Criminal Injustice and Avatars
by Sherry F. Colb
In my Verdict column for this week, I discuss a proposal offered by Adam Benforado, author of Unfair: The New Science of Criminal Injustice, that instead of trying criminal cases "live" before juries, we instead record the trials and edit out inadmissible material, objectionable questions, along with objections and rulings on those objections, only then showing the results to juries and thus protecting jurors from information that would likely taint their consideration of cases. In the interview with Benforado in which he discussed this proposal, he mentioned a second one as well, one that I will examine in this post.
Benforado suggested (in the interview and, presumably, in the book as well, which I have only just begun to read) that instead of the jury watching a recording of the actual witnesses at trial, the recording of the trial should replace the witnesses with avatars. Let us evaluate both upsides and downsides of this proposal. Because it may sound preposterous and even bizarre to some readers, I will begin with the upsides.
One of the injustices embedded in the existing criminal justice system is the way in which juries tend to evaluate the credibility of witnesses. Research supports the conclusion that the following characteristics of witnesses make them appear more credible to juries: whiteness, attractiveness, and slimness. There are other such traits, but even just these three are alarming. That a jury might accept the testimony of one witness as true because he is white and reject the testimony of another because she is a large person or unattractive is extremely disturbing, and what makes it even worse is that jurors seem to be completely unconscious of harboring these biases. That is, if asked "why did you believe him?," jurors are likely unable to detect the fact that his race (or weight or attractiveness) played any role. If the first step toward fixing a problem is acknowledging it, I would think it might be difficult to acknowledge something that occurs outside of consciousness and correspondingly hard to "fix the problem."
Not without reason, then, Benforado wants jurors to be able to hear and see witness testimony without knowing witnesses' race, weight, or attractiveness. The way to do that, he concludes, is by replacing the faces of actual witnesses with avatars. I assume that all of the avatars will be equally attractive or unattractive, equally slim or heavy, and colorless (or perhaps all of the same unnatural facial color, like gray or light blue). In this way, we could perhaps eliminate the biases that ordinarily interfere with jurors' making valid judgments about whether or not to believe witnesses.
I find this proposal intriguing. I cannot say I have ever even thought of the possibility, and it could indeed have the salutary effects that Benforado anticipates. However, I am deeply skeptical not only of its legality but of what, in fact, would remain of the criminal trial as we know it if we were to adopt the proposal.
In terms of legality, the Sixth Amendment right of confrontation guarantees a criminal defendant the right to be confronted, at a jury trial, by the witnesses against him. It seems to me likely that any court addressing the question would consider the inability of the jury to see the faces of the actual witnesses who are testifying (and instead viewing computer-generated avatars) to be incompatible with the confrontation right. If we look closely at why trials have witnesses (as opposed to a pile of affidavits), it is that we want the jury to be able to observe the demeanor of witnesses and the way in which they talk, gesture, and behave generally, and thereby make a kind of "human lie-detector" assessment of the witnesses' credibility. Absent a view of the witness him- or herself, it is hard to know how a juror would go about assessing witness credibility. It would seem, to say this differently, impossible to take away the witness's race, weight, and attractiveness while somehow retaining enough of his or her appearance to allow the jury to evaluate demeanor evidence.
One answer may be to suggest that assessments of demeanor evidence are overrated and myth-based (as Benforado does suggest, relying on scientific evidence) and that jurors should focus instead on inconsistencies in the witness's statements, on motives the witness might have to shade the truth, on capacity defects of the witness (an inability to remember key details about the crime) or on prior acts of dishonesty. If so, then perhaps what we need is not avatars but instead a simple sound recording for the jurors, though there might be clues to (or indicators of) weight and of race in the sound of one's voice, so we might also need to use sound disguisers. At that point, it would seem that we are extremely far away from the trial that the Sixth Amendment contemplates and far closer to affidavits (with Siri narrating).
At the same time as I raise these problems, I remain curious about how an avatar system might work. Perhaps we could take old trials where innocent people have been convicted and run them with avatars for mock juries and see whether that improves matters. In either event, I am impressed with Benforado's creativity in trying to generate ideas that could help fix our profoundly broken criminal justice system. And who knows? Maybe avatar witnesses will prove some day to be such a great improvement on what we have that our Constitution can withstand it.
