One of the most important issues that California voters face for the November 2020 ballot is California Proposition 25, Replace Cash Bail with Risk Assessments Referendum (2020). Previously known as Senate Bill 10 (SB10), Prop 25 is a new measure that would effectively eliminate cash bail in California and replace it with a system that evaluates a defendant’s likelihood to attend all court appearances based on an algorithm instead of the insight of a judge. Proponents of Prop 25 argue that these new measures would create a fairer, more just system, but the issues at hand are far more complicated than that. California voters should understand that Prop 25 is not a full-proof system and has several flaws that outweigh its benefits.
The Current Cash Bail System
Under the current cash bail system, if you are arrested for a crime, you will likely have to pay a bail bond in order to be released from jail. A bail bond is essentially a financial agreement between you and the court that states that you will attend all court appearances and abide by certain restrictions until your trial is complete. If you miss a court appearance or attempt to flee the state or country, the court could refuse to return the amount you paid in bail to you and put a warrant out for your arrest. However, if you do attend all court appearances, you will have your money returned to you, whether or not you are found guilty.
The amount of bail you will have to pay depends on several factors, including a bail schedule that is determined by California judges each year, where you live, where you work, and the likelihood that you will not show up to a court appearance. The annual California bail schedule sets specific bail bond prices based on certain crimes, which can have minimums and maximums. While one charge may be as low as $1,000, another could be in the millions. To help defendants pay these costs, agencies like Balboa Bail Bonds can pay the full amount of the bond. In turn, our clients will only have to pay us a small, 10% fee, effectively making the system more affordable for Californians.
How Does Prop 25 Change Bail?
Californians have two options when voting on Prop 25:
- Voting Yes on Prop 25 would replace California’s cash bail system with a pretrial risk assessment system.
- Voting No on Prop 25 would keep California’s cash bail system in place with no changes.
Under a pretrial risk assessment system, each defendant would be reviewed by a computer algorithm instead of having defendants pay a cash bond to be released from jail during their trial. This algorithm will consider nine factors, which include their age, current charges, previous convictions, and whether or not they have failed to appear in court before. By analyzing these factors and adding them into a points system, the algorithm will assign a risk level to a defendant, which includes:
- Low-risk: If the system determines that you are a low-risk of missing a court appearance, then you will be released from jail without any restrictions.
- Medium-risk: For medium-risk defendants, the court will schedule a Prearraignment Review to determine if they can be released or if they must remain in jail until the trial.
- High-risk: In general, all high-risk defendants will be held in jail until their trial.
In the current system, judges use their own mathematical processes to determine bail amounts and risk factors, while the new system would eliminate the human factor and place the decision in the hands of a computer algorithm. Proponents have argued that this will eliminate issues like racial or class bias and create a fairer system, but experts and civil rights leaders disagree.
How Algorithms Can Be Racially Biased
When people think of machine-learning and computer algorithms, they like to believe that these systems are completely free of bias. They hope artificial intelligence will rely on hard facts and data to make decisions. However, what if that data is flawed? What if the system the program is working within is biased? Can a computer algorithm truly be 100% objective?
According to experts in machine-learning, the answer to that final question is, in fact, no, it cannot.
In a joint statement from researchers at MIT, Harvard, Princeton, NYU, UC Berkeley, and Columbia University, experts in artificial intelligence and machine-learning have criticized pretrial risk assessment programs for their reliance on flawed data sources, unreliable predictions, broad definitions, and overestimations of danger. For example, these algorithms are designed, in theory, to predict the likelihood of a released defendant committing a violent crime before their trial. However, there is very little data for these programs to rely on, which can lead them to over-predicting violence. In some instances, algorithms have labeled defendants high risk even when the crime they were arrested for was not violent.
In addition, these programs often cite arrest records when making judgment calls. But, as the researchers have noted, African-American and Latino communities are often overpoliced and have higher rates of arrest, which can drive the programs to enforce harsher restrictions on them, even labeling them high risk for minor crimes. While the program is not necessarily asking for the defendant’s race when making a judgment, it is influenced by distorted data that is laced with racial bias.
One need only look at a case study in Florida to see the negative impact of pretrial risk assessment programs. According to an in-depth investigation by ProPublica, a computer program utilized in Broward County, Florida was “almost twice as likely” to label black defendants as high risk when compared to white defendants who had committed the exact same crime, even when the white defendants had longer criminal records. In addition, when ProPublica dug deeper into the data, it discovered that the program determined that black defendants were “77 percent more likely to be pegged as at higher risk of committing a future violent crime and 45 percent more likely to be predicted to commit a future crime of any kind.”
