Presumption of Innocence - Episode 93

Episode 93 • 8th September 2026 • Fox Rothschild: The Presumption of Innocence • Matt Adams & Matt Lee

Speaker: The views expressed in this podcast are those of the participants and should not be considered the views of Fox Rothschild LLP or its attorneys.

This podcast is for informational purposes only, is not legal advice, and does not create an attorney-client relationship

Matt Adams: Welcome back to The Presumption of Innocence, a podcast brought to you by the White Collar Criminal Defense & Government Investigations Practice at Fox Rothschild. My guest today is Professor Brandon Garrett. He's the David Eichel Distinguished Professor of Law and the director of the Wilson Center for Science and Justice at Duke University School of Law.

He's one of the nation's leading scholars on criminal justice outcomes, evidence, and constitutional rights using empirical research to demonstrate how criminal proceedings go wrong. Professor, it is a real privilege to have you on the program. Welcome.

Brandon Garrett: It's great to be here. Well, and given that we're, we're speak- I'm speaking to you at the home of your white collar practice, although it's not really the topic today, one of the big long-term empirical projects I've worked on for years has been tracking federal corporate prosecutions.

And I wrote a book years ago called “Too Big to Jail” describing all the challenges when prosecutors go up against corporations and what are the dynamics in those cases. We’re still updating the corporate prosecution registry. Every year there are more, you know, cases to add to that registry. It's like, you know, more than 3,000 cases going back to, going back 25 years.

So that's-

Matt Adams: Well, yeah ...

Brandon Garrett: That’s not our topic today really, but, but …

Matt Adams: One, one of, your certainly eight books that's been out, but your most recent book is one of the topics I wanna pivot to quickly at the outset here, which is it's called “Defending Due Process: Why Fairness Matters in a Polarized World.”

This is a transcendent topic that really covers the range of criminal and regulatory enforcement. You write in your book that people across the political spectrum are tempted to put desired outcomes before a fair process. For a show called The Presumption of Innocence ... that feels like a real critical tension.

Can you tell us professor, based on your empirical research, is due process winning or losing? What's the deal and what's the score?

Brandon Garrett: Well, I mean, we're just starting with the book title, which you mentioned, “Defending Due Process: Why Fairness Matters in a Polarized World.” I had no idea when I started working on the book or even when I submitted, you know, the full draft of the book a couple years ago, that basic due process protections would become such a national topic this past year in the news but also in our courts.

We've seen all kinds of due process challenges on topics that are new and plenty of due process challenges on topics that are old. But, you know, one of the reasons I wrote the book, it's kind of defunct now. I was trying to show that low-level due process issues and the types of proceedings that no one focuses on unless it happens to them, like suspension of your driver's license, bail hearings, Medicaid hearings.

I, I mean, I talked about more serious criminal cases and, and other types of situations too, but I wanted to highlight at the local level, there are all these neglected procedural due process problems, and it amounts to something bigger. It amounts to a national problem. Well, we don't need to be convinced so much that it's a national problem so much because there's been so much attention to due process, which is good, over the last year, but there also been so many due process challenges, which is not so good.

Matt Adams: The central thesis of your book, and perhaps I guess the book's core argument, is that the democratic value of due process is somewhat up for grabs, and it's up for grabs in, it's hard.

Brandon Garrett: It’s hard, it’s never easy, yeah ...

Matt Adams: It's up for grabs in courtrooms, colleges, police stations, and, and in this amorphous echo chamber of our online world that has just become so prevalent in our society.

It's pretty broad and pretty high intellectual stakes for, for that to be occurring. What do you think the root cause is, professor, for such a disinterest is, is the best way I think I can put it, in notions of a fair process and some value to a large segment of our population, and just skipping right ahead to the outcome without worrying about what happens in the middle.

Brandon Garrett: I think, you know, and I talk about this, there's a... I've done a lot of research and others have done a lot of research on people's attitudes, and people really do care about fairness and due process a lot. But they, it hits people the most when it's happening to them. And so, you know, obviously, like, if it's, if you're the one who got a traffic ticket for no reason, if you're the one who gets some government notice saying, "We're gonna seize your car, your house," and you don't know why, that's serious and you get really upset and you're like, "Why, why am I learning about this now?Why … there’s gotta be some way to challenge this. I gotta have a day in court."

People really care about their day in court when it's them. But it's, it's, it's understandable, but really easy when it's not you, when it's third person, when it's someone else, to say, "That person got arrested for that? Like, why do we even have a trial? Like, that person just needs to be put away." Or, "That person committed fraud? Well, they should seize their house. They should take their car. They should take their driver's license. Someone accused of something like that shouldn't be walking the streets."

And you know, we, we even see some of the same people having it both ways and, and it's, it is kind of human nature. We've seen... I give the example of police unions, which have said, "We need to have the power to just arrest people and put them straight in jail. Why do we have these bail reforms or even bail hearings at all? Why should someone get a hearing when we arrest someone for a serious crime? Straight to jail." But of course, when we-

Matt Adams: Take our word for it, we're right every time ...

Brandon Garrett: But on the flip side, when police officers get accused of misconduct, they're like, "No, we need a police officer's bill of rights. We should have, like, a cooling off period, time to meet with our lawyers. We should have a hearing. We should have a lot of protections in place to make sure police officers don't get unfairly accused of misconduct." So when it happens to you, fairness really matters, but when it happens to them, it's harder to see it.

Matt Adams: You'll never, you'll never hear it on this program, us advocating for shortcuts in due process, that's for sure.

But let's go back to the beginning. You started your career at Neufeld, Scheck and Brustin, the firm synonymous with the Innocence Project and all of its good works.

Brandon Garrett: Yeah, two of the lawyers founded the Innocence Project.

