Why Bail Reform is a Bad Idea
There has been a continuous argument by activist, judges, and advocates seeking to abolish bail practices on the grounds of unfairness. The activists argue that indigent defendants cannot post similar bonds to the wealthy ones. Bails offer yet-to-be adjudicated defendants a chance to be free as they await trial, allowing them to participate in their defense process. Bonds also provide surety to the society that the accused person will appear in court for a hearing. In the recent past, many states across The US have advocated for bail reforms, and some states have already implemented the proposed changes. It is essential to understand the importance of upholding the bail system and the shortcomings of the bail reforms.
In Support of a Bail System
Donald Verrilli, who served as a solicitor general to the former US president Barack Obama was in support of the bail system. He stated that bail served as a reconciling mechanism that accommodated both the interests of the defendant and society. Bond offers the accused pretrial liberty and fulfills the society’s interest in making sure the accused will appear in court.
There are several bail mechanisms, including cash, property, or third party responsibility. Irrespective of the mechanism used, bail is useful in ensuring that the accused honors their obligation to undergo trial under the criminal justice system.
What Solutions Do Bail Critics Offer?
Bail critics boldly state that algorithms should decide who remains locked-up and who walks free. They seek to eliminate human discretion and the bail system, and instead release defendants based on some algorithms generated by computers. The proposed reforms are not as effective as bail in ensuring that a defendant reappears for the court trial, and they are not as effective as bail in ensuring public safety.
Shortcomings of Bail Reform
Bail critics argue that abolishing bail bonds will help achieve fairness and safety in the criminal justice system. However, it is impossible to make equity and reduce disparities in judicial systems by using algorithms. The critical shortcomings of bail reforms include:
- Bias
An independent, nonprofit newsroom that conducts investigative journalism with moral force, pointed out that the software used to predict future criminals is biased against blacks.
In 2014, the then US Attorney General Eric Holder warned that risks scores from risk assessment software might be injecting bias into the courts. He observed that although the bail reform measures had good intentions, they undermined efforts to ensure individualized and equal justice. He pointed out that the measure escalated unwarranted and unjust disparities in the criminal justice system.
Upon visiting Rensselaer Polytechnic Institute in April 2017, Chief Justice John G. Roberts Jr. received a startling question from the then president of the college. She asked the Chief Justice if he could foresee a day when smart machines, run on artificial intelligence, would assist with courtroom fact-finding and even judicial decision-making. To the amazement of everyone present, the Chief Justice observed that the day was already there and it was putting a significant strain on the judiciary’s way of doing things.
Two cases of petty theft bring out the bias exhibited by the use of algorithms. While running late to pick her good sister from school on a spring afternoon in 2014, Brisha Borden saw an unsecured kid’s blue bicycle and a scooter. Brisha and a friend took the bike and scooter and attempted to ride them before realizing they were too small for them. Suddenly a woman came running after them claiming that they should leave her kid’s bike. By this time, a neighbor who was observing had already called the police on them. Both Brisha and her friend faced petty theft charges amounting to eighty ($80) dollars.
The previous summer, 41-year-old Vernon Prater shoplifted tools worth eighty-six dollars and thirty five cents ($86.35) from a home depot store. Prater was a criminal, had earlier armed robbery charges, and attempted robbery charges. He has even served five (5) years in prison for the crimes. Borden also had a record, but for misdemeanors she had committed while she was a juvenile.
After both Brisha and Prater were booked in jail, a computer program issued a score predicting the likelihood of each committing a future crime. Brisha Borden, who was black, had a higher score while Plater, who was white, had a lower score. After two (2) years, it was evident that the computer program got it all wrong. Brisha did not have additional criminal charges. However, Plater was serving an eight-year jail term after breaking into a warehouse and stealing electronics worth thousands of dollars.
It is clear that scores similar to those used in deciding Brisha Borden and Prater’s cases are biased. However, risk assessments are common in courtrooms across the United States. They come in handy when making decisions about whom to set free at various stages of the criminal justice system. In places like Arizona, Delaware, Kentucky, Oklahoma, Washington, Wisconsin, Colorado, and Louisiana, judges receive results of these assessments during criminal sentencing.
