Annie Kuo: Hello, welcome to Discovery, a University of Washington School of Law podcast where we interview guests to the law school and our distinguished faculty. On this episode, we'd like to introduce you to Mary D. Fan, our Jack R. MacDonald endowed chair who just published an article, “Big Data Searches and the Future of Criminal Procedure” in the Texas Law Review. She's got another article forthcoming in the Minnesota Law Review called “Suspecting with Data.” Mary teaches criminal law, evidence and criminal procedure investigation at UW Law. She has a J.D. from Yale, a master's from Cambridge and is a Ph.D. candidate in epidemiology here at the University of Washington. She has been awarded the large section Professor of the Year twice at UW Law and been nominated for the University-wide Distinguished Professor of the Year. Welcome, Mary.
Mary Fan: Thanks, Annie. Glad to be here.
AK: So, Professor fan you begin your articles like you began your criminal procedure class with an unsolved mystery. In the big data searches article, you start with the unsolved murder of an immigrant family burned to death in their Denver home with a community terrified about potential hate crimes. How does this case illustrate, as you put it, both a great power and peril of data to solve crimes by unknown perpetrators?
MF: Well, Annie, as you powerfully described these heartbreaking facts, you know, this is a true story. It's a tragedy, and it starts with a stolen iPhone. And it ends in these arson murders that consumed five innocent lives, immigrant family members who are asleep in their home when vengeful arsonists burned it down. It's also a story about using data to identify criminal suspects. So, let's talk about two suspect identifications in this story.
First was a mistaken identification involving the use of the Find My iPhone feature, and it resulted in death by vigilantism. The perpetrator of these murders, along with his friends, he was angry because he was robbed of his iPhone and other things. And he wanted to find the people he believed robbed him and so he mistakenly thought he had located his iPhone in the home of the people he ultimately murdered. They were innocent.
The second set of big data-based identification strategies used geofencing and keyword searches. And these techniques actually caught the alleged vigilante murderers after the case had lingered unsolved for more than six months. These arson murders we’re growing cold. The community was terrified, especially the immigrant community was terrified. As you say, people didn't know if it was a hate crime. Ultimately, it wasn't a hate crime. It was in fact a story about a terrible, terrible mistake.
AK: And you say the police obtained a geofence warrant and then a keyword warrant to crack the cold arson murder case. How is a geofence or keyword warrant different than a regular warrant?
MF: Sure. Well, let's go back to Fourth amendment first principles. And let's talk about what a so-called regular warrant is. So, the Fourth Amendment to the U.S. Constitution requires that government actors, including the police get a warrant based on probable cause before you or I, or any of us, can be searched or seized unless an exception applies. Now, there are a lot of exceptions and complexities to the seemingly simple 54 words of this constitutional right. It's a whole course, in fact, a very full course that you can take if you come to the University of Washington. And we'll talk about these homicides and investigations, as well as many other cases and pragmatic questions about how to solve these cases, while still protecting our civil rights and liberties.
So, a search warrant. A warrant is in the text of the Fourth Amendment, and it's usually based on what we call a probable cause affidavit filled out by law enforcement. A law enforcement officers says here is the reason that we believe that there is a fair probability — what's probable cause? Just a fair probability. A flexible, low standard — there's probable cause to believe that evidence will be found in the place that we're asking that you allow us to search. That's for a search warrant. There's also arrest warrants. And an arrest warrant is to arrest someone as opposed to search a place. An arrest warrant is based on probable cause to believe that someone committed a crime.
What's really confounding the courts about keyword warrants and geofence warrants are these techniques are used to identify unknown perpetrators, usually. So, courts are very familiar with arrest warrants. Let’s say there's probable cause to believe this individual committed a crime. So, let us, give us this warrant to arrest. Even with respect to search warrants.
