PROSECUTORIAL ANALYTICS.

• Author: Kreag, Jason

ABSTRACT

The institution of the prosecutor has more power than any other in the criminal justice system. What is more, prosecutorial power is often unreviewable as a result of limited constitutional regulation and the fact that it is increasingly exercised in private and semi-private settings as the system has become more administrative and less adversarial. Despite this vast, unreviewable power, prosecutors often rely on crude performance measures focused on conviction rates. The focus on conviction rates fails to capture and adequately evaluate the breadth of prosecutorial decisionmaking. We can do better by fully implementing analytics as a tool to evaluate the prosecutorial function. This tool has revolutionized crime-fighting. Yet, it has been conspicuously absent as a tool to improve other aspects of the criminal justice system. This Article demonstrates the promise of prosecutorial analytics to improve oversight and to promote systemic interests in justice, fairness, and transparency. It offers concrete examples of how analytics can 1) help eliminate race-based jury selection practices; 2) minimize prosecutorial misconduct; 3) uncover whether undesirable arbitrary factors shape prosecutorial discretion; and 4) provide better metrics for the judiciary, practitioners, and the public to evaluate prosecutorial performance.

TABLE OF CONTENTS INTRODUCTION I. THE RISE OF ANALYTICS IN THE CRIMINAL JUSTICE SYSTEM A. Analytics as a Crime-Fighting Tool B. Constitutional Criminal Procedure's Small Data Focus C. Factors Contributing to the Analytics Imbalance 1. Political Considerations 2. Structural Limitations 3. Measurement Challenges 4. Competency Limitations II. EXTENDING ANALYTICS TO THE PROSECUTORIAL FUNCTION A. Checking Prosecutorial Power B. Moving Beyond Conviction Rates C. Improving Constitutional Decision-Making With Analytics III. ANALYTICS BEYOND CRIME-FIGHTING A. Jury Selection B. Identifying Supposedly Irrelevant Factors 1. Over Politicization 2. Potential Arbitrary Factors C. Charging and Bargaining Decisions D. Prosecutorial Disclosure Obligations IV. THE CHALLENGES OF PROSECUTORIAL ANALYTICS A. Risks of Analytics B. Mandatory Prosecutorial Disclosures CONCLUSION INTRODUCTION

The oral argument before the Supreme Court in Tim Foster's case began with Foster's counsel asserting, "Mr. Chief Justice, and may it please the Court: The prosecutors in this case came to court on the morning of jury selection determined to strike all the black prospective jurors." (1) Beginning an argument with such a sweeping assertion is a risky strategy. An unsympathetic jurist might interrupt with a question immediately challenging the factual basis for the assertion. The ensuing back and forth may prevent discussion of the underlying substantive claim. But Foster's case was different.

Unlike the vast majority of cases involving claims pursuant to Batson v. Kentucky, (2) in which courts must determine whether a prosecutor committed purposeful discrimination in jury selection without access to the most relevant evidence--the prosecutor's jury selection notes--Foster had the notes and they were before the court. (3) Thus, the Court had the opportunity to review the contemporaneous, and un-scrubbed, evidence of the prosecutor's decisions free of any post-hoc rationalizations. (4) The notes were damning and consistent with the prosecutor's actions. (5) After striking all of the potential black jurors, the prosecutor appealed directly to racialized fears, asking the all-white jury to return a death sentence to "deter other people out there in the projects from doing the same again." (6)

In the typical Batson case, the Court does not have the benefit of such direct evidence. Without it, judges are forced to perform Batson's critical third step, which requires them to be lie detectors, by simply evaluating the prosecutor's demeanor to determine whether the prosecutor lied to mask purposeful discrimination. Even under the best of conditions, this amounts to guesswork. (7)

There is an alternative approach. Courts could supplement their review with fairly basic statistical analysis and data analytics to diminish the demeanor-based guesswork. (8) The incorporation of analytics could simultaneously improve judicial decisions and decrease discriminatory jury selection practices.

Analytics can also help evaluate charging practices, identify misconduct, and uncover prosecutorial decisions based on factors that should not be considered. (9) This Article identifies these examples to demonstrate that analytics is an underused tool. It argues that analytics offers promise as a tool to: 1) regulate prosecutors' expanding power; 2) more accurately measure prosecutorial performance; and 3) improve constitutional decision-making.

