California’s Race-Blind Charging Law
Amidst discussions of big data, automation, and artificial intelligence, there has been a great deal of focus in ensuring that equity is included as a pillar of responsible development. In many aspects of life, outcomes aren’t equitable and we know that automating with bad data leads to bad results, just at a faster rate.
One of the areas of focus in ensuring just outcomes is in prison sentencing, where reports from the Sentencing Project found that minorities are more likely to be arrested, convicted, and receive long prison terms as compared to their white peers.
Because of this, some district attorneys have decided to begin race-blind charging processes for criminal defendants. In Yolo County, California, District Attorney Jeff Reisig implemented a race-blind prosecution process, partnering with Stanford University to deploy software that redacts racially identifying information before a prosecutor receives a case.
The project’s goal isn’t just to remove identifiers of race, but to compare how charging results would differ across a redacted and unredacted report. In addition to removing specific mentions of race, the software will also remove certain physical and location descriptors that might have an influence.
Race-blind charging isn’t something that those concerned with unequal sentencing are on board with wholesale though, some even argue that it only makes the problem worse. Florida International University researchers looked at a couple state attorneys offices and found that they embrace race-blind charging. These researchers argue that by avoiding race, prosecutors are allowing systemic injustices to continue without inquiry.
While these Florida prosecutors believe that being blind to race will produce the most equitable outcomes, it’s not clear that the commitment extends beyond an ethos, rather than a process that removes racial indicators from documents. Outside the subconscious influence seeing this information has, regardless of an intent to disregard it, the FIU researchers note that a race-blind attitude can result in alternative cultural identifiers or stereotypes which can be more insidious as a veiled substitute for race.
The topic of race-blind charging has gotten more focus in the news lately because of California’s Assembly Bill 2778, which will require distract attorneys across the state to redact racially identifying information regarding suspects, witnesses, and victims in charging documents. The bill, which was sponsored by Yolo County DA Reisig, passed 117-0, but was not without its opposition. The San Diego County Deputy District Attorneys Association contended that the exceptions to the law make it complicated and less effective, that it’s expensive, and that California shouldn’t have to be the testing grounds for it.
California, though, is no stranger to testing out new concepts, just as they did by requiring all of their counties to create a plan to redact discriminatory restrictive covenants in land records. Extract has been helping California counties tackle this challenge with automated software that understands context clues and references.
While Stanford has created software to spot racial identifiers, it isn’t clear that all the DA’s will be using it. Guidelines show how to find and redact across multiple pages using popular redaction tools and indicate that AI could help with accuracy, but DAs are on their own in selecting a tool.
Some offices, like in San Diego County, are soliciting bids for technology that can handle the process. While each DA’s needs may be unique, they’ll all need a solution that can work with their case management system and produce consistently accurate results.
For counties looking for an alternative to Stanford’s technology, Extract offers an automated redaction solution that can differentiate things like when the words ‘black’ or ‘white’ are used to describe race, a color, a last name, or something else.
If you’re interested in learning more or would like to see a demonstration of the redaction solution, please reach out.