Now search warrants, there's probable cause to believe there's evidence of a crime in that place. But even if it's just — that's just the burden of the officer in getting a search warrant, but in the story, that probable cause affidavit, oftentimes will get a sense of who the discreet perpetrator is. So, I'm searching the suspect's mom's place, because I think the suspect hid the murder weapon in his mom's house. But in my story, you are getting a sense of who the suspect is. In contrast, geofencing and keyword warrants, we’re saying, we know this crime happened. We know it happened here. And we know it happened at this time in this place. We don't know who the perpetrator is. That's why we're asking you for this so-called geofence or keyword warrant.
What's a keyword warrant? A keyword warrant draws on keyword search data. So, these big tech companies — Google, primarily, the major market player in this area — keeps records of lots of details about the things that we search, and those records are connected to other records. So, not just the things that we search, but when we log on, our Gmail account, and so forth. They have a lot of data. These big tech companies have a lot of data on us. And so, a keyword warrant is saying, you know what, let's make it concrete. In the murder case, the actual keyword warrant in that case, the police said, how do the arsonists know where to go to find the victim's home? Let's see who googled the address shortly before the murders. And that's in fact what led them to the perpetrators. Because you're not going to Google your own home. And if you're just a private person, who else is going to Google your address, right? So, that's an example of a keyword warrant.
Geofences, you know, are pretty familiar to you. If you ever logged on to the internet, or you're just surfing the web, or you're on Facebook, or whatever social media account you use, we often find targeted advertising. And you may find advertising about businesses that are near you. So, private companies use geofencing all the time T They see where you are, and they target products and services to you based on lots of things, your preferences, but also based on where you are. For law enforcement purposes, a geofence is “I know this house burned down at 2 a.m. Who was there at 2 a.m.?”
These days, cell phones are just a ubiquitous part of our bodies, like the Supreme Court have said it's almost like another bodily appendage to us. And our cell phones leave traces, data traces, because of all the geolocation services that we get on our cell phones and our apps, etc. And so, what a geofence warrant does is it goes to, let's say, let's just let's just make the company Google, because a lot of most of these requests are to Google. It says to Google, tell us who was in this neighborhood at 2 a.m. in the morning. It's great if you have a crime that occurred in a place that there wouldn't be other people except the perpetrator and the victim who you can roll it out, right? Let's say you have a series of burglaries or robberies or assaults, whatever, with similar modus operandi. So, who was at place A place B, and place C? You can really weed out a lot of people and really focus in on the potential perpetrators. Not great if you're saying who's in the U-District at 7 p.m. on Friday night? There are a lot of people beyond a potential suspect in, let's say, an assault.
AK: Has the U.S. Supreme Court ruled on the constitutionality of geofence or keyword warrants?
MF: No. We rarely give short answers a lot. But no.
AK: And your article delves into the confusion in the lower courts over these kinds of warrants, these digital search strategies for solving crimes. Can we talk first about geofence warrants? How have lower courts split over whether the police can use a geofence warrant to determine who is present at the time and place of a crime?
MF: Courts are very badly split as to whether or not geofence warrants are a warrant, as contemplated by the Fourth Amendment, or — and this calls for more history — this hated general warrant which is what prompted the Fourth Amendment. This general warrant historically was a license to rove, and if you had this license, you could go into any person's home to search for uncustomed goods. And this is actually what made the founders really upset, these general warrants and writs of assistance in the colonies that the crown used. And so, opponents of geofence warrants say it's not a warrant like the Fourth Amendment requires. It's actually like the bad writs of assistance, or general warrants, that prompted the framers to require a warrant based on probable cause particularizing the place to be searched, or the person or thing to be seized. So, that is a lot of the debate in the lower courts.
There's also a question about whether or not these keyword warrants or geofence warrants are a search at all because a search, within the meaning of the Fourth Amendment, is not a search the way you and I think of it, when you say, well, I've been searched. A search is a term of art for purposes of our Fourth Amendment rights. It is an intrusion on our reasonable expectation of privacy, or a trespass for the purpose of gathering information, a trespass on our property. For our purposes, let's focus on the intrusion on the reasonable expectation of privacy because there is no trespass. Just trust me, I don’t want to break it down.