ProPublica carried out a study of the risk assessment software Correctional Offender Management Profiling for Alternative Sanctions – COMPAS, to examine the powerful and mostly hidden effect of algorithms in American life. They found out that:
The risk scores were unreliable in predicting future violent crime, and only 20% of the people forecasted to perform violent crimes did
In forecasting who would re-offend, the algorithm was likely to label people of color as future criminals while the white defendants were termed as low risk and unlikely to commit further crimes.
- Inaccuracy
Criminal justice would be fairer if only computers were accurate in predicting who is likely to perform future crimes. However, if the computer scores are wrong, a dangerous criminal may walk free. Still, if the computer gets it wrong, a defendant may unfairly receive a harsher statement than deserved or wait much longer than deserved for parole.
Judges use algorithms for predicting recidivism while making decisions for pretrial, parole, and sentencing. Activists for bail reforms argue that the systems use big data and advanced machine learning that make the results accurate. They also say that the results obtained from COMPAS risk assessment software are free from human error and bad judgment. However, a research paper published on Science Advances on January 17, 2018, titled “The Accuracy, fairness, and limits of predicting recidivism” revealed that COMPAS is no more accurate than predictions made by individuals with little or no expertise in criminal justice. COMPAS has a collection of one hundred and thirty-seven (137) features, but it is easy to achieve its accuracy level using a simple linear classifier with two (2) elements.
- Failure to Appear in Court
Bail bonds acted as a force that ensured that the defendant reappeared in court after the pretrial release. Failure to appear in court would mean that the defendant had decided to forfeit the bail placed. This is not the case for release based on computer algorithms. Advocates for bail elimination often refer to success stories like New Jersey where bail reforms have taken effect. The activists claim that the number of pretrial detainees in New Jersey dropped by twenty (20%) percent after the state adopted the reforms. However, when sharing the success story, New Jersey does not reveal the recidivism and failure to appear in court rates for the released defendants. It would be transparent if the state would release the statistics for non-appearance in court for defendants released after bail reforms as well as statistics for defendant released during the bail system.
- Increase in Crime by Released Defendants
Defendants released through COMPAS algorithms have a higher likelihood of committing crime after their release. Kentucky was among the first states to adopt risk assessment software in the judicial system. Kentucky is often quoted as the state with the best practices in the pretrial area. The 2011 Kentucky law that made risk assessment a necessary part of bail decision made a remarkable change in the bail system but only recorded a small increase in pretrial release. The increase in pretrial release came at a cost; failure-to-appear and pretrial crime increased as well. The bail reform changes in Kentucky eroded overtime and judges favored their old way of making a judgment.
- Increase in Racial Disparity
Bail reform activists argue that risk assessment tools help reduce racial disparity. However, studies reveal that the risk tools could be increasing the racial inequality they seek to reduce. In 2016, Jennifer Skeem researched at the University of California together with Berkeley and Christopher T. Lowenkamp from Administrative Office of U.S. Courts. The team evaluated the validity of a risk assessment tool, not COMPAS, used to make probation decisions. They evaluated the scores of about thirty-five thousand (35,000) federal convicts. The results indicated that blacks had higher average scores but made a conclusion that the ratings were not due to bias.
Further Limitations of Risk Assessment Scores
In Napa County, California, the probation department makes use of risk assessments while suggesting to the judge the appropriate probation or treatment plan for individuals facing sentencing. A Napa County Superior Court Judge, Mark Boessenecker, pointed out that the suggestions are helpful. However, the same judge later stated that the risk assessment scores do not reveal whether a person is dangerous and if they should go to prison. Boessenecker trains other judges around California State on evidence-based sentencing.
In the words of Boessenecker, a judgment based on risk assessment scores may see a person who has molested a child every day for a whole year come out as a low-risk person. On the other hand, a drunken person would come out as high risk simply because he is homeless. The risk factors scores should not be used to determine whether a person is dangerous or not. Instead, they help to determine whether a person needs probation treatment or not.
Risk Assessment Tools Puzzle Defendants Declared Low Risk
At times, the reports from the risk assessment tools used under new bail reforms do not make sense to the defendants. James Rivelli, a Hollywood, Florida man, was arrested and charged five (5) years ago after he shoplifted seven (7) boxes of Crest Whitestrips from a drugstore. James had a criminal record that included multiple thefts, aggravated assault, and felony drug trafficking. However, the commonly used COMPAS classified him as being a low risk of offending.