Yeah. I mean, this was separate from the Innocence Project. It was a civil rights firm. But, you know, after I clerked for a federal judge, I, I worked at this firm where a lot of the cases involved either more regular civil rights police brutality cases, but also the science-y cases, and I actually tried to steer clear of the science-y cases at first.

Like, because these were cases where people had been exonerated by DNA testing and they were suing for for damages for being wrongly convicted all those years. But I was like, "Oh, DNA testing, that involves statistics." I went to law school because I don't like math and science. But, but I couldn't steer clear of those cases because they had a lot of them and they were, they were really interesting, and I ended up representing people like in the Central Park Five case and some well-known exonerees and, and getting into the world of forensic evidence and some of the science issues that I was hoping to avoid.

Matt Adams: Chris Fabricant from the Innocence Project has been on this very program.

Brandon Garrett: Yeah, yeah ...

Matt Adams: Talking about some, some of the blind spots that exist within the forensic science community that lead to these horrific wrongful conviction scenarios.

Brandon Garrett: And it was actually the forensics work, and so, and I, and I know Chris, and I've worked…We've written amicus briefs in some of the cases that he and his team have worked on, and I've, I've written a lot about forensics.

Some of the work I was doing was looking at, you know, well, okay, so we know, for example, that some forensic methods are more reliable than others, and that raises due process issues. If you're gonna convict someone based on something that's totally unreliable evidence, at some point that might even be an unfair trial.

The Supreme Court just said that last year in a case, that evidence that's so unreliable and so prejudicial could deprive people of due process. Like it would deprive people of due process maybe if you had a fortune teller testify at trial that, "I know because of my crystal ball this person did it."

Anyway, I was doing studies looking at, well, for more bread-and-butter forensic evidence, like fingerprint evidence, do people appreciate that there's some limits to that, that fingerprint examiners are...there's some subjectivity involved. There can be a lot of detail on a fingerprint, but they can make errors and they have made errors.

And so I was doing these studies and in these studies when you're testing out mock jurors, you often have a control condition. So the control condition was this: And, you know, I assume that you'd say good things to a client if they came to you and said "Look, I'd like you to represent me. You know, as far as I know, the police, you know, arrested me, but there were no witnesses, and they don't have any evidence. They just said, you know, 'We, we got a guy. We got you.' And I was, you know, within a mile or two of the crime scene, but there was nothing connecting me to the crime. There, there was no, there's no evidence in, in this case. The, the police just said, 'We got you. We got a guy.'" And so that was one of the conditions. There was no evidence. I assume presumption of innocence, that person shouldn't get convicted.

Matt Adams: Yeah. No doubt. It's hard not to see how those early experiences in practice in that civil rights firm really led and shaped the trajectory of your scholarship.

Brandon Garrett: In that study, in the condition where there's no evidence and there, and there's a presumption of innocence, and we instruct these mock jurors online, we surveyed, you know, tens of thousands of people at this point. In that particular study, we had this control condition with no evidence, and a lot of people convicted in that condition.

And so my psychologist colleague, Greg Mitchell, and I were "What?" How is it and we did see people will convict differently depending on how strong the fingerprint evidence was. These were people in general that were paying attention to the experiment. They were trying to do the right thing.

We had no reason to think that these were just like, there's something strange about those people. We're like, "What is going on here? Why are people willing to convict?" They'll convict, many fewer people will convict, but still it was like, I don't know, 45% of the people would vote to convict, and we figured like, are they just assuming police have some other evidence that they haven't heard about?

They just figure if police got a guy, there must be a good reason for it. Like, what is happening? And that started me on the road, on the research road to this book because we started trying to unpack, well, what is it about people? Do they not understand beyond a reasonable doubt and the presumption of innocence?

Or do they really think there must be evidence there? They're like reading things into the evidence. Like, what is happening? Why are people so willing to convict even when the evidence seems weak?

Matt Adams: And so what's the answer to that question? What is the evidence, what does the evidence-based empirical data say?

Is it that we are just willfully blind because we don't know better, or because it's just easier?

Brandon Garrett: What we have found over time is... And so we asked a question in this paper, and, and didn't even think about it. We just thought the... We didn't realize that people were gonna be convicting based on no...voting to convict with no evidence. But we figured people are gonna be different, people are gonna view the evidence differently, and it might have to do with their attitudes towards what the right answers are in criminal cases. And there's just a standard question that gets asked in all the national and international surveys about people's attitude towards the courts.

It's like a social scientist question. And it's... so this is the question. And so I, I'm guessing I know what your answer will be, and, and maybe your listeners too, since they're listening to the Presumption of Innocence podcast. So the question is this: Which type of error is more harmful to society?

One: Convicting an innocent person; Two: Acquitting a guilty person; or Three, I don't know. Like, which one, which one would you say?

Matt Adams: Easy, trivial. It is innocents.

Brandon Garrett: Yeah ... convicting an innocent person. And it goes back to, we like... We, as lawyers, we like to quote William Blackstone, you know, "Better that 10 guilty people go unpunished than one innocent person," be convicted.

Matt Adams: Deprived of their liberty. I think it's 100 people.

Brandon Garrett: Deprived of their liberty. Yeah. Yeah,

Matt Adams: Yeah.

Brandon Garrett: So maybe it... But even if there's no 10 to one ratio, I mean, like, we don't really want like 10 guilty people on the streets.

Matt Adams: I think it's 100 guilty people in the actual saying. Yeah.

Brandon Garrett: And the Supreme Court says this is the reason...that principle is what the Supreme Court cites when it says, "This is why we have beyond a reasonable doubt." Like, we wanna err on the side of not convicting innocent people. But, what we found was that — And, and we didn't ask the question that way because we're kind of like, "Wait a minute. It's either one or the other, or you don't know?" That- that's... You don't normally force people to choose when you do survey questions. So we also added a third option, which is “they're both equally serious.” What we found was two-thirds of the respondents said they're equally serious. And there was like, you know, maybe 15% of people who said the worst thing that can happen is you free a guilty person, and they were predictably more likely to convict.