So, with respect to reasonable expectations of privacy, the Supreme Court has in a number of cases held that you have no reasonable expectation of privacy against the police in information you share with any third party. So, for example, with respect to our financial information. Well, you shared it with the bank, the bank knows it, you may want it to be private, but you have no Fourth Amendment reasonable expectation of privacy, the Supreme Court has said. Now, Congress can and has given you some protections because they were mad when the Supreme Court said that you have no reasonable expectation of privacy. But for Fourth Amendment purposes, there's this thing called the Third-Party Exposure Doctrine, which means that if you share it with some third party, then you don't have privacy in it. Well, guess what, all our Google data, who has it? Google. As well as other social media companies and all the other entities that our data gets resold to potentially as well.
So, there's a question as to whether or not we have a reasonable expectation of privacy, and the Fourth Amendment applies in the first place. There may not even be a need to get a warrant. Now, courts have just jumped over this question. And this shows the power, frankly, of big tech companies. Why have they jumped over this question? Because Google says we will not give it to you, and we will fight you in court unless you get a warrant. And that's why law enforcement officers have been turning to these warrants. They say, well, fine, we'll get this warrant, we'll get a magistrate judge to sign off on it. And we'll give it to you. So, you can tell your customers we made you, not that you were just giving out customer data willy-nilly and making your customers mad at you. So, it's cover for Google. Right? And all these companies that have this data.
So, there's a question about whether or not you have fourth member protections. There's a question about whether these warrants are really warrants within the meaning of the Fourth Amendment. And then there's also a question of sometimes some courts say if it's if the warrant is tightly defined, then they'll find probable cause and particularization as required by the Fourth Amendment. Other courts might say, if you did a better job of defining the geofence parameters, so it's not capturing a lot of potentially innocent parties, then we might find it satisfactory for Fourth Amendment purposes. But you didn't do a good enough job here. So, you see courts, splitting all over the place. It would be tempting, I wish I could, but I can't give you a one-size fits all rule. Like the more particular you are, the less people you affect with your request, then the more likely it is to be upheld. But frankly, the splits are kind of chaotic.
So, I'll give you an example. Remember, the January 6 riots on the U.S. Capitol? To figure out who the perpetrators of those assaults and trespasses were during the attempt to disrupt the certification of the U.S. presidential election, well, there's a lot of people there. And so, to unmask them — a lot of them had cell phones, they were taking selfies — to unmask them, law enforcement used a geofence warrant to see who was on the grounds of the Capitol at the time of the riot, because they were trespassing, they're not allowed there. And so, there's like a three-step process. I'm not going to go into all the nitty gritty of it, but multiple district judges have upheld a process that yielded 5,518 anonymized devices. That's what we call step one. There's this multi-step process. As step one, Google will say, here are these anonymous devices that were in this geofenced area. Now, I'm not going to give you the identity, the names of these folks. You, government, come back and tell us of all these devices, who really do you think are the potential perpetrators? So, we'll just narrow it down. So, at step one, with respect to the January 6 investigation, there were 5,518 anonymized devices. By step three, when there was disclosure of the identifiers for the suspects, it was 1,535 user accounts. And there was a, it was a pretty complex process. I'm not going to go into that. And that's been upheld.
In contrast, a court rejected, rule that was invalid, to disclose just three devices, one of which was the perpetrator. So, do you see I can't tell you, it's not just a one-size fits all based on the impact by numbers. Courts have just varied. It depends on the judge, frankly, and how they construe the Fourth Amendment.
AK: How have courts addressed keyword warrants if at all?
