The Pattern Problem

Mar 29, 2018
Originally published on March 30, 2018 8:48 am

Content Warning: This episode contains descriptions of sexual abuse.

A panel of judges sits to decide the fate of the young woman. She's the child of addicts and an ex-addict and ex-felon herself, and she's asking the court to trust her to become an attorney. The outcome of her case hinges on a question we all struggle with: are we destined to repeat our patterns, or do we generally stray in surprising directions? - a question increasingly relevant in an age when algorithms are trying to predict everything about our behavior.

Special thanks to the following musicians:

Rick Klaras for his song "Rain"

Aimee Mann for her song "Stuck in the Past"

Copyright 2018 NPR. To see more, visit http://www.npr.org/.

ALIX SPIEGEL, HOST:

It's natural to look for patterns. There's a certain safety in knowing that the man was a murderer because his mother violently abused him. Children who were violently abused don't seem to do that well. That's a pattern. And if there's a pattern, then there's something we can do about it - not violently abuse our children. Problem solved.

(SOUNDBITE OF ARCHIVED RECORDING)

UNIDENTIFIED PERSON #1: All rise. Supreme Court of the state of Washington is now in session.

SPIEGEL: On November 16, 2017, the Washington state Supreme Court was called to session, as a tall lawyer with a choirboy face named Shon Hopwood sat sweating at the defense table. Behind him was his client, a responsible-looking woman named Tarra Simmons who had graduated near the top of her law school class and was now trying to become a lawyer.

The question in front of the court was, would this version of Tarra, the responsible citizen Tarra, last? Or would that Tarra be undone by the other Tarra, the one who had demonstrated a clear pattern of criminal behavior? Criminal-behavior Tarra did have a long history of undermining responsible-citizen Tarra. So which would appear in the future, and could you tell by looking at the patterns in her past?

That was the question for the court to decide. But also, in a way, it's a question for all of us listening to decide, who would Tarra be in the future?

(SOUNDBITE OF ARCHIVED RECORDING)

UNIDENTIFIED PERSON #2: Shon Hopwood will have 20 minutes for his opening and 10 for response.

SPIEGEL: Now, I got to say there were lots of things that were unusual about this case - really, so many things - the curious role that Shon played in the life of Tarra Simmons, how Tarra ended up in front of the Supreme Court in the first place. But for my money, probably the strangest was this little tidbit. Shon Hopwood, the lawyer arguing the case, was an armed bank robber.

(SOUNDBITE OF ARCHIVED RECORDING)

SHON HOPWOOD: Good morning, and may it please the court. This case presents the court with the core question of how long must we...

SPIEGEL: This is INVISIBILIA. I'm Alix Spiegel.

HANNA ROSIN, HOST:

And I'm Hanna Rosin. INVISIBILIA is a show about all the invisible things that shape human behavior, our beliefs, our emotions, our expectations. And today, we're looking at patterns.

SPIEGEL: Because we are always trying to find patterns, quietly but compulsively searching for them. We meet an attractive man, then find out that he cheated on his ex. And instantly the possibility of a pattern haunts us. Would he cheat on us, too? Is that his pattern?

We go to the doctor. He asks about the family cancer history. He's looking for a pattern. We hope he doesn't find one in us. Every time we go online, to Amazon, Google, companies are tracking our patterns, using them to guess what we'll do next.

ROSIN: So the thing we're trying to figure is, how much does pattern predict in an individual life, in a living, breathing person, like Tarra Simmons, but also in ourselves?

SPIEGEL: I tagged along as these Princeton scientists staged a massive pattern-finding competition that harnessed the power of machine learning. It involved eating insane amounts of sugar. So clearly, the conclusions they came to will be revelatory.

ROSIN: Stick around.

(SOUNDBITE OF MUSIC)

SPIEGEL: There's a lot to say about the complicated life of our main character, Tarra Simmons. But I want to begin with her lawyer, Shon Hopwood, because it was Shon's strange path that made Tarra believe that she could break free of her old patterns. You see, Shon's life really didn't conform to the normal pattern. Like Tarra's, it was full of wild detours.

HOPWOOD: I don't have a great excuse as to why I did these things. And everybody always wants that because - I don't know - it, like, closes a circle for people. That's not really how it happened.

