Reus ex machina: Defense counsel's first appearance arguments as a human intervention to mitigate algorithmic risk scores

Harris County implemented a pre-trial risk algorithm the same week as counsel at first appearance to address magistrates’ lack of individualized information when setting bail. The use of money bond, especially without individualized information, results in access to financial resources as the driver of pre-adjudicative incarceration, which has been linked to higher rates of guilty pleas and incarcerative sentences. Decision-making algorithms are marketed as providing defendant-based data in a race, gender, and income neutral way; however, the tools repackage incomplete and disparity laden information and create path dependency for defendants unable to question the categorizations. Rothgery v Gillespie County, Texas provides, even mandates, an operationalizable solution with a constitutionally recognized right to counsel at bail. Including defense counsel at a defendant’s first appearance in front of a judge has been repeatedly linked to lower bond amounts, higher rates of pre-adjudicative release, and increased use of no cost bonds. No study has explored how positive defendant outcomes occur. This work investigates how by conducting interviews with lawyers representing clients at their first settings, transcribing arguments presented to judges on felony arrests over a week period, categorizing the arguments, collecting bail amounts suggested by defense, requested by the prosecutor, set by the court, and included on a schedule, and reviewing the county’s training materials. These data help determine what defense argues at the first court setting following arrest and show what, if any, impact the argument has on the position of bail set relative to the prosecutor’s request and the risk assessment’s recommendation manifested as the bail schedule. The investigation found defense counsel: argues factors from United States v. Salerno and Texas’s Code of Criminal Procedure 17.15 while presenting individualized defendant information and reducing opportunities for self-incrimination; mitigates the use of the pretrial risk tool by correcting and contextualizing elements and scores to the point that they create what Barthes refers to as a reconstituted myth for the score; and may offer a counter-anchor to the bail amount recommended by the prosecutor and/or bail schedule. The results of this project identify how defense impacts bail amounts leading the positive results of counsel at first appearance. It offers an actionable strategy to counteract the negative aspects of algorithmic decision-making. For rhetoricians and technical communicators, these results steer us towards using Barthes as an operationalizable strategy instead of his common place as a method of critique.

Embargo status: Restricted until 06/2174. To request the author grant access, click on the PDF link to the left.