Print
Opinion

Is it our DA or the algorithm that makes S.F. less safe?

Vote no on Proposition 25
Why are people accused of gun crimes getting released back onto the streets of San Francisco? Photo: Tumisu

Sadly, San Francisco has become a laboratory for social experiments that are proving destructive and have incentivized the criminal trade.  At the heart of the matter is the Public Safety Assessment (PSA) algorithm and a district attorney with dangerous ideas about criminal justice. Proposition 25 will codify using computerized decisions to replace the experiences of human attorneys in the DA’s office. 

FAILURE TO PROSECUTE

For starters, below is a window into District Attorney Chesa Boudin’s implementation of criminal justice and his failure to prosecute. It would have been too voluminous to include all of Boudin’s first 10 months of dispositions, so below is a snapshot of just his first three weeks in office. And to document that Boudin has not changed his prosecutorial stripes, some recent egregious Boudin dispositions are also included. 

From the chart below, it’s important to observe the following:

  • 1) The seriousness of the crimes (i.e., guns, domestic violence, sexual assaults on children),
  • 2) How quickly each arrestee was released back out onto the street, and
  • 3) The quantity of cases that the district attorney has elected not to prosecute.

THE PROP 25 ALGORITHM

A computer algorithm generally assesses whether an arrestee will likely return to court for his assigned date, but the algorithm can’t evaluate whether an SFPD report lacks sufficient probable cause or the extent of a victim’s injuries in photographs. The domestic violence arrestees (above) who were released hours after they were arrested are directly attributable to Boudin’s leniency ⎯ and not the algorithm. 

An algorithm is primarily used for routine crimes, like burglaries, auto break-ins, and thefts. The legitimacy of using an algorithm as a substitute for a district attorney’s judgment is one of the two components of Proposition 25. The other component is whether arrestees should be allowed back on to the streets without bail as collateral.  

Former DA George Gascon started “experimenting” with the algorithm called Public Safety Assessment (PSA) in 2016. Boudin continues to use the PSA algorithm and claims he is not in a position to influence its disuse. However, on the San Francisco district attorney’s official website is a page dedicated to the Sentencing Commission, which “was formed in 2012, and is led by the District Attorney’s Office.” The first power and duty of the commission is to “recommend evidence-based criminal sentencing strategies.” “Evidence-based methods” is defined by epic.org as a euphemism for algorithms. Considering all of these aforementioned facts, it is completely inaccurate for Boudin to claim he is powerless over the PSA algorithm.

PROP 25 WELL-FINANCED BACKERS WITH CONFLICTS OF INTEREST

John Arnold, a Houston, Texas billionaire, has donated $3.5 million to the Yes on Proposition 25 campaign. He owns the PSA algorithm currently used by San Francisco and stands to gain the most if Proposition 25 passes, because his algorithm will be in the incumbent’s position for the no-bid SF contract. Arnold also donated $915,000 to the San Francisco government’s Financial Justice Project, which allows homeless people receiving citations to receive social services instead.

The tangled PSA algorithm financial web can be further dissected to the $2 million the San Francisco DA’s office received from the MacArthur Foundation. Sitting on the MacArthur board and approving the donation is Sendhill Mullainathan, made famous as a zealous algorithm supporter in Malcolm Gladwell’s 2019 book Talking to Strangers. Mullainathan is a founder of the ALJ Poverty Lab, which was funded by the MacArthurs and John Arnold. And John Arnold and the MacArthurs also both sit next to George Soros’ Open Society in the biggest donors circle for the Marshall Project.

All of this spells financial pressure to accept and promote the PSA algorithm in the DA’s office and in the upcoming election. 

THE PSA ALGORITHM’S FAILURES

How is a “routine case” defined so that it comes within the parameters of the PSA algorithm? Who knows? Should an attorney in the DA’s office or a computer program decide whether a habitual burglar or felon with a gun should be released back into the public? Boudin has been silent on the delineation.

In June, a burglar was released a couple days after he was arrested. In August, that same burglar broke into a home where a 16-year-old and a 10-year-old girl were home alone. The burglar attempted to rape the 16-year old. Boudin scapegoated the release of the burglar on the PSA algorithm. 

In June, “Vermond” was also released from San Francisco County Jail on robbery and gun charges. Vermond was required to wear an ankle tracker, but the career gang member cut the monitor off and subsequently went on a robbery spree. He was also wanted in both Fremont and Union City for armed home invasion robberies. His robbery series partner was also not held in San Francisco custody despite gun charges on September 22. In the commission of an October 1 armed robbery in the Union Square area, Vermond dropped his gun and was shot and killed by the intended victim.  

This incident follows the San Francisco algorithm release of a parolee with gun charges in 2017 only to be involved in the Twin Peaks robbery/murder of Edward French just 7 days later.

These three tragedies raise two questions:

Is the algorithm or Boudin responsible for the release of the non-violent (sic) criminals? (Though judging by the disposition of the cases in the table above, it appears it is Boudin’s goal to let arrestees back on the street as soon as possible whether an algorithm is employed or not), and

Might the 16-yearold girl not have been traumatized and would Vermond and Edward French still be alive today if there had been more severe consequences for those perpetrating previous crimes? 

VOTE NO ON 25

John Arnold has kept the secret sauce of his PSA algorithm walled-off from users by rationalizing that he doesn’t want competitors to steal his secrets.  

Conclusion #1: Neither Boudin nor anyone in his office knows how to monitor the algorithm’s filters or processes. 

Conclusion #2: A Houston-based political donor is attempting to earn monopolistic profits while gaining an outsized voice on the safety of our streets, using a monster computer algorithm that no one in San Francisco understands or can control 

Conclusion #3: You should vote to stop the algorithm takeover by voting no on Proposition 25!

Lou Barberini is a CPA. Send feedback to [email protected]

Send to a Friend Print