(SOUNDBITE OF MUSIC)

SPIEGEL: To the naked eye, it looked like Shon Hopwood was born into a really good pattern. He grew up in the neighborly, low-crime community of David City, Neb...

HOPWOOD: A sprawling metropolis of 2,500 people.

SPIEGEL: ...To a great Christian family that tried to steer him away from dark forces that might corrupt...

HOPWOOD: I can tell you my mom confiscated many Ozzy Osbourne and Motley Crue cassettes.

SPIEGEL: ...Which is not to say that Shon's parents oppressively monitored their children. They didn't.

HOPWOOD: Me and my brothers would roam the neighborhood. It was a good childhood.

SPIEGEL: Fresh air, safe community, loving family - pretty good patterns. But for some reason, in college, Shon started veering off the graph. He wasn't that interested in school, so dropped out and returned to David City to work, spend time with friends and family. And that was all going fine, really nothing out of the ordinary.

HOPWOOD: Until my friend called me one day and said, hey, come down to the neighborhood local bar. I want to talk to you about something.

SPIEGEL: Shon's friend, Tom (ph), like Shon was born into a middle-class family, but wasn't exactly thriving. He was studying criminal justice at college but felt depressed and so had come up with a novel plan to shake himself out of his rut.

HOPWOOD: And he just asked me - he said, what do you think about robbing a bank? And I don't know where the idea came from. And, you know, most people would have said, no, or that's - what are you talking about, or walked away or a million other responses. And my response was, yes, this is a great idea.

SPIEGEL: Now, to be clear, it wasn't like Shon didn't have second thoughts. He had several, had them right up to the moment when he walked into the bank dressed as a handyman with a toolbox on his hip.

HOPWOOD: I'm wearing coveralls and a hard hat and have a toolbox. And I walk in the bank, and I pull a mask up. And I drop the toolbox on the ground. It makes a huge noise. Everyone turns and looks at me, and I unzip my coveralls, pull out a .22 rifle and yell, everyone get down; this is a robbery.

SPIEGEL: After that, Shon recruited a small group of friends and just hit one bank after another. It was all going great until one night, while living it up at the DoubleTree in Omaha, when four guys from the FBI tackled him. He was sentenced to 12 years in prison, which was bad enough.

But what really cut was that a bunch of people in his hometown disowned Shon's completely blameless parents because the apple doesn't fall far from the tree, right?

HOPWOOD: They have to have some reason for why I did these things because otherwise it just doesn't compute.

SPIEGEL: They need some pattern in order to make them feel stable.

HOPWOOD: Yeah, to make - to make sense of it all because it didn't make any sense that me and the people that were involved with me had robbed these banks. I mean, one of my codefendants was the son of the town attorney. And so nobody could understand why this happened.

(SOUNDBITE OF MUSIC)

SPIEGEL: We need to find a pattern. And when it eludes us, we ache for it. Shon was worried that in the eyes of the world, his pattern was now set.

But in prison, this kind of random miracle happened. He was working in the prison law library checking out books when a friend asked for help with his case. The guy wanted to try a long-shot petition to the Supreme Court.

Shon had never studied law and only had a high school education, but he wanted to be helpful. So he agreed, spent two months working on an argument and then sent it off and basically forgot about it.

HOPWOOD: Then one day, I'm walking out to the recreation yard at 6:30. I always went and lifted weights at 6:30 in the morning. And a friend of mine comes running and screaming out of the housing unit. And this being federal prison, my first thought is, what did I say to this guy yesterday that he wants to come fight me at 6:30 in the morning?

SPIEGEL: The Supreme Court had accepted his appeal.

(SOUNDBITE OF MUSIC)

SPIEGEL: That highly unusual lottery ticket win led to other unusual turns in trajectory until one fine day, Shon found himself moving a box of belongings into a small office at Georgetown Law. He'd become a law professor, proof the past was not necessarily prologue.

He laid out that lesson in an autobiography called "Law Man" and then embarked on a long uphill battle. He would help other criminals convince the world that the past was not necessarily determinative. It could be peeled away, left behind like a set of clothes that left no lasting mark on the body.