On a small scale, this program demonstrated clear racial bias when making important and life-changing decisions for defendants while only overseeing a single county, but if implemented across all of California, the results could be disastrous. While proponents may hope that these programs will eliminate racial bias, the fact of the matter is that these programs only further enforce it.
How Are Pretrial Risk Assessments Funded?
An important element that is often forgotten when discussing bail reform is how the pretrial system is currently funded. With cash bail, bail bond agencies handle many of the administrative tasks of monitoring defendants when they are released before their trials, including paying for ankle bracelets, supervision, and ensuring they arrive at the court on time. Under the new system, the costs would ultimately shift to taxpayers.
But it is not just the administrative costs that would be an issue for taxpayers. Each county and city will need to have these systems put in place, court officials trained on how to properly interrupt and use them, and they will need to be monitored, which will expend both administrative and IT costs. There will also need to be additional court hearings to overturn an algorithm’s judgment. All of this will put the financial burden on taxpayers who are already struggling with an unreliable economy.
Given the massive cutbacks that California has suffered due to the COVID-19 pandemic, adding on more costs to implement an unreliable and untested system state-wide is extremely risky. If you look at another heavily populated state like New York, you can see that pretrial risk assessment programs are also ineffective. In 2019, the state implemented a progressive transformation of its pretrial system to eliminate bail; however, just as early as April 2020, New York rolled back the bail reform, citing the high costs of managing the system in the wake of the COVID-19 outbreak.
Even in our own San Diego County, pretrial risk assessment programs were utilized in conjunction with bail up until 2015, when the programs were ultimately shut down due to the budget cuts. This shows that even a hybrid may be too expensive for California taxpayers.
Pretrial Risk Assessments Criticized by Major Civil Rights Leaders
In conjunction with both the inherent flaws and costs of pretrial risk assessment programs, more than 100 civil rights organizations have called out these programs in a joint statement for worsening racial bias within the criminal justice system. Ranging from the ACLU to the NAACP, the organizations have criticized the proposition for applying sweeping changes to the courts without providing transparency and deterring decision making to unreliable computer programs. Rather than focusing on eliminating racial bias at its core within the justice system, Prop 25 would only shift the blame to computer algorithms, ultimately not fixing anything.
Effectiveness Is Still Untested
One of the biggest concerns with Prop 25 is whether or not it will make California safer. Proponents have argued that the pretrial risk assessment programs can accurately predict the likelihood of a released defendant committing another crime, but this is not supported by substantial data.
When looking at the situation in Florida, the programs were more likely to release defendants with extensive criminal histories rather than first-time offenders based solely on the defendant’s race. In this situation, the program was also found to report false positives and release violent white defendants at a higher rate than non-violent black defendants. This suggests that it was far more influenced by racial bias than actual public safety.
In turn, the New York situation is a little unclear. By January 2020, the NYPD reported that violent crime increased following New York’s massive bail reform program; however, this data has not been fully analyzed and has been subject to criticism from public defenders. Given that the program only lasted for a short period, it is hard to say whether it was effective or not.
But, as stated earlier, employing these programs on a large-scale across California is dangerous, and our counties should not be used as testing sites for these unreliable programs. Until a better, more effective system can be properly reviewed, tested, and implemented in a safe and efficient manner, Californians should avoid making drastic changes to the bail system.
Prop 25 Creates More Problems Than Solutions
Ultimately, the decision is down to California voters. Prop 25 is a veto-referendum, which means that it has previously passed through the California legislature and governor’s office but is contested by voters. Given the information and criticisms provided by civil rights leaders, data scientists, and experts from across the country’s finest research institutes, approving the massive changes brought by Prop 25 could be disastrous for our criminal justice system and economy.
When you look at the case in Florida, you can see how these programs can lead to more unfair, racially motivated detentions rather than actual reform. In turn, New York is a clear warning sign that these lofty changes cannot be implemented in the long run without massive costs and unreliable results. No one will dispute that California needs reform within its criminal justice system, but completely eliminating bail is not the solution. Our state is complex and diverse, and it needs more intricate solutions than Prop 25 offers. Voting no on Prop 25 gives California the time to develop more effective and fair systems for its citizens without subjecting them to costly and erratic software.