They typically were older and conservative. There are also maybe like more like 20 something percent of people who tended to have degrees and be more liberal who said convicting the innocent, that is the worst. But then just the vast bulk of the public, and across these studies we've seen this, two-thirds of Americans, both errors are equally serious.

We care about both. We don't wanna see wrongful convictions. We don't wanna see unfairness. But we also care about public safety. We don't wanna see guilty people going free. We care about these values equally. So it, it's not like most people don't care about due process, but they don't care about it like as the most paramount important thing that matters above everything else.

And, and we see this, you know, Democrats, Republicans, old, young, men, women, Black, white, Hispanic, just across races, across political parties, just most people view both of those errors as equally serious. And that's the challenge. It's not like Americans don't care about due process. But if you tell people, like, "If you wanna be safe in your homes, well, forget about due process. We can't have these bail hearings. We just need to be able to throw people in jail when they, endanger your safety." Or if you, if you tell people it's a zero sum... people also think the reverse. Like, if they hear like, "Wait a minute, innocent people are being convicted? Well, we need to like...We need to do something right away. This is a disaster." The whole point of the book is to show that these things are not incompatible. In fact, like, just back to, like, the simple criminal case example, if you convict an innocent person, and I, I worked on some of these innocent guys' cases, the guilty person is not convicted, by definition, if a crime happened.

Like, both bad things happen in that situation. It is not good for public safety to convict an innocent person. The guilty person not only didn't get convicted, but they figure, like, "I'm never gonna get caught now." So there's some impunity involved.

Matt Adams: But society…but society sure does feel better about taking that punitive measure out against somebody.

Brandon Garrett: Against somebody. Against somebody.

Matt Adams: Yeah.

Brandon Garrett: Yeah. But it's obviously, it's a public safety disaster, and there are examples of, like, whole, like, repeat offender crime waves because innocent people got convicted, and the serial murderers and the like acted with impunity. These things are often not zero-sum.

I describe in the book how it's often not that expensive to provide better due process to people, and public safety can suffer because if you're punishing innocent people or fining people who don't need to be fined, unfair processes are really error-prone. You can be sanctioning the wrong people, and that doesn't, that doesn't help communities, doesn't actually make people feel safer, and so there are just so many negatives.

It’s better to slow things down and get it right when the government is doing serious things to us, like taking our liberty and taking our property.

Matt Adams: Well, one of the phrases that you use in the title of your book, Professor, “Defending Due Process: Why Fairness Matters” in a, quote, “Polarized World,” is probably the understatement of the year.

We are living in perhaps one of the most polarized times in our nation, at least, history, if not, I don't know, the history of the world. But the fact of the matter is you chose the phrase to be part of the title of your book, and what is it about that polarization that leads to this zero-sum game you describe, the central thesis of your book about there not really being much room for people to slow down, take a breath, and let the process play out?

Brandon Garrett: Yeah, when there's some group that you just think is like, "These people are no good," then you don't think that they deserve fairness. It's very easy to say, "No, no due process for them. Like, that's not…Why should we be giving them the luxury of a day in court? Why should we be acting like the law, the rule of law applies to them?"

“They're just you know, outcasts that should just, we should just punish them right away. We should just get rid of them.” And what goes around can come around. There, there are real serious problems in any society once you start eroding the rule of law, eroding due process like that.

The hope is, like, when you see it happening to you, then you realize how, how serious it is to treat someone in an arbitrary, summary way and not give them the benefit of a fair shake, give them the chance to actually explain their case, to make sure that if the government is doing something, it's doing it for an accurate reason.

But the biggest threat to due process is this idea that, you know, the process is that those are technicalities. We don't need to be doing that for everyone. Maybe for me, but not for them. And, interestingly though, I also think that, you know, people may just assume that even if these are people that, you know, politicians or even myself think like, "These are bad people, these are criminals, or these are fraudsters," whatever the category is, these are people who the government should, should be doing something to.

Many people also assume that they do get a fair day in court, and that there is more due process than there really is. We just did a big my, my psychologist friend that I mentioned earlier, Greg Mitchell and I, a big study asking people, "What kind of process, like, do you think people should have in a bunch of different common settings that aren't a criminal trial where there is, there, you know, you can, there's a jury, there's a defense lawyer, you have a right to counsel?"

It's, so these non-criminal trial, non-trial settings. And so we ask people like, "How much due process do you think someone should get if the government wants to take their car for not paying their taxes? How much due process do you think someone should get if it's a... " And it's obviously big in the news because of all the immigration enforcement increases in the country.

"If someone …If the government wants to deport someone from the country, how much due process should they get there? How much due process should someone, someone get if it's just a hearing about whether they're released or not on bail while they're facing criminal charges?" Before they've been ch, you know, there's a trial or anything like that.

Pre-trial. "How much due process should someone get if their partner is seeking a protective order, having accused them of domestic violence?" But that isn't, you know, going to a trial yet either. It's a question of like, is there a, do you do an emergency protective order? So we ask about a bunch of these different settings, like how much due process should people get?

And overwhelmingly, across all these settings, like, everyone answers well, yeah, of course there should be a presumption of innocence. Of course the burden of proof should be really high. Of course people should be provided a lawyer if they can't afford one. In all these settings, people consistently answer like there should be like really robust due process, and that's not how it works in this country.

You have no right to a lawyer at a bail hearing, or at least it's not very clear, and in most places, people don't have public defenders. Immigration hearings, no right to counsel. Most people do not have, have counsel in immigration hearings. You know, there's no civil version of Gideon versus Wainwright.

So if it's your car that the government wants to forfeit, there's no right to a lawyer there. You know, if you can afford one, good. That should help you. But you don't have a right to one. And in none of these contexts, there's no, like, beyond a reasonable doubt or presumption of innocence if the government's seizing your car for not paying your taxes.