In my Verdict column for this week, I discuss a proposal offered by Adam Benforado, author of Unfair: The New Science of Criminal Injustice, that instead of trying criminal cases "live" before juries, we instead record the trials and edit out inadmissible material, objectionable questions, along with objections and rulings on those objections, only then showing the results to juries and thus protecting jurors from information that would likely taint their consideration of cases. In the interview with Benforado in which he discussed this proposal, he mentioned a second one as well, one that I will examine in this post.
Benforado suggested (in the interview and, presumably, in the book as well, which I have only just begun to read) that instead of the jury watching a recording of the actual witnesses at trial, the recording of the trial should replace the witnesses with avatars. Let us evaluate both upsides and downsides of this proposal. Because it may sound preposterous and even bizarre to some readers, I will begin with the upsides.
One of the injustices embedded in the existing criminal justice system is the way in which juries tend to evaluate the credibility of witnesses. Research supports the conclusion that the following characteristics of witnesses make them appear more credible to juries: whiteness, attractiveness, and slimness. There are other such traits, but even just these three are alarming. That a jury might accept the testimony of one witness as true because he is white and reject the testimony of another because she is a large person or unattractive is extremely disturbing, and what makes it even worse is that jurors seem to be completely unconscious of harboring these biases. That is, if asked "why did you believe him?," jurors are likely unable to detect the fact that his race (or weight or attractiveness) played any role. If the first step toward fixing a problem is acknowledging it, I would think it might be difficult to acknowledge something that occurs outside of consciousness and correspondingly hard to "fix the problem."
Not without reason, then, Benforado wants jurors to be able to hear and see witness testimony without knowing witnesses' race, weight, or attractiveness. The way to do that, he concludes, is by replacing the faces of actual witnesses with avatars. I assume that all of the avatars will be equally attractive or unattractive, equally slim or heavy, and colorless (or perhaps all of the same unnatural facial color, like gray or light blue). In this way, we could perhaps eliminate the biases that ordinarily interfere with jurors' making valid judgments about whether or not to believe witnesses.
I find this proposal intriguing. I cannot say I have ever even thought of the possibility, and it could indeed have the salutary effects that Benforado anticipates. However, I am deeply skeptical not only of its legality but of what, in fact, would remain of the criminal trial as we know it if we were to adopt the proposal.
In terms of legality, the Sixth Amendment right of confrontation guarantees a criminal defendant the right to be confronted, at a jury trial, by the witnesses against him. It seems to me likely that any court addressing the question would consider the inability of the jury to see the faces of the actual witnesses who are testifying (and instead viewing computer-generated avatars) to be incompatible with the confrontation right. If we look closely at why trials have witnesses (as opposed to a pile of affidavits), it is that we want the jury to be able to observe the demeanor of witnesses and the way in which they talk, gesture, and behave generally, and thereby make a kind of "human lie-detector" assessment of the witnesses' credibility. Absent a view of the witness him- or herself, it is hard to know how a juror would go about assessing witness credibility. It would seem, to say this differently, impossible to take away the witness's race, weight, and attractiveness while somehow retaining enough of his or her appearance to allow the jury to evaluate demeanor evidence.
One answer may be to suggest that assessments of demeanor evidence are overrated and myth-based (as Benforado does suggest, relying on scientific evidence) and that jurors should focus instead on inconsistencies in the witness's statements, on motives the witness might have to shade the truth, on capacity defects of the witness (an inability to remember key details about the crime) or on prior acts of dishonesty. If so, then perhaps what we need is not avatars but instead a simple sound recording for the jurors, though there might be clues to (or indicators of) weight and of race in the sound of one's voice, so we might also need to use sound disguisers. At that point, it would seem that we are extremely far away from the trial that the Sixth Amendment contemplates and far closer to affidavits (with Siri narrating).
At the same time as I raise these problems, I remain curious about how an avatar system might work. Perhaps we could take old trials where innocent people have been convicted and run them with avatars for mock juries and see whether that improves matters. In either event, I am impressed with Benforado's creativity in trying to generate ideas that could help fix our profoundly broken criminal justice system. And who knows? Maybe avatar witnesses will prove some day to be such a great improvement on what we have that our Constitution can withstand it.