McKinsey and NYC Corrections Officials Rig 'Restart' Program Data at Rikers Island

McKinsey tests its new anti-violence strategy in what the firm calls ‘Restart’ housing units at Rikers Island, implementing its centerpiece algorithm called the Housing Unit Balancer (HUB) designed to predict each inmate’s propensity for violence. By July 2015, eight Restart units housing about 250 inmates are in place at the George Motchan Detention Center.

However, jail officials and McKinsey consultants jointly rig the Restart program in its earliest phase to virtually guarantee there will be few violent episodes. For each Restart unit, jail officials replace inmates selected by McKinsey’s HUB algorithm but considered troublesome with alternates believed to be more accommodating, chosen from a list provided by McKinsey. One email forwarded to several consultants shows a jail guard carefully curating gang members on McKinsey’s initial list, noting one inmate was a Blood ‘but acceptable for program.’

McKinsey reports that violence has dropped more than 50% in the Restart facilities, but records later reveal this number is bogus. The data manipulation allows McKinsey to claim success and justify contract extensions, even though the methodology is fundamentally fraudulent. In October 2016, the Department of Correction announces that violence in the Restart units is ‘down by over 70% and assaults on staff are down by 82%’ - statistics based on cherry-picked, docile prisoners rather than the algorithm’s actual selections.

This manipulation exemplifies how consulting firms can manufacture success metrics to justify their fees while providing no real value. The rigged data helps McKinsey secure contract extensions that will ultimately bring total payments to $27.5 million for a program that not only fails but coincides with worsening violence.