nine warrants and make two arrest per month, or they’d risk losing their jobs. This led to routine lying on warrants and bullying and intimidation of informants. What we don’t know is how many people were wrongly raided, arrested, and jailed because of all of this.

Compensation systems are really hard to design, as I wrote many years ago:

The agency problem is the fancy economic term for what most of us already knew intuitively; what benefits the stockholders doesn't necessarily benefit management. For example, I can think of many executives I've worked with on "re-engineering" projects, who, if they wanted to be honest about what would make their department work better, would "re-engineer" themselves right out the door and let somebody competent take over. Somehow, however, it was always one of their minions, usually one they didn't like too well, who was found to be superfluous. There are all sorts of ways in which this agency problem affects managers actions to the detriment of their shareholders, but one of the most widely known is in compensation.

If you know anyone in corporate sales, you probably already know of the hilarious shenanigans in which the sales force engages in order to meet their quotas. The purchasing manager at my old job was good for 10K or so of thoroughly bogus orders at the end of the month or the quarter to help our sales reps meet their quotas; in return they gave us a little extra off our regular purchases. These orders were invariably cancelled, after a decent interval, due to the whims of our fictitious clients. None of this was good for the companies for whom these sales reps worked; it benefited only the sales force.

But trying to prevent these gymnastics has proved futile. Change the quota from orders to sales and they'll ship the stuff out and have it "returned", incurring shipping charges both ways; change sales to "final sales" and they'll leave, because no one's willing to have their income that dependant on the whims of people they don't know. This applies even more to executives, who have much more power to manipulate matters so that they keep their job.

I was talking about CEO pay. But the problem matters more in the criminal justice system. We grade our prosecutors not on being right, but on winning. Cops are given stupid quotas rather than a mandate to, you know, actually reduce crime. Years ago, I read Joe McGuinness's Fatal Vision, on the once-infamous Jeffrey MacDonald case. At the time, it seemed like a fascinating insight into how you investigate a case. Then a few years ago, I came across Fatal Justice, a book on the case--and the role that McGuinness played in it--which offers pretty compelling evidence that MacDonald is possibly innocent and certainly the victim of forensic incompetence compounded by gross prosecutorial misconduct. If this happens in the spotlight, what's going on in the dark corners of the criminal justice system?

Well, things like this outrage, apparently. The Atlanta PD needs to remember that its job is to fight crime, not just put people in jail.