MF: So, there are very few decisions, or even law review articles on keyword warrants. It’s very new. There's a reason for this. A lot of keyword warrants are secret, they're filed under seal. You can think about why that might be. These keyword warrants reveal very intimate details, potentially, even if it's just the victim's address, which is sensitive, right? You've already been victimized. And then do you really want this out in the public where you live, for example. Especially because law enforcement investigate a number of sensitive issues such as kidnapping or sexual assaults and you don't want the release of, for example, the victims address in these cases.
AK: So, there's like no scholarship, barely any on keyword warrants because a lot of the cases that involve it have been sealed, or it's just so new.
MF: I'm very proud to say that one of the very first, or perhaps the first piece on keyword warrants was my student, written in criminal procedure. It started in a criminal procedure investigation class as an essay developed into an award-winning comment.
AK: We should give her a shout out.
MF: So, shout out to Chelsa Edano. Yeah, so one of the very first was my student developed class. And then there's, of course, my article. There's a growing number of cases and literature on geofence warrants, smaller handful, a much smaller handful of cases, much less scholarship, for now. But the thing about technology is once people start sniffing it out, there's going to be an explosion. This stuff is so interesting.
AK: Yeah.
MF: It's also so useful to law enforcement.
AK: So, let's talk to you about this, let's pivot to this other interesting dynamic, originalist romantic Luddism and Fourth Amendment interpretation. In your Texas Law Review piece, you eliminate the ridiculous, or the ridiculousness, or I think I saw it written as like the comic anachronism.
MF: Yeah, it's, it's kind of, it is ridiculous. So, we talk about this romantic Luddism, a combination of fear of technology and nostalgia for these simpler times, these times past when we had more privacy before the technology sort of bore down and eroded the space of personal privacy. And I don't mean it as a pejorative. We often, sort of, think it's a pejorative to call someone a Luddite. But Luddism actually arose from an antitechnology stance of English textile workers who wanted to impede industry efforts to innovate by smashing new machines and staging uprisings. And it was a revolt, it was a rising up against technology, logical innovation, crowding out humans, human craftsmanship, human labor. So, there's actually deep and interesting roots to this view, and let's be real, you, I, a lot of us have a little bit of a romantic Luddite in us. Right? We don't want to be completely dependent on technology.
I use this lens as a critique on originalism, which is so dominant these days. And what's originalism? Originalism is a technique by which we try to interpret how the Constitution applies to new questions by looking at the original meaning way back when the Constitution was framed. What did the framers have in mind? It is ridiculous to interpret the Constitution when it comes to cutting edge questions like keyword searches, or geofences through the perspective of a time when we didn't even have electrical power.
AK: Mary, you also talk about an empathy deficit in the law on how we treat the risk of affecting the rights of innocent third parties in Fourth Amendment law. What do you mean by that?
MF: There's an empathy deficit throughout many areas of law and that's one thing we'll learn when we take a critical perspective in law school. The empathy deficit is particularly severe when it comes to criminal law and procedure. And then with respect to what I'm talking about, specifically affecting the rights of innocent third parties in Fourth Amendment law. Well, the Supreme Court has said that, you know, innocent parties may be searched. And that's just something that could happen because probable cause is a low and flexible standard. Innocent parties may lawfully be detained. Let's say you're living in a group house, your roommate — nothing to do with you — committed a crime. You just are living in that house. Well, when the police execute a search warrant on that house, the police have the power automatically to detain everybody on site. Moreover, if they have a reason to believe that you pose a danger while you're being detained for let's say two hours or three hours for the execution of that search warrant, you may be cuffed if there is a reasonable basis to believe that you might pose a risk, even if police know that you're entirely innocent.
That's what I mean by collateral impact and a collateral impact could be collateral harms. And I talk about a case where a family who couldn't afford to live in their own house had to share a house had nothing to do with the person who was being searched, but they were detained. And while they were being detained a five-foot two woman was asked about her immigration status that led to consequences for her. So, these can be pretty severe harms as well for innocent folks who police know have nothing to do with the warrant they're executing. And the Supreme Court in their doctrine has said, this is just what happens. It's a cost we're willing to tolerate. And yet, you get all this conflict and consternation in the courts over heaven forbid the government might find out I was in the area of a bank in a busy downtown area at 2 p.m. at the time of a bank robbery — me and 200 other people.