James Rivelli was genuinely surprised when a reporter informed him that he had scored a three (3) out of a possible ten (10). He said that he was astonished that his score was as low as he had spent five (5) years at a state prison in Massachusetts. He concluded that his criminal history did not count in Broward County. This is not true as criminal records from across America should always be included in the assessment scores. Less than one (1) year later, James Rivelli faced two (2) felony charges after shoplifting tools worth one thousand ($1,000) dollars from Home Depot. In defending himself, Rivelli argued that he committed the crimes due to drug addiction and that he was then sober.
Bail Reforms in California
On August 28, 2018, the then California Governor Jerry Brown signed the SB10. He touted the Senate bill as bail reform. In the words of Jerry Brown, California reformed its bail system so that the rich and the poor would receive equal treatment. The bill sought to end money bail as well as the provision of bail bonds. However, the proposed bail reform does not affect profit people and may only cause Californian’s more suffering. According to the new laws, people no longer pay dollar bails to get out of prison. Instead, algorithms developed by private institutions help determine the likelihood of a person to recommit a crime and the possibility to miss court dates.
The scores obtained from the risk assessment tools fall under low, medium, or high risk. The low-risk people are set free, although the judges have the liberty to cancel the assessment recommendation and order incarceration. Based on the rules created by local jurisdictions, the defendant with an average risk score may be released or go to jail. Unlike under some special conditions, people with high-risk assessment scores go to jail.
The risk assessment tools do not make decisions based on a person’s circumstances. Instead, they compare individual facts about a person such as age and past criminal convictions to others with similar profiles. The risk tools assign scores based on whether other people in similar circumstances missed court dates or were re-arrested.
Usually, the factors the tools consider are secret and only known by the designing company. It becomes hard to determine how the algorithms arrived at a score. It is therefore tough to challenge the scores in court.
The risk assessment tools determine the risk of re-arrest. Usually, police mainly patrol neighborhoods where poor people live and where people of color live. Police will rarely patrol the affluent white communities. Frequently patrolled areas are likely to have high rates of re-arrests. The risk assessment tools that assess the risk of re-arrest predict police behavior.
The risk assessment tools have errors where black people are classified as high risk nearly twice as high as white people. The system also wrongly classifies white people as low risk at a much higher rate than black people. The assumed errors made while assessing risk undermines the claim by Senate Bill 10 supporters that rich and poor people will have a fair and equal treatment.
The scope of each risk category such as low, medium, or high, is not derived scientifically; thus, it is not objective but discretionary. The judges who are in control of the scoring may adjust the size of each category to jail as many or as few people as they desire.
Under the reformed law, jail means there is no option for release. Under the bail system, a poor person would acquire bail bonds and later make payments to a bonds provider. This would enable the poor to get out of jail.
Under the California bail reforms, the judges can make preventive detention. The judges do this if they have an adequate belief that non-monetary conditions or a combination of conditions of pretrial release will not assure protection of the public or the victim, as well as the appearance of the defendant in court. However, this provision is not supported by any guiding principles, and any limits of judicial discretion do not bind it.
The Court of Appeals condemned judges for setting money bail so high making it function as a form of preventive detention in the California bail case In re Humphrey. Under the bail reform, judges can order preventive detention without setting money bail. The new law does not constrain judges from making any decisions. It does not define the standards under which judges are to jail people without an option for release. Therefore, the new law gives judges the power to take away a defendant’s policy without due process.
Legislative proponents of the bail reforms argue that without cash bail; judges will use the risk assessment tools in a way that puts a few people in detention. However, in California, judges are popular for expanding pretrial incarceration. A considerable percentage of people in California jails are pretrial detainees. The California bail reform is a major setback for the criminal justice reform movement.
Bail Reform is Not a Fair Idea
The increasing use of risk scores in the judicial system is controversial. Risk assessment tools were initially useful to judges and gave them insight into the various treatment types an individual might need, such as drug treatment option and mental health counseling. However, using the tools to judge defendants during a sentencing hearing may result in incarceration. Usually, defendants do not get any chance to challenge the scores from the risk assessment tools. Risk assessment tools are not as effective as cash bail in ensuring that defendants honor court dates after release.
