By Strong Ma, J.D. candidate, Harvard Law School
STUDENT VOICES: The views expressed below are those of the student author and do not necessarily reflect the position of the Access to Justice Lab.
“Replacing cash bail: Fairer justice or robopocalypse?” This headline, emblazoned across a California news article in 2020, encapsulates the rhetoric at the time surrounding Proposition 25, a statewide referendum on California’s efforts to do two different things simultaneously: 1) eliminate cash bail, and 2) provide judges making pretrial release decisions with a risk assessment instrument. The headline represents more than a time capsule of an intriguing but internal state political scuffle. The association of ending cash bail with algorithmic decision-making embodies a complexity to bail reform efforts that members of the bail reform movement, inside and outside California, should remember in their push for greater access to justice for pretrial detainees. Ending cash bail does not necessarily espouse a preference for algorithmic (or “robotic”) decision-making, nor a distaste for judicial discretion. Moreover, a complex criminal justice reform movement need not (and, indeed, likely cannot) address all bail issues at once.
Cash Bail: A Background
After an arrest, courts often set an amount of money defendants must deposit as condition of release before trial. If a defendant cannot pay the amount, they have two options: stay in jail, or pay a nonrefundable premium to a bail bonds company that will post the full amount. In fact, bail bonds are the most common method for defendants to obtain release, with only four states banning them. Individuals unable to afford either the bail amount or the bail bonds fee must choose the former option.
Critics of cash bail argue that it creates a two-tiered system of justice where pre-trial freedom depends on one’s wealth, as bail schedules and courts can prescribe amounts unrelated to a particular defendant’s ability to pay. Critics further say that cash bail exacerbates racial disparities and increases pretrial incarceration—incarceration which in turn leads to unjust and harmful outcomes. Defenders have argued that cash bail can improve accountability and public safety. Yet, a 2017 New York Times article observed a then-growing consensus against cash bail that “cuts across party lines, and includes law enforcement leaders, prosecutors, defense lawyers, the courts and religious leaders,” with the only holdouts being (predictably) the bail bond industry.
California’s Effort to End Cash Bail
In part riding this wave of the then-growing consensus, California enacted SB10 in 2018, which eliminated cash bail and replaced it with a system of pretrial risk assessment. The law would have required courts to use a “validated risk assessment tool” for all pretrial detainment decisions. Such tools assign points based on certain defendant characteristics (such as age and criminal record) that are ultimately tallied to a “score” often categorized as low, medium, and high risk. Under the SB10 schema, before arraignment, low risk defendants would be automatically released, and high risk automatically detained. Judges maintained the discretion to make release or detention decisions with the score serving as guidance.
SB10 would have gone into effect in 2019, but opponents of the law obtained enough signatures to put it up for a referendum vote through California’s direct democracy system. Under the title Proposition 25, California voters could either vote yes and support ending cash bail through SB10’s provisions, or no and strike SB10 from California law. Despite California voters in the same election approving parolee voting rights and rejecting harsher criminal penalties, Prop 25 ultimately lost at the ballot box, with 56.4% voting no and 43.6% voting in favor.
Post-Mortem of the Rhetoric Surrounding Prop 25
Prop 25 presents an interesting conundrum: In light of the seemingly broad consensus against cash bail and reform friendly results on other 2020 criminal justice propositions, why did California’s efforts to end cash bail fail?
One possible explanation stems from the successful harnessing of the boogeyman of “algorithms.” In other words, narratives surrounding Prop 25 eventually focused not on the injustice of cash bail but on the potential pitfalls of the risk-based assessment system that SB10 meant as a replacement.
One can see this in the “No on 25” campaign’s messaging. Opponents of Prop 25 repeatedly invoked the threat of algorithms making judicial decisions: “Imagine a spouse, son, daughter or close friend stuck in jail at the mercy of computers and the bureaucracy”; “[Prop 25] will force Californians to rely on computer algorithms managed by government bureaucrats to decide who can and can’t get out of jail.”
In addition to trying to instill a general concern for arbitrary or untransparent decision-making by algorithms, opponents also argued that such “computer profiling” would discriminate against minorities. To support this point, the “No on 25″ campaign heavily publicized that some civil rights organizations had joined their coalition, including the Southern California ACLU and Human Rights Watch, making them the odd bedfellows of the bail bond industry. Indeed, the President of the California NAACP wrote a strong statement in opposition to Prop 25 in California’s official voter guide (though there may have been unsavory financial ties motivating that particular instance of support).
Prop 25’s Rhetoric vs. Reality
The arguments against Prop 25 could be challenged on multiple fronts. One is that SB10 neither mandated a particular algorithm nor completely eliminated judicial discretion in release decisions. Moreover, while there are legitimate critiques regarding bias, risk assessment tools in practice likely do not match the popular conceptions of “computer profiling.” In many cases, risk assessments are merely a checklist of factors that any person could use—such “algorithms” need not be designed nor applied by machines.
Yet, the narrative-setting by opponents of Prop 25 seemed to have worked. For example, CalMatters, an award-winning political news organization known for its balanced reporting, ran the “Robopocalypse” headline as part of its coverage. Its voter guide summarized the proposition as replacing cash bail with an algorithm. Other news coverage was similarly dominated with an algorithm framing. More concerning is that this domination of the debate seemed to spread to bail reform discourse outside of California. Many publications have framed the broader ending cash bail effort as naturally coming with both the use of risk assessment algorithms and the reduction of judicial discretion.
Takeaways
We can learn two things from the California example: 1) the perceived threat of algorithms can poison discourse around ending cash bail, even in arguably favorable political environments; and 2) cash bail lies at the nexus of various critiques levied at the justice system—that it privileges the wealthy, remains biased against minorities, and overall advances an over-carceral institution—not all of which can be simultaneously or perhaps even immediately addressed.
Jurisdictions can begin tackling the wealth disparity critique by eliminating cash bail without changing all that much else—perhaps by leaving it up to judges to decide detainment and release decisions without resorting to money bonds. After all, cash-based bail did not exist at the beginning of United States criminal procedure history. On the other hand, addressing bias and over-incarceration likely will need to involve some further combination of reexamining judicial discretion, liberalizing statutory penalties, and other systemic changes. Cash bail reform often comes packaged with such ideas (with all of their perceived costs and benefits) but it need not. Neither does ending cash bail need to be synonymous with using algorithmic assessment tools, whose merits can and should be debated.
Despite these complexities, the hope for ending cash bail has seen a resurgence. In 2023, Illinois became the first state to abolish cash bail after their highest court upheld the 2021 Safe-T Act. Unlike California, Illinois did not focus on nor mandate a statewide risk assessment tool. Instead, it called for individualized assessments. While the law does permit the use of pretrial risk algorithms in such assessments, they cannot be the sole factor in a decision. Thus, the Safe-T Act keeps risk assessment tools in the background. While California’s specter of a “robopocalypse” may still cast a shadow on bail reform efforts, the Illinois example may illuminate a path forward on one of the most visible issues in access to justice.
If you’re interested in more on this topic, listen to our Proof Over Precedent podcast episode.