There's no presumption of innocence beyond a reasonable doubt if it's an immigration hearing or a bail hearing. Bail hearing is supposed to be clear and convincing evidence, but it's not exactly in most places. So people assume that that things are pretty fair. I think maybe people, you know, see crime shows and, you know, you're into criminal law, I'm into criminal law.

Like, there's a lot of people who aren't lawyers who follow criminal law and assume that in their head, like, that's what a day in court looks like. That's what you get. And unfortunately, there are wrongful convictions, there are serious challenges in our criminal justice system, but the level of procedure is way, way scantier.

It, I mean, it's just, like, it's not exactly a judge, it's not exactly a court, and you don't really get a lawyer in all these other contexts. And people, and a lot of people don't realize that until it happens to them.

Matt Adams: How do you see the role of an academic center like the Wilson Center that, that you run at Duke in actually changing criminal justice outcomes as opposed to just studying them and articulating the empirical results out into the ether so that somebody else can deal with it?

Brandon Garrett: Yeah, obviously as a law professor, I'm teaching future lawyers and teaching them about due process, and I teach evidence, so talk about reliable evidence. So, you know, teaching future lawyers feels really good, and it's a way of educating the next generation of lawyers who are defending their clients and defending the rule of law.

So that's, that's something. And we also have a lot of students that work with us, and same thing. We want to teach them about the justice system and encourage them to get involved in their future careers. So working with students is one way. But you're right, like we can do studies, and if no one reads them and no one cares about them, then they won't have much impact in the world.

A lot of work that we do, although we definitely do studies, is in collaboration with government offices that really do want to change things or at least study things and decide what to do about it. And so we've worked with prosecutors that wanna track what's going on in the plea bargaining system.

On the due process front we've partnered with the Administrative Office of the Courts in North Carolina because there's a lot of concern with how many people have their driver's licenses suspended. And so we maintain that data each year that sort of the patterns are out there. There, there is a fair amount that local officials can do, including mass dismissing old cases, which...

And if they sort of see how many cases there are, they can do that. And so there, there, there's more of that kind of like a, I guess, engaged work that we do, where we're doing research, but we're doing it for government officials that are interested and wanna take some action based on it. And that, that feels particularly good.

It's sort of like at the intersection between public service and research.

Matt Adams: Well, you, you mentioned one of these registries earlier, but you also maintain a couple of available data sets that you put out to the public, the DNA Exonerations Database... the Corporate Prosecution Registry being the one that you've mentioned in our, in the course of our discussion here today, and then some death sentencing data that other scholars, practitioners, and the public writ large are able to rely on. What motivated you all to make that data open access, and how has it been used in ways that has surprised you? Because this notion of putting numbers, putting empirical support behind some of these loftier concepts that we like to talk about – truth, justice and the American way, if you will really, it gives them a shot of adrenaline for sure.

So talk to us about why.

Brandon Garrett: The very first project that I worked on which involved counting things, it was math that I felt like I could handle because at the time, there were only 200 DNA exonerations in the US, and so the percentages were easy to calculate.

Matt Adams: I can count to 200. I can count to 200.

Brandon Garrett: Exactly. I can... Like, that was a spreadsheet. That was easy. Now we're getting closer to 500. It's still... It's not as big as the National Registry for Exonerations, which includes non-DNA cases, so that's thousands of exonerations. This is hundreds of exonerations. But I went back and tracked down trial transcripts in those cases and, and that's been a big resource because, for example, there've been a few different death penalty cases recently where the question was like, okay, there's some DNA evidence which points to someone who's not the person sentenced to death, but the person confessed in detail.

And, what jumps out in the DNA exoneration cases where we, where the person has already been exonerated, is that lots of innocent people confess in detail. And it's because officers, if the interrogation isn't recorded, there's no way to know that they, in fact, supplied the key facts that made the confession statement seem really detailed and powerful.

And so, the evidence was corrupted in quite powerful and…ways that made the confession seem like detailed ones. An innocent person really can't say much about what they did. If you ask them to confess, they may resist doing it. If eventually they get pressured into falsely confessing, all they can say is, "I don't know, I did it."

Because if they're innocent, they don't know what happened. And what we've seen is that in the vast bulk of these cases, innocent people confessed in detail because prosecutors supplied the facts. There are a few cases actually where the innocent person just kept repeating like, "I don't know what I did, but sure, I did it," and they were never able to get the person to repeat after them.

But for the most part, officers succeeded in feeding facts to these innocent people. And so, it's been really important to show courts that can happen, that a confession that seems really detailed and corroborated can nevertheless be false. So, that, that's an example of...lawyers have used those examples around the country.

And to have a, a database where you can say, "Oh, this is a New York case. This is a Kentucky case. Well, we have examples from New York. We have examples from Kentucky," that's been really helpful too

Matt Adams: Well, I, wanna dig back into those issues some more because your, your scholarship is highlighting a lot of the patterns and trends that this data is actually driving.

So let's start with wrongful convictions. Your, your work entitled “Convicting the Innocent” analyzed the first 250 DNA exonerations in the United States and attempted to identify systemic patterns, false confessions, contaminated eyewitness iden-identifications, flawed scientific testimony and..

Brandon Garrett: We did. I just finished a big update of that database too, so we're now up to 455 DNA exonerations. There's a big set of new false confessions and the new cases that have been added to that registry of DNA exonerations, so we just did a big website. Haven't updated the book. I don’t know. Publishers don't always wanna do second editions of books. And I'm not sure I wanna do a whole second edition of the book, but I am updating the resource that's available to the public, to lawyers to look at cases.

Matt Adams: Well, looking back now after the passage of something like a decade which of those systemic failures have been most resistant to reform as you've allowed that data to sort of chart the course of an extended trajectory? And as you said, to add the new data onto it, I mean, what, what... Which of those things is the hardest to fix?

