(Chesa Boudin campaign account/Facebook)

Boudin eliminated cash bail and restricted pretrial detention in San Francisco. He also reaffirmed a flawed quest to predict who should be jailed for what they might do.

Last week, San Francisco’s newly elected district attorney, Chesa Boudin, released his office’s new bail policy. Following his campaign promise, the new policy forbids prosecutors from requesting money bail under any circumstances. In addition, it allows them to request pretrial jail time only for people who face certain violent charges and who prosecutors believe pose a high risk of violence or flight. 

With these provisions, Boudin has adopted what is easily the most progressive prosecutor bail policy in the country. His reforms are more comprehensive and transparent than those adopted by other leading progressive prosecutors, such as Boston’s Rachael Rollins and Philadelphia’s Larry Krasner, let alone those of DAs elsewhere in the country.

But the policy also reveals how tough-on-crime norms limit the contemporary vision of progressive prosecution. Boudin’s office will continue the practice of assessing risk to justify incarcerating legally innocent people for their future crimes. It has defended this approach with unproven appeals to public safety and predictive accuracy, that in a prior era of bail reform would have been seen as an intrusion on fundamental civil rights.

For decades, activists and some legal scholars have denounced the practice of incarcerating people pretrial as a violation of the presumption of innocence, unlikely to improve public safety, destructive to communities, and racially discriminatory. Awaiting trial from a jail cell, people suffer worse case outcomes and risk losing their jobs, their homes, and custody of their children.

Money bail has been the primary mechanism for pretrial incarceration: A judge conditions the release of a person upon the person posting a money bond, often on the recommendation of the prosecutor. Amounts are often set beyond what the person can afford. On any given day in the United States, hundreds of thousands of people are detained pretrial because they are unable to post bail.

In recent years, some progressives have won DA elections on promises to transform this bail system. Boudin, Krasner, and Rollins have released three of the most emblematic prosecutor policies on the issue so far. Each expands the circumstances under which people not yet convicted of a crime will be released from jail without financial conditions. 

Only Boudin’s reform outright eliminates money bail, though. 

According to the policy he set, prosecutors in his office can never request money bail, and they can never defend someone’s incarceration on money bail. That’s a big shift in policy that should enable many people to avoid jail pretrial. 

On the other side of the country, prosecutors in Rollins’ office can still request money bail in situations where they believe there is a flight risk. In Philadelphia, Krasner’s prosecutors face looser restrictions. Krasner instructed his prosecutors to not request cash bail for certain low-level offenses (misdemeanors and some felonies classified as nonviolent). Otherwise, prosecutors there are free to request money bail. 

So even in jurisdictions with progressive prosecutors, incarceration on unaffordable money bonds continues to be a problem. Through extensive court-watching efforts, the Philadelphia Bail Fund has found that, although the pretrial jail population has declined, Krasner’s prosecutors continue to request unaffordable bail amounts that result in pretrial incarceration, as The Appeal reported in the summer of 2019

By contrast, prosecutors in San Francisco will no longer be able to ask judges to detain people by imposing financial conditions on their release, at least if Boudin’s directive is properly implemented. In San Francisco, people’s freedom will not depend on their ability to post bail.

The second aspect of Boudin’s policy is that it limits when someone can be detained pretrial. If a jurisdiction only eliminates money bail, there’s always the risk that, instead of facilitating people’s release, courts and prosecutors will turn to other methods of pretrial incarceration. This is the trade-off that critics say was made with California’s 2018 bail reform law (Senate Bill 10), which is currently pending a 2020 public referendum.

That is because money bail is not the only mechanism for jailing people pretrial. In many states, including California, a judge — often acting on a prosecutor’s recommendation — can determine that a person is dangerous or is a flight risk and order that person to be jailed until the criminal case is over.

Boudin’s policy restricts when this can happen. It sets a presumption against it, and enables prosecutors to recommend pretrial detention only for people facing charges for certain violent felonies. And they can do that only if the prosecutor is convinced that the person has a “substantial likelihood” of committing “great bodily harm” or fleeing the jurisdiction if released. These are narrower circumstances than SB 10 would allow. They are also somewhat narrower than Rollins’s policy, which enables prosecutors to request pretrial incarceration for all statutorily eligible offenses in Massachusetts — mostly violent felonies and gun charges — if there is a “clear safety risk to an identifiable victim or witness.” 