It seeks to bring the institution of the prosecutor in line with one of society's defining characteristics--a seemingly insatiable desire to measure everything, to record as much as possible, to quantify our actions, and to mine the data for trends, patterns, and distortions to solve problems and make better decisions. This Article refers to this process as analytics. Others have offered a similar definition for Big Data. (10) Analytics is a better term here because not all of the applications of this tool to prosecutorial decision-making involve the "three V's" of Big Data--volume, velocity, and variety." Rather, analytics includes the use of small, large, and big data sets.

While the prosecutorial function has largely escaped the wave of analytics, the technique has revolutionized law enforcement's crime fighting efforts and is ubiquitous in modern life. Outside of the criminal justice system, the desire to measure and count things plays out on a grand scale (12) and on the personal level. (13) Organizations use this data to market products, measure risk, search for cures to disease, rank colleges, identify potential voters, and win sporting events. (14) Within the criminal justice system, actors have been quick to adopt analytics as a crime-fighting tool, promoting so-called smart policing or predictive policing techniques. (15)

Yet the rapid adoption of analytics as a crime-fighting tool has been spotty, and, at times, not thoughtful. (16) More importantly, it has been almost entirely prosecution-based in ways that raise serious concerns about fairness to defendants and that makes for a lost opportunity in advancing other worthy prosecutorial ends. Seeing this, several scholars and policymakers have called for more refined reflection about the implications of its use. (17) They worry that reliance on analytics distorts values, (18) risks exacerbating longstanding inequities in the criminal justice system, (19) and masks motives and biases. (20) As a result, they have proposed regulations that would limit the use of analytics as a crime-fighting tool. (21)

These calls for analytics speed bumps have great merit. But they miss a crucial step: expanding the use of analytics to even the playing field. This Article seeks to effect a better analytics balance in the criminal justice arena, proposing specific ways of deploying analytics to serve better prosecutorial justice, fairness, and transparency, and to assure that defense-specific interests are part of the data calculus. It argues that analytics can be a powerful justice-enhancing, bias-detecting, and innocence-uncovering tool. (22)

The need for defendant- and fairness-sensitive analytics is especially compelling when applied to criminal prosecution. The prosecutor is unique in the criminal justice system because of her near unreviewable power and discretion. (23) Furthermore, the prosecutor's power has grown in absolute and comparative terms over the last several decades. (24) Yet, prosecutors and those who observe them still often rely on crude measures of their performance, i.e., crime rates and conviction rates. These measures overlook important questions about how prosecutors exercise their discretion in charging defendants, negotiating plea deals, meeting constitutional and ethical disclosure obligations, and, among other decisions, selecting juries. (25) These factors should be considered in addition to crime rates and conviction rates when evaluating whether prosecutors adequately serve broad interests injustice, fairness, and transparency.

The existing regulatory regime for prosecutors is inadequate to protect these systemic interests. (26) Many prosecutorial decisions are insulated completely from the reach of constitutional and ethical regulations. And where applicable, the Constitution and ethical regulations only provide minimal guidance. The democratic process fills too little of this regulatory void, (27) because the work of the prosecutor increasingly takes place in private or semi-private proceedings, (28) and prosecutors' broad discretion leaves prosecutorial elections susceptible to manipulation. (29)

Analytics can help. Analytics can mine historical data to help identify prosecutors who might be more likely to commit misconduct in the future. It can help identify prosecutors or prosecutorial offices that engage in or are more likely to engage in race-based jury selection practices. It can also help identify undesirable trends in charging and plea bargaining. Simultaneously, it can provide the public with better data to evaluate prosecutorial performance. We are unlikely ever to approach the equivalent of open-source prosecuting, but analytics gets us closer. And in doing so, it promises dramatic increases in transparency.

Extending analytics beyond the crime-fighting side of the criminal justice system will require access to the relevant data--data that is often controlled by or in the possession of the prosecutor. Some prosecutors have demonstrated a willingness to share data with outside evaluators to measure justice and fairness. (30) In other instances, public interest organizations and researchers have undertaken herculean...