Brandon Garrett: I mean, there, there's so many things that haven't been fixed. Many more, like, it's very common now for policing agencies to videotape interrogations, so that way you can't have officers feeding facts to someone and deny that they did it.

But, you know, at, at The Wilson Center, we've, we're just finalizing a report that a wonderful lawyer here, Marcus Pollard, is finishing. And the 50 biggest policing agencies in the country that usually have the most up-to-date policies, the most best practices, even many of those agencies still don't videotape all interrogations and still don't have a lot of other protections during interrogations.

No particular policies for what if the person we're interrogating is, has behavioral health issues, they're disabled or…no policy on that. No policy on lying to suspects or using aggressive ploys. So we still have a long way to go, even though we have, like, half the states that require videotaping interrogations, often because of what they learned from these wrongful convictions, these DNA cases.

Things are somewhat better in the eyewitness area. Even more states have changed the way they do lineups to prevent, a lot of these DNA cases involved eyewitnesses that misidentified people, and for the most part, not, you know, very few were eyewitnesses that did it on purpose. They were, they were misled by suggestive lineups, and they picked someone who may have looked a little bit like the attacker, but their memory was altered by bad lineup procedures.

There's been a lot more progress on that front, but you still have small agencies that are doing lineups the same way. And there, technology is creating a new challenge because now so many agencies are using facial recognition to do searches, and you can really pick a face that looks a lot like the culprit's face but isn't if you have a big database.

Like you or I, like if, if you're searching 300 million photos on a database, like there's... The, the algorithm is designed to, and will pick, it, it'll find, you know, 20 or 50 faces if you ask it to find 20 or 50 faces that look almost exactly like ours. And maybe our faces are there, but the other 49 faces are not us and will look a lot like us.

The bigger the database, the more like near-miss images you'll have. So that's a big problem. That's a big technology problem. The area of jailhouse informants, not many states have done anything about it, it's a big problem. And the area of forensics, all kinds of challenges. There've been leading scientific organizations, including in big reports that came out after my book that have said we need to do something about unreliable forensics.

We need more guardrails. Federal rules of evidence were just amended to highlight the concerns with forensics when you have scientific expert testifying in court, but there are still lots of problems. Lots of problems around the world too. I mean, it's been exciting to see that Convicting the Innocent book has had traction in countries all around the world.

And and these problems are kind of the same everywhere, but with some differences. Like I was just in Taiwan and they've... my book was translated there. In East Asian countries, you can interrogate someone for days, and sometimes they'll allow it for like 21 days, 27 days, like weeks, and there's no Miranda.

You have no right to remain silent, and we can interrogate you for, for weeks. And so everyone confesses. And so false confessions, the false confessions chapter of my book translated in Taiwan has had an impact because so many of their cases involve confessions, and they've had many clients in Taiwan at their innocence project that, that falsely confessed.

And, in other countries, one of the challenges is that judges just don't meaningfully review new evidence of innocence. They just don't. They don't want to reopen convictions. They're just very reluctant to ever look at new evidence of innocence. And so that's, that's a global problem, and there's now a a network of lawyers that's trying to elevate the right to claim innocence because it's just not something that's well protected around the world.

the same as they were back in: 2011

Matt Adams: Pivoting then to another of your works, “Autopsy of a Crime Lab.”

Brandon Garrett: Yeah, that's the focus on forensics.

Matt Adams: You document... Yeah, Focusing on forensics, you document more than a half of DNA exonerees were convicted on the basis of fingerprints, bite mark analysis, blood splatter analysis, or other, what I would call, junk science forensic analysis. And, and your colleague, Chris Fabricant, came on this program to talk about that in the...particularly with respect to blood splatter and, and bite mark analysis. What would reforms to the forensic system look like if you were following the data with a captive audience of lawmakers who were willing to accept all of your recommendations?

Brandon Garrett: Yeah. I mean, one of the most important things that we've, that the scientific community's been talking about for a year is that, for years, is that crime labs should be scientific labs.

You know, I work at a university. At our university hospital, you wouldn't have, like, analysts just sort of, like, eyeballing someone's scan saying, "Yeah, I don't know, that kind of looks like cancer. Let's give them chemo." Like, you have to only use tests that have been validated, and there are requirements for all clinical labs to have staff. If there are some types of images that you really have to just look at, and it's not, like, a chemical test or a DNA test if you have to assess things visually, you have proficiency testing to test the reliability of the people doing that work. You don't just take their advice on it.

And if, and if you can use diagnostics and use statistics, you absolutely do, because you don't want to just be relying on judgment calls. I mean, sometimes they're just tough areas where a diagnosis is tough, and you have to rely on a doctor's judgment call. And there you, you know, you get a second opinion, and you use other diagnostics to try to inform that judgment call.

Crime labs are almost entirely unregulated in this country, and so they range widely in terms of do they have any quality control functions? Are they even accredited? Is it just sort of people who learn this work on the job and start looking at fingerprints, and now they look at fingerprints?

Or is it, no, it's a Ph.D. chemist that's operating drug testing equipment that is calibrated, and it's more like a scientific lab, which is good. I mean, there are more people with a science background in crime labs in this country. That's been progress, but they're still almost entirely unregulated.

Like, Texas has a forensic science commission, which has done things like say, we shouldn't be using bite mark evidence in Texas because there's no evidence that's reliable. But in every other state, it's entirely unregulated. It's not regulated like basic consumer products are regulated, and it's definitely not regulated the way that, that clinical labs, like any lab at a hospital, any strep test you get, that's regulated.

Crime labs using evidence to convict people in criminal cases, not regulated

Matt Adams: Crazy, crazy. Well, you argue on this topic, sticking with it for a moment, in forensic science and, and your work entitled “Autopsy of a Crime Lab,” you argue that judges must think, rethink their role as gatekeepers under the Daubert standard.