But in defining the circumstances where pretrial detention ought not to occur, Boudin, like Rollins and supporters of SB 10, is also endorsing the premise that it sometimes should. 

These policies affirm that pretrial incarceration is justified based on predictions of future dangerousness. This conception of pretrial justice reflects the limits not just of particular  policies but of the current horizon of progressive prosecution. Rollins’s office has made use of such “dangerousness” holds, The Appeal reported in October, though she has requested these hearings with significantly less frequency than other Massachusetts DAs who use reform rhetoric. 

Pretrial incarceration based on dangerousness assessments, a policy first proposed as legislation by the Nixon administration, swept the country in the 1970s and 80s. It permitted courts, for the first time in American history, to legally jail people awaiting trial based on a public safety rationale. These new laws were a turn away from the conclusions of the Johnson administration’s seminal 1967 report “The Challenge of Crime in a Free Society,” which had considered pretrial detention as way to reform bail but had determined that it “might well create more of a problem than the imposition of money bail, in the light of the difficulty of predicting dangerousness.” 

In a 1987 opinion written by Chief Justice William Rehnquist, the Supreme Court found the federal government’s new preventive detention scheme to be constitutionally permissible. In dissent, Justice Thurgood Marshall chastised the court for “disregard[ing] basic principles of justice.” He warned of “the coercive power of authority to imprison upon prediction” and “the dangers which the almost inevitable abuses pose to the cherished liberties of a free society.”

Boudin’s bail policy sides with Nixon and Rehnquist over Johnson and Marshall. The policy assumes that prediction can save us from the harms of pretrial incarceration. 

Like recent reforms around the country, the plan is to develop better ways to identify the “right” people to detain pretrial. The press release accompanying Boudin’s policy promises that the office will release “people who are safe” and detain “those who pose a serious threat to public safety.” 

But whether that determination is made through prosecutors’ judgment or actuarial risk assessment tools, the project cannot succeed. Neither prosecutors nor algorithms can know in advance who will commit violent crime.

It turns out that predicting violence is really hard. The information prosecutors see and the judgments they make are racially skewed. And although risk assessment tools are often touted as a solution to implicit bias, a perspective that the press release endorses by evoking the “equitable decisions” born of “objective data,” the data used to build actuarial assessments is itself deeply flawed and biased. Even the best actuarial risk assessments’ predictions of pretrial violence are frighteningly poor. Data from jurisdictions that use the Public Safety Assessment (PSA), a risk assessment tool that Boudin’s policy includes as a factor in bail decisions, show that 86 to 99 percent of the people that the algorithm flags for potential pretrial violence will not get arrested for a violent crime if released. Arnold Ventures, the company behind the PSA tool, has cautioned that the tool should not be used as the basis for detaining someone. Largely because pretrial violence is so rare, it is also hard to predict. 

Even high-risk people are unlikely to commit a violent crime while awaiting trial, so a preventive detention system involves incarcerating thousands of legally innocent people for what a fraction of people might do. The fallout of these policies disproportionately impacts poor communities and communities of color — especially Black communities. 

Boudin himself has researched the shortcomings of risk assessment tools. In a paper released last week, he and his co-authors show that, by overcharging cases, prosecutors can inflate a defendant’s risk score, leading to a recommendation for a harsher pretrial outcome, even if the charges are eventually dropped. Such outputs are a far cry from the “objective data” promised by his office press release. 

Boudin’s policy prevents prosecutors from recommending pretrial detention without supervisor approval, in an effort to foster the presumption against it. But this only moves the problems with making reliable predictions up the hierarchy. Acknowledging the gravity of pretrial incarceration does not by itself make the decisions sounder. 

By eliminating financial conditions for release and restricting when prosecutors can request to jail people pretrial, Boudin took more progressive action on bail policy than his contemporaries. 

But Boudin’s policy also upholds a pretrial incarceration scheme justified by tough-on-crime conventions. Today’s bail policies expect prosecutors and algorithms to know in advance who will commit crimes and who won’t. But no one can know that. At its core, a preventive detention scheme endorses the view that with good enough math or intuition we can overcome the moral stain of incarcerating people — not for what they have done but for what they might do.

Colin Doyle is a staff attorney at the Criminal Justice Policy Program at Harvard Law School. He works on bail and pretrial reform across the country at the local and state levels.