Po­lice end data-gath­er­ing ef­fort

Chicago cops rated res­i­dents’ risks of links to vi­o­lence

Chicago Tribune (Sunday) - - FRONT PAGE - BY JEREMY GORNER AND AN­NIE SWEENEY jgorner@chicagotri­bune.com asweeney@chicagotri­bune.com

Chicago po­lice qui­etly stopped rat­ing the risk of peo­ple be­ing caught up in vi­o­lence.

Chicago po­lice have qui­etly ended a con­tro­ver­sial data-gath­er­ing ef­fort that rated tens of thou­sands of res­i­dents on who was most likely to be caught up in vi­o­lence — ei­ther as a vic­tim or crim­i­nal.

A re­port last week by the city’s gov­ern­ment watch­dog dis­closed the demise of the eight-year pro­gram — re­ferred to by po­lice as the “Strate­gic Sub­ject List” or SSL — while rais­ing con­cerns about its op­er­a­tions on a num­ber of fronts, in­clud­ing the ac­cu­racy of the rat­ings and the shar­ing of the data with other law en­force­ment agen­cies.

The Po­lice Depart­ment had re­vised the SSL nu­mer­ous times since its in­cep­tion in 2012, in­clud­ing a sig­nif­i­cant over­haul last year that in­cluded a name change.

In qui­etly dump­ing the pro­gram in Novem­ber, the depart­ment ac­knowl­edged that the data-scrap­ing ef­fort hadn’t re­duced vi­o­lence, cit­ing a na­tional study re­leased last year that found the SSL in­ef­fec­tive.

The re­ver­sal fol­lows a sim­i­lar move by Los An­ge­les po­lice last year.

One na­tional ex­pert said CPD’s de­ci­sion was long over­due, con­sid­er­ing the in­her­ent bias in a depart­ment tar­get­ing peo­ple based on its own data as well as the risk of vi­o­lat­ing in­di­vid­u­als’ civil rights and con­cerns that such lists lead over-po­liced com­mu­ni­ties to dis­trust the po­lice.

“It is a big deal,” said An­drew Fer­gu­son, a vis­it­ing pro­fes­sor at Amer­i­can Univer­sity Washington Col­lege of Law who has writ­ten ex­ten­sively on big data polic­ing risks. “It fi­nally puts the brakes on a sys­tem of tar­geted per­son-based pre­dic­tive polic­ing that was largely flawed at the out­set. I think there is a clear fi­nan­cial cost — money not spent deal­ing with the un­der­ly­ing so­cial and eco­nomic is­sues that cause vi­o­lent crime. But there is also a per­sonal cost to the in­di­vid­u­als who were on these lists or tar­geted by these lists.”

‘We know who they are’

The SSL pro­gram was borne out of Chicago’s ex­per­i­men­ta­tion in the 2000s with pre­dic­tive polic­ing, a sci­en­tific ap­proach to fig­ur­ing out which “hot spots” would see spikes in vi­o­lence so more of­fi­cers could be de­ployed there to help out. Chicago ul­ti­mately be­came the only big-city po­lice depart­ment to win a fed­eral grant al­low­ing it to place in­creas­ing em­pha­sis on those be­hind the vi­o­lence.

Us­ing a com­put­er­ized al­go­rithm de­vel­oped by the Illi­nois In­sti­tute of Tech­nol­ogy, po­lice scored peo­ple on the list from zero to more than 500 by look­ing at myr­iad fac­tors, in­clud­ing their crim­i­nal his­to­ries, es­pe­cially for weapons or drug of­fenses, as well as if they had ever been shot or af­fil­i­ated with a street gang. The higher the score, the greater their risk to be in­volved in vi­o­lence.

Sim­i­lar to other depart­ment data­bases, the SSL was ac­ces­si­ble to street cops — and soon be­came in­te­gral to their daily op­er­a­tions. For a time, then-Su­per­in­ten­dent Ed­die John­son noted “we know who they are” in dis­cussing the city’s crim­i­nal el­e­ment — a ref­er­ence to the SSL.

Po­lice brass of­ten cited the data in re­sponse to high-pro­file crime in­ci­dents, telling re­porters if sus­pects scored high on the SSL. But it was never clear that it pro­vided a prac­ti­cal way to re­duce vi­o­lence. A 2019 study by the Rand Cor­po­ra­tion con­cluded as much.

In the re­port, the public pol­icy think tank con­cluded that the Po­lice Depart­ment’s last ver­sion of the SSL was “op­er­a­tionally un­suit­able” for sev­eral rea­sons, in­clud­ing that the data used to de­ter­mine in­di­vid­u­als’ risk as­sess­ments was too out of date to keep up with the quick pace of Chicago’s vi­o­lence.

The re­port also found a “level of public fear” about the pro­gram that over­shad­owed its ef­forts to re­duce vi­o­lence. Those public con­cerns stemmed from the Po­lice Depart­ment’s fail­ure to be fully open about the pro­gram’s in­ner work­ings, the re­port said.

The study noted as well that be­ing on a “bad guy list” tar­nished shoot­ing vic­tims.

In ad­di­tion, since ar­rest data was a key fac­tor in de­ter­min­ing those most atrisk of vi­o­lence, any­one with a re­cent ar­rest was au­to­mat­i­cally sub­jected to “ad­di­tional scru­tiny and pun­ish­ment,” Rand said.