For our listeners who litigate these cases, what should defense counsel be doing at Daubert hearings to challenge the validity of forensic evidence?

Brandon Garrett: And so Rule 702, the Federal Rule on Expert Evidence, and, you know, almost all states follow the federal rules, even if the state hasn't yet adopted the new amendment to Rule 702, the advisory committee that amended Rule 702 two years ago said, like, "These are amendments. We're just restating what should have been clear to judges and to litigants, and yet it hasn't been. We're correcting their error, but we shouldn't have to amend this rule, so really, that amendment should apply no matter what." And what they said was, the burden is always on the party trying to introduce an expert to show by a preponderance that this expert meets all the reliability criteria of Rule 702 and of Daubert.

You need to show that this person is using reliable methods, that they're applying their methods reliably, that they're reaching conclusions reliably. You need to show all that if you want to put on your expert. We don't just presume experts are okay because, because they've been around or because the method has been around.

And so that, and they even highlight in the notes that this is a special problem in criminal cases where judges have often not asked basic questions about forensics, including, like how reliable is this method? What's the error rate? Do we even know? And if we don't know, well, what do we know about it?

There's also been some real progress. You know, I've digested a lot of case law in recent years where judges have started to ask these questions informed by some of these scientific reports, and all of a sudden you really are seeing much more detailed rulings on firearms evidence, in particular fingerprint evidence too.

There was just a big ruling out of New Jersey sending a case back for hearings on reliability of fingerprint comparison evidence. A lot of courts have sort of tried to kind of split the baby Solomon-style and say, "Well, we're not sure how reliable this method is. There are a lot of questions. It's kind of unknown how reliable this method is because there's been so few studies of it."

A lot of these disciplines, these traditional forensics, like these pattern forensics like fingerprints and firearms, there are very few studies that have ever been done. You know, DNA testing was developed statistically, and so we, we know what statistics are associated with certain types of DNA profiles.

Fingerprints and firearms, no statistics. We don't know how good people are at comparing these things. So some courts have said, "Well, since we don't know, we're not gonna throw it out, but we'll just limit what the expert can say." And so they, they just have to use more cautious language.

They can't say it's a match. They can't say it's a source identification. All they can say is, like, it's a, to a reasonable scientific certainty or a reasonable fingerprint certainty. Or more recently, courts have said, "Look, like, no, they can't say anything about how probable it is that this fingerprint came from someone.”

We don't have any statistics in this field. All they can say is like, “I can't exclude this firearm. I can't include it either though, I just can't exclude it.” And so courts have tried to compromise to say like, "We're just not gonna let this expert reach all these aggressive conclusions if we don't know what the research is to support them."

Matt Adams: But if, but if that's what they're saying before a jury, the jury is hearing it's a match

Brandon Garrett: That's so, we've done some studies on these, on the different forms of firearms language. And, and for the most part, if they use more cautious language, jurors still hear it's a match. If they say, "I can't exclude," so therefore they're not exactly matching, they're just saying, "I can't exclude," that actually did have some impact on mock jurors.

Still a fair number of them would convict. But that at least had some impact. And the other thing that we've seen in our studies, all the same studies where we're testing these due process questions, like which error do people care about more, convicting the innocent or acquitting the guilty?

Consistently in those studies, we see that people really do care about error rates. Like we, for better or for worse, we don't have a problem in this country that people just assume that experts should be trusted. Like people don't... Just because someone says, "I do science," or, "I'm a forensic scientist," people don't say, "Oh, well, you're an expert. We trust you." People do place a lot of weight on forensic evidence in criminal cases, but people do wanna know, like, "Well, okay, Mr. Expert, like how good are you really? Maybe there are experts out there, but are you a really good expert? Like, how reliable are you?" And people really wanna know, "Oh, you say you're using a scientific method? Well, how, how good is it?" I mean, we, we worry about like people being too paranoid about basic science and not understanding math or science, and, you know, as a lawyer that didn't major in anything scientific, I get that. I'm not great with numbers. I work with scientists. But people really do wanna know, like, well, what are the actual stats?

Like, how reliable is this method, and how reliable are you, Mr. Expert? How, how good are you at these things that you say that you're so proficient at? And so when people hear about error rate studies, to the extent that they've been done, or if they hear that they just haven't been done, like, really? No one has studied the error rate.

We don't really know how good you are at, doing blood spatter analysis because no one's studied that? Or there have only been two studies done on the reliability of fingerprint matching? Really? It's been used for 100 years, and, and there have only been two studies done? Jurors are really upset about that, and they wanna know.

And if we hear, okay, so you're... Maybe it's there can be very detailed information of fingerprints, and fingerprinting can be accurate under some circumstances. But how good are you? And if you say, "Well, actually, I don't, we don't really test proficiency in my lab. Like, I, I mean, I, I think I've always gotten it right. I'm confident that I'm always right." People are like, "Wait, wait a minute. No. But are you really an expert? Have you been tested?" And so jurors really wanna know about that. And so I've, I always encouraged lawyers to try to dig into those reliability issues and try to bring them out for the jury.

Matt Adams: Before we come full circle and go back to defending due process and this current moment that we're living in explaining part of the title, Why Fairness Matters in a Polarized World, I wanna use just another pivot to some of your other data-driven observations about some of the systemic issues in our criminal justice system and, and draw a little bit on “Too Big to Jail,” the corporate criminal accountability paradigm that you wrote about in that work, presenting a lot of detailed data showing a pattern in which prosecutors negotiate DPAs or Deferred Prosecution Agreements, demand admission of wrongdoing, impose penalties, and require restructuring that provides some reform to those corporations.

But those are usually somewhat vaguely defined, and-

Brandon Garrett: Yeah ...

Matt Adams: High-level employees tend to, what you call, get off scot-free. Has anything materially changed in the data since you published that book about the corporate criminal paradigm?

