California Gov. Jerry Brown on Aug. 28 signed state Senate Bill 10 into law, which will entirely replace California’s cash bail system with a risk-assessment bail system.

Under the new law, which will take effect Oct. 1, 2019, California judges will no longer impose monetary bail conditions on criminal defendants who are in jail while they await trial.

While that’s California’s prerogative, it must be recognized that this is a policy determination that is based on good intentions, but without much data to support it.

Traditionally, in a cash bail system, a defendant may be released from jail while awaiting trial in exchange for a payment to the court. Judges set and determine the amount of the payment at the defendant’s arraignment or bail hearing.

With California’s new risk-assessment bail system, defendants facing felony charges will no longer be required to pay monetary bail to be released from jail prior to trial.

Instead, judges will make bail determinations according to a defendant’s “risk rating,” which represents the risk that a defendant, if released on bail, will not show up for court or be rearrested for committing another crime.

Risk-assessment bail systems use computer algorithms, data analytics, and machine learning to calculate risk ratings for defendants. Typically, risk-assessment tools consider a defendant’s sex, age, personality, criminal history, and socioeconomic background to calculate the defendant’s rating.

In California, judges will be required to use a defendant’s risk rating to determine whether the defendant should be released on bail or remain incarcerated prior to trial.

Judges will only be allowed to depart from the rating under a specific set of circumstances.

The system will rate defendants on a low-, medium-, or high-risk scale. The higher the risk rating, the more probable it is that defendants would not appear in court or would be rearrested for committing another crime.

Individuals rated as high risk most likely will be denied bail, while low-risk individuals will most likely be granted bail, so long as they agree to abide by non-monetary bail conditions, such as house arrest.

Medium-risk defendants will be denied or granted bail following an additional pretrial hearing in front of a judge.

According to Brown’s office, the system is being implemented to “[preserve] the rights of the accused, while prioritizing public safety.”

The California law comes at a time when many Americans—on both sides of the political aisle—are urging lawmakers, at all levels of government, to make significant reforms to the criminal justice system, including reforms to the cash bail system.

When it comes to cash bail reforms, some—including Brown, a Democrat, and the California state Legislature—think the current system is biased, unfairly keeps individuals locked up in jail simply because they cannot afford to pay bail, and then produces wealth-based, as well as racial, disparities in the criminal justice system.

Whether that is actually true is a question for another day, but there’s good reason for state legislators to scrutinize their cash bail policies.

Pretrial detention of individuals who do not pose a risk of committing crime upon release or fleeing before trial can waste jail space that could otherwise hold a more harmful offender, and can lead to needless separation of families and to loss of jobs and housing, among other social costs.

Over the past several years, several states have rushed to reform their cash bail systems by adopting risk-assessment tools to aid judges in the bail-setting process.

These states think that risk assessments will more fairly, effectively, and objectively determine whether criminal defendants belong behind bars while they await trial, and as a result, eliminate problems such as wealth-based or racial disparities, allegedly caused by cash bail systems.

Yet risk assessments in the criminal justice system are still in their relative infancy, and it’s not quite clear that risk-assessment tools that currently are being used are eliminating much of anything.

ProPublica has published one study about a risk-assessment tool known as COMPAS, which concluded that the tool produces biased risk ratings for African-Americans. The developers of COMPAS responded with a study of their own and concluded that no such bias exists and that the tool is actually very accurate.

Who’s right? That’s anyone’s guess.

According to Megan Stevenson, an economist, criminal law scholar, and law professor at George Mason University’s Antonin Scalia Law School, “virtually nothing is known about how the implementation of risk assessment affects key outcomes: incarceration rates, crime, misconduct, or racial disparities.”

Stevenson further asserts that “[t]he empirical research evaluating whether outcomes are improved by incorporating algorithmic risk assessment into the decision-making framework is beyond thin. It is close to nonexistent.”

Stevenson is one of the first to conduct an empirical study on the effects of risk-assessment tools, which will be published in a forthcoming law review article. (The draft is available here.)

In Kentucky, Stevenson writes, the risk-assessment tools “led to neither the dramatic efficiency gains predicted by risk assessment’s champions, nor the increase in racial disparities predicted by its critics.”

Therein lies the problem with risk-assessment tools at the moment.

These tools are highly complex, and a sophisticated understanding of statistical analysis and computer science is necessary to understand how these tools operate.

Before we can draw any conclusions about their effectiveness, or lack thereof, we—including government—must first understand whether and how these tools work, and what, if anything, they accomplish.

Even then, some things might make perfect sense in theory, but not in practice.

Yet, despite the current lack of reliable evidence, more and more states are incorporating risk-assessment tools into their bail systems—as well as their sentencing and parole systems—without actually knowing whether these tools are capable of achieving a desired result.

It’s possible that these tools will make matters worse. It’s also possible that they will make matters better. Until we know more about what outcomes they produce, rather than the good intentions they represent, the jury is out.

That’s why it’s essential that states that utilize risk-assessment tools evaluate their effectiveness thoroughly, constantly, and with an open mind, fully prepared to adapt or drop the tool, based on the results of such evaluations.

The stakes are way too high on all sides not to do so.

California has become the first state to opt to completely eliminate its cash bail system and is set to replace it with a promising—yet unproven—risk-assessment bail system.

After Brown signed Senate Bill 10 into law, he said, “Today, California reforms its bail system so that rich and poor alike are treated fairly.”

Let’s hope he’s right, but in the meantime, one must wonder, how is he so certain?