I mean, okay. Your device — moreover, at step one your device is anonymized too. So, we seem to be so concerned about the impact of just even anonymized devices being released to law enforcement to really no pragmatic effect on the people, most of the people. And yet the doctrine is explicitly accepting and permits real harms. And so, I think there's a real empathy deficit. I think that there is a failure of imagination and understanding and empathy when it comes to impact on the people who have the least power to resist. And there is a as a huge surfeit of concern and outrage over data sometimes.
AK: Early in your response, you mentioned probable cause. In your paper, you argue for what you term, quote, technological probable cause. So, what is that? And how might that matter for the constitutionality of geofence and keyword warrants?
MF: What's confusing the courts is there's no named perpetrator in these keyword warrants and geofence warrants. Frankly, there's confusion at multiple levels, because frankly, a search warrant can be on an innocent third parties home too, you don't have to just execute a search warrant on a named suspects home. So, that's one level of confusion. But with respect to technological probable cause, what am I saying there? I'm saying that one can have probable cause predicated on advances in science and technology that give us a reasonable basis, a fair probability, to believe that there will be evidence in the place to be searched. And I use an analogy to what we call John Doe warrants. So, some John Doe warrants are, let's say, a homicide or a sexual assault, something with DNA evidence left behind, is unsolved. We don't know who the perpetrator is. These are used particularly in sexual assault cases because there's a statute of limitations there. To be able to preserve the ability to bring the perpetrator to justice after the statute of limitations may have run, what prosecutors will do is they will file what they call a John Doe warrant. We don't know their identity yet. But we have, very precisely, we have the DNA profile. So, we will append to our arrest, John Doe warrant, the very precise DNA identifier of this individual, and that helps preserve the ability to bring that individual when found to justice. So, that's an example of how science and technology can give you alternative ways beyond a name, a known individual’s name, to identify someone, and also to get probable cause.
AK: Fascinating. How would you protect against police roving through our data on a hunt for people who have sought abortions, for example?
MF: Now, this is a real and important concern in these changing times when a growing number of states are criminalizing what used to be a fundamental constitutional right. So, opponents of techniques like geofence and keyword warrants might say, Professor Fan, you start with this example of an arson murder — I start with it because that was one of the key cases in the canon so far, the very short canon of cases — but what about misuse and abuse of these big data searches to hunt for people who are trying just to exercise what used to be a fundamental constitutional right, aren't you concerned? And yes, I am very concerned. I think there absolutely needs to be safeguards. And I propose a number of safeguards. I'm not going to detail all of them here. But for one thing, there's a reason why I'm focusing on big crimes. I don't think these techniques should be used in roving searches for unknown perpetrators. For there to be technological probable cause there has to be an existing discrete crime that we know was committed by someone on this date and time. Do you see that's a very different posture than hunting for suspects. So, that's one protection because it simply falls outside of what even a more sophisticated understanding of probable cause would consider to be sufficient, first of all. Second of all, I think these techniques, they do impose risks. Opponents have good points, and one of the safeguards should be a limitation, not just on the fact that there's a discrete crime, but that there's a discrete serious crime. Serious crime that involves, for example, homicides, sexual assaults, serious bodily injury.
AK: Thanks for being here on the podcast.
MF: It's a pleasure Annie. Thanks so much.
AK: Professor Mary D. Fan is the Jack R. MacDonald Endowed Chair at the University of Washington School of Law. Her scholarship has been cited by judges including U.S. Supreme Court Justice Sonia Sotomayor and in major media venues.
RSS feed
Apple iTunes
Spotify
Amazon Music