Brandon Garrett: No. I mean, I've done some updates since then, and for the most part, You know, there was a, there was a decline in corporate prosecutions during the first Trump administration.

They, although there were still some big cases with big fines. There was a, another sort of renewed focus on corporate prosecutions and white-collar cases during the Biden administration. We're still kind of early into the Trump two administration, but so far it looks like there are many fewer corporate prosecutions.

There was even a moratorium where the administration said for whatever it was, like four months, that we're not gonna bring foreign corrupt practices cases. Now there's no moratorium, but we're seeing fewer cases. It makes it a lot easier to update the registry when there are fewer cases, but we haven't seen like fundamental changes in approaches.

In general, they've said, "Look, we want to give leniency to companies that really cooperate and show that they want to fix these problems." Which can be a good approach if you really are actually making sure that companies fix these problems and there's accountability. I mean, I think there's a better model for this and I am practicing what I'm preaching, so I'm finishing my work.

We have like six months left as a monitor in a not corporate case. It's a civil rights case that I talked about in my due process book. It's a case involving misdemeanor bail reforms that were instituted about seven years ago in Harris County, Texas. And including because it's a civil rights case and the consent decree that they negotiated calls for this, the monitoring is long, like we, you know, it'll be seven years next March.

So they wanted a long period of oversight over this bail system which had constitutional violations that the judge found. We're not just sort of saying, you know, "Yep, it's all good, they're complying. End it." We have to issue detailed reports. All of our data is... There's a public dashboard that the Harris County makes available so people in the community can see what's going on.

Whereas in all of these corporate cases that were brought federally, there isn't really a judge involved. We have no idea what the company did when it was being monitored or whether it's fixed problems or not. The reports of any monitors, they're not public, and you just don't have a sense of when the oversight ends.

Like, are things better at this company or not? Was this worth it? Was this like a good use of government resources to oversee this company, or was it maybe it was an abuse of government resources and of due process? They forced the company to do these things, and they weren't real problems, or maybe the problems were really bad and they haven't been fixed.

We have no idea. It could be either. And so, it's been, it's been really interesting to be on the other end and be a monitor and to be in that role of trying to independently look at, you know, how is it going? But fortunately, in our monitorship, it's our... We're, we're filing public reports with the court, the county is itself making data public, and so everyone can see how it's been going, and they can see that these due process protections are doing a lot of good in the county.

Matt Adams: In episode 84 of this very program, your colleague at Duke Law, Samuel Buell, was on the show discussing Enron's legacy and his, his role i- in the Enron task force. In the post-Enron era, if you will agree with me that was perhaps the high watermark for corporate accountability do you believe that prosecutors have gotten better or worse at holding corporate actors individually accountable in response to the policy paradigms that were brought about by that Enron scandal?

Brandon Garrett: I mean, in general as a non-former prosecutor, maybe I was an outlier, but in my “Too Big to Jail” book and before I wrote the book in some articles and after the book in some articles and in testimony in Congress I continued to say that even if it's a very serious case, even if it's a major corporate crime case that involves billions of dollars we need to take due process seriously, and prosecutors should not be negotiating agreements out of court.

If criminal cases are resolved, they should be resolved in court publicly with the details public and with happens public. And that due process is also a, a matter... I mean, people may not think that fairness matters as much when it's a corporation as opposed to individual people, but there are a lot of individual employees of that company.

And it's just good government and good for the public to have due process there because as members of the public, if a company is paying fines, we should know, well, what did they do and is it being fixed? Was this abusive? Was the company fined too much with thin evidence, or was this not even enough?

Is this company being treated as too big to jail and no one's being held accountable? Which is it? If there isn't due process and there isn't an actual proceeding in court or at least more transparency and, and clear documents filed along the way in court, then we have no way of judging, and I think that's bad for the justice system.

I think due process may slow things down a little bit, but like with all these other areas, whether it's someone's driver's license being suspended or a multibillion-dollar corporate prosecution, due process serves the public interest.

Matt Adams: Well, let's channel all of your work back to where we started and have this full circle moment with “Defending Due Pprocess, Why Fairness Matters in a Polarized World.”

You argue that due process is under pressure from multiple directions, and a quote that struck me is that, quote, "In overheated debates, people argue that others do not deserve any presumption of innocence." What are the most dangerous fault lines in our system that you see right now, having dedicated a career to dissecting empirical data about those fault lines, dissecting empirical data about what it means to have an imperfect criminal justice system where these terrible things happen and happen more than once?

History repeats itself. Where are we at right now and what is the most dangerous systematic problem that you see in our polarized world.

Brandon Garrett: I'll just mention two things before we end the program. One is the new threats to due process that AI poses, and that's a theme in the book. And I, I think it's...

We all appreciate it much more now than even when the book came out because we're seeing AI get rolled into more aspects of our lives, and including government decision-making. And it's, you know, it's one thing if the government coaches an eyewitness to testify that I saw Garrett's face at the convenience store when it was robbed.

It was him. And you know, the only picture they showed the eyewitness was, was my face. It was suggestive. They kind of coached the eyewitness. It's another thing if it's an AI platform that says, "We identified Brandon Garrett at the convenience store." And well, how do you question it? How do you know whether it was coached?

How do you know how it even works? And we have more automated systems like that that have never been adequately tested, and they aren't designed to be tested. They're not designed to provide any information how they work, and that's a choice, and I talk about that. Like, it is a choice to use black box AI that can't explain itself, and that might be fine.

Like, maybe we don't care how it works if it's just like, giving us fun movie recommendations.

Matt Adams: Or a recipe that maybe is good or maybe it sucks.

Brandon Garrett: I don't know ... I feel very strongly. I don't want to waste those ingredients and make some, like, subpar carnitas. If I'm gonna spend hours and hours roasting the pork, I want it to be an excellent carnitas, and I want to be sure of that.