The Po­lice Depart­ment touted its use of the SSL by po­lice of­fi­cials, clergy and com­mu­nity lead­ers dur­ing door-to-door vis­its with some of those with the high­est scores, warn­ing them about the dan­gers of a crim­i­nal life­style while of­fer­ing so­cial ser­vices to try to help steer them away from trou­ble.

In its re­port re­leased last week, the city in­spec­tor gen­eral’s of­fice found fault with the Po­lice Depart­ment for, among other things, not prop­erly train­ing its of­fi­cers on their use of the SSL.

The IG re­port also crit­i­cized the Po­lice Depart­ment for shar­ing the data with the Cook County sher­iff’s and state’s at­tor­ney’s of­fices as well as the mayor’s of­fice with­out pro­vid­ing guide­lines on how to use in­for­ma­tion.

The re­port also faulted po­lice of­fi­cials for im­prop­erly fac­tor­ing in­di­vid­u­als’ ar­rest his­to­ries into their scores with­out con­sid­er­ing whether they had been con­victed.

“If an in­di­vid­ual was re­leased with­out charges or was charged and ac­quit­ted, or if the charges were dis­missed, that in­di­vid­ual would have had a risk score … which fac­tored in an of­fense which they were never found to have com­mit­ted,” the re­port said.

Fer­gu­son, the law pro­fes­sor, noted that what he calls “tar­geted peo­ple-based” pre­dic­tive polic­ing also raises prob­lems be­cause it is based on ar­rest records gen­er­ated by the depart­ment, re­sult­ing in a built-in bias. For ex­am­ple, drug ar­rests can be a prod­uct of over­polic­ing, pos­si­bly un­der­cut­ting the va­lid­ity of such data to de­ter­mine an in­di­vid­ual’s true risk.

Fer­gu­son said us­ing such data in de­vel­op­ing polic­ing strate­gies can be ef­fec­tive, but he crit­i­cized CPD — fac­ing pres­sure to re­duce vi­o­lence — for push­ing the idea with­out an ef­fec­tive plan for how it would be used.

“Chicago felt like they had to do some­thing, and this was their an­swer,” he said. “All of these things are a fail­ure of imag­i­na­tion, of vi­sion at the front end.”

Last year a sim­i­lar list kept by Los An­ge­les po­lice of the city’s “chronic of­fend­ers” drew crit­i­cism in a po­lice over­sight re­port for be­ing poorly main­tained and in­con­sis­tent. This led the depart­ment to an­nounce ma­jor changes, Fer­gu­son said.

A shift in strat­egy

In late Septem­ber, the nearly $4 mil­lion in fed­eral grant money that CPD used to de­velop the SSL ran out, lead­ing to its can­cel­la­tion on Nov. 1, ac­cord­ing to the in­spec­tor gen­eral’s re­port.

In an interview Fri­day, An­thony Guglielmi, the Po­lice Depart­ment’s chief spokesman, said the de­ci­sion to end the ef­fort also re­flected a shift in strat­egy to rely on tech­nol­ogy cen­ters in­side po­lice dis­tricts and de­tec­tive ar­eas as well as sup­port for a grow­ing net­work of com­mu­nity or­ga­ni­za­tions that pro­vide street out­reach, so­cial ser­vices and vic­tim ad­vo­cacy across the city.

The tech cen­ters, funded largely by a $10 mil­lion pri­vate do­na­tion, of­fer a wide ar­ray of com­puter pro­grams and spe­cial­ized tools — in­clud­ing gun­shot de­tec­tion tech­nol­ogy called ShotSpot­ter and an ex­ten­sive net­work of sur­veil­lance cam­eras — for of­fi­cers to track vi­o­lence in real time.

The cen­ters have been ex­panded to all but two of the city’s 22 dis­tricts, pro­vid­ing crit­i­cal in­for­ma­tion in solv­ing shoot­ings, the depart­ment said.

But Fer­gu­son warned that, just like the SSL, the tech cen­ters’ ex­panded use of sur­veil­lance needs to be mon­i­tored to pro­tect the con­sti­tu­tional rights of ci­ti­zens.

“We can’t take our eyes off the ball,” he said.

In re­sponse, Guglielmi said the tech cen­ters gather data mostly on the places and times that crimes hap­pen to help com­man­ders de­ploy of­fi­cers to re­spond. The cen­ters are used to quickly gather any rel­e­vant video footage. The city, he said, has a long his­tory of de­pend­ing on sur­veil­lance cam­eras, and he stressed they cap­ture crimes tak­ing place in public spa­ces only, not pri­vate homes.

The Amer­i­can Civil Lib­er­ties Union of Illi­nois, which has been crit­i­cal of the Po­lice Depart­ment’s em­pha­sis on pre­dic­tive polic­ing, praised its de­ci­sion to end its use of the SSL as well as the in­spec­tor gen­eral’s re­port for shin­ing a light on its prob­lems.

In a state­ment, Karen She­ley, direc­tor of the ACLU of Illi­nois’ po­lice prac­tices project, called for con­tin­ued trans­parency on the Po­lice Depart­ment’s part for the “pre­dic­tive polic­ing tools” it is still us­ing.

“Trans­parency is crit­i­cal to the on­go­ing ef­fort to cre­ate faith in polic­ing in Chicago,” she said. “Un­reg­u­lated, so­phis­ti­cated, pow­er­ful data­bases in­volv­ing hun­dreds of thou­sands of names only harm that trust.”


Chicago po­lice of­fi­cers process a crime scene on Jan. 16. Po­lice qui­etly ended a con­tro­ver­sial data-gath­er­ing ef­fort to pre­dict crime in Novem­ber.

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.