Anyway, so I don't know about the recipe example. I get really upset if it's a terrible AI slop recipe. Although I have a really funny example where my son just graduated high school. He's on his high school tennis team, and there was a coach that was like, "I'm gonna try something innovative. I have this really interesting tennis drill for you all." And it made no sense, and it was super complicated. And after about a half hour, the kids were like, "Can we just go back to some, like, playing some matches?" And the coach was like, "Sure." He's like, "Yeah, but what did you think of those drills? Weren't they great?" And then he confessed "I asked ChatGPT to come up with some tennis drills." Anyway, but still, like, using, using LLMs and using AI can, can be fun. It can be helpful. Very different if the government is using it to decide who to deport, whose driver's license to terminate, whose passport to cancel, who to put on a no-fly list, who to be audited by the IRS, whose grant to terminate, whatever it is.

We, and we've seen examples across all these areas of the government making really impactful decisions that cost people their jobs, their grant, their passport, their whatever it is, their immigration status, their Medicaid. And if you can't ask, "Well, why are you taking my passport? Why are you taking my Medicaid? Why am I being fired from the Forest Service?" And the government says, "Well, I don't know. The AI said thumbs up." That's not a good answer. Just like they can't just say, "Well, like, we're firing you from the Forest Service, we're taking your passport, we're seizing your car, we're auditing your taxes just because."

You can't just say, "But just because." That's not due process. You have to explain why. And so there's a real concern that these AIs being deployed by the government in ways that can't possibly satisfy due process. The due process challenges have already begun, and there have been some victories.

But we need due process guardrails when technology gets rolled out, and it's being rolled out fast and in lots of ways that some of which are really positive and help us out. But it's very different if it's our rights that are at stake.

Matt Adams: In a recent episode of this program, professor, we had a former FBI agent, a supervisory FBI agent, who dealt prom- predominantly with healthcare investigations, and I, I asked her that very question.

Now, how much of this is analytics driven, big data being culled to start criminal healthcare investigations? And the answer was really shocking. Something I knew already was that a large percentage of these investigations were formed that way, but it took human intelligence to get them over the finish line, was her working hypothesis.

Do you think that human nature being what it is, the easy way out sometimes being too alluring to pass up, the people tasked with that public trust, the people tasked with that sacred responsibility to investigate before they potentially try to take someone's liberty, are actually using AI in the right way?

Or do you take the more cynical view-

Brandon Garrett: Unfortunately, we know that some of these tools are just being used, including because people may not be trained or, or there are just no regulations or guardrails. But there have been false arrests around the country because of errors due to facial recognition.

And some departments say, "Oh, no. We, have a human in the loop. You know, we look for corroboration. We're not gonna arrest someone purely just on a facial recognition hit." But sometimes the corroboration is just, "I don't know, it looks like the guy to me. I saw the surveillance video. It looks like him."

And so we just don't have regulation, even of government use of AI, and vendors are coming up with interesting new systems all the time, and it's really tempting. Like, we could save time. You know, we can gain accuracy. Why not use these tools? Even something as basic as police reports, which are filed.

It's like one of the foundational documents in a criminal case. There are now systems that will transcribe and digest the audio from a body cam and turn that into a police report. And one of the defenses is, "Well, you know, but there's still a human in the loop. The police officer has to read that draft and has the opportunity to make any kind of edit, and then they sign off on it."

There have already been hilarious examples of officers signing off on police reports that had extended quotations from Disney's “Princess and the Frog,” because that was... the movie was playing in the background when the arrest was made or whatever. So, technology can be helpful. It can provide shortcuts, but shortcuts when people's rights are at stake are...

Due process is designed to prevent those kind of shortcuts.

Matt Adams: In our waning moments today, I wanna ask you a question that I ask all of our guests on this program, and that is what I refer to as our crystal ball question.

So Professor, dust off your crystal ball here, and if you had to project yourself 10, 30, 50 years into the future, what is the next generation of quantitative criminal justice scholars like yourself going to be tracking to ensure that the system is not corrupted by these influences, whether they be junk science, whether they be AI, what is it going to be that your successor will be looking at and saying, "This is something that we need to get out there so that we can undertake meaningful reforms"?

Brandon Garrett: Right now, criminal justice data is just really bad. And it's a problem with a lot of government databases. We have people whose due process rights are violated because their, their health benefits are cut off or their driver's license is suspended, sometimes because just, like, the wrong address is on file or just basic errors.

And so ideally, we'll actually have better data to study these problems, and the government itself will have better data and not make all kinds of terrible mistakes, not, you know, deny bail to someone because they have a prior arrest, but it turns out, no, that's a typo. That's a different person.

Problem is, is that, you know, the technology may get better, but to actually have good data, you actually have to care about its quality and invest in that. And so that, that's a real question. Are we actually gonna invest in due process and in accuracy over just bigger databases, bigger systems?

That's the question. There, there will be a lot of work for people like me if the technology gets better but the data and the accuracy doesn't. And unfortunately, I think that my students will have lots of great work to do after I'm long gone. I see this tension.

It, it goes back to these surveys about due process. People care about fairness, but they also care about outcomes and public safety, and those things are just always in battle with each other and intention. And so I don't see that, that going away anytime soon. So there'll always be a need for lawyers to fight for due process, and there'll always be a need for researchers to look at the data.

Matt Adams: Well, Professor Brandon Garrett of Duke University Law School, it's been a tremendous honor to speak with you about your quantitative analysis into the systemic failures within our criminal justice system. Hopefully, shedding some light on some of these issues will help bring us to a better place where we don't have to have databases of wrongful convictions and other things like that.

But until we do get to that better place, it's refreshing to understand that there are people like you out there, professor, who are holding onto this information so that we can use it in the trenches day over day. Thanks a lot for being with us on this episode of "The Presumption of Innocence." That's all the time we have.

We'll see you next time. Take care.

Share Episode

Shownotes

Transcripts

Follow

Video

More from YouTube