That was Shon's belief and probably why he was so open to Tarra Simmons, a small woman with pretty blue eyes who was born into the kind of patterns that no one wants and who had tried over and over to break free of them.

TARRA SIMMONS: My parents were both addicts all their lives. My mom, she was a victim of childhood sexual abuse. And ever since she was, like, 15, she lived on her own.

SPIEGEL: Tarra Simmons, like her parents, grew up in an environment so dismal and harsh, it would make any judge charged with predicting her future worry. I mean, how could anyone escape that? Abuse, crime, drugs - that was the life of the first Tarra.

To give you some sense, the one bright spot in her life was her father, who she says was a crack addict and who never seemed to have a job but was always generous. She says he was the kind of guy who would literally give you his last $10 and still remembers how he once bought her a stuffed monkey.

SIMMONS: I mean, there were tough times, too. You know, he would be cooking crack in the kitchen, and there was all kinds of gang members. And so there were situations and places that - where I was abused, but I didn't blame my father for that.

SPIEGEL: Tarra was on the streets of her hometown, Bremerton, Wash., by 13, where there was more abuse. She says she was even abducted by a man in Seattle who briefly forced her into prostitution there.

SIMMONS: He would put me out on a street and then make me give him the money. And it took about three or four weeks, and then finally one of the men that picked me up, he made me perform oral sex on him. But then he dropped me off at the ferry and told me to run. And I almost think that he was an undercover cop. I really do to this day. Yeah.

SPIEGEL: So Tarra was the victim of terrible crimes, but also, at moments, a perpetrator of crime. As a teen, she was arrested for shoplifting, she hung hard with gangs, got involved in physical fights - the kind of stuff that sometimes blossoms into more serious criminal behavior. But, for a while anyway, that's not the path Tarra's life took, really because of one person, Davon.

SIMMONS: He was the best baby ever. He's really been the best kid. He has the most gentle soul of anyone you'll ever meet.

SPIEGEL: When she was just 15, Tarra gave birth to her first son, Davon. And she says that her little boy made her want to live a better life so that he could live a better life. So Tarra started making grand plans.

SIMMONS: I remember getting the little Gerber flyers in the mail saying, you know, put some money away for your child for college. And I think I even signed up for it and would put like $5 a month away for his college because I just knew he was going to college, even though I hadn't graduated yet.

SPIEGEL: But she was going to an alternative high school, taking the heaviest course load she could, but working so hard she still got really good grades.

SIMMONS: Trying to show him that it didn't matter where we came from, that we could overcome obstacles and still do whatever we wanted.

SPIEGEL: After graduating, Tarra found a government program that allowed her to go to college. And within six years, she had a nursing degree and a job in the ER. By 22, she bought her first house, a two-story home on a cul de sac.

(SOUNDBITE OF HOME VIDEOS)

UNIDENTIFIED CHILD: Hello. Hey.

SPIEGEL: If you look at home videos from this time in her life, you see kids circling on bikes playing with each other. This scene feels a million miles away from the first Tarra, from child prostitution and crack cocaine. And if we were to stop the clock here and ask you to predict Tarra's future, I'd bet, in surveying the scene, you would feel that glow of hope that comes from watching a person escape a terrible past and conclude that the story would continue that way. It looked so lovely.

(SOUNDBITE OF HOME VIDEO)

SIMMONS: I love you.

UNIDENTIFIED CHILD: I love you.

SIMMONS: I really felt like I was on a good path. My life was going to be good now.

SPIEGEL: It was clear sailing?

SIMMONS: Yeah, that's what I thought.

(SOUNDBITE OF HOME VIDEO)

UNIDENTIFIED CHILD: (Laughter).

SPIEGEL: And it's true that for a while, Tarra did maintain her new life. She got married and worked as a nurse, took Davon on trips to Disney World. But then came stress, and with stress came old patterns.

It began when Tarra started getting into fights with her husband. They eventually separated, so Tarra started dating again. And for fun one night, she went out with a man in town. But the date turned into a nightmare. Tarra says she was raped.

SIMMONS: And it was really traumatic for me.

SPIEGEL: Tarra says she called a rape line and went to the hospital.

SIMMONS: I mean, I was bleeding, and I wanted to be evaluated.

SPIEGEL: But the experience activated some old part of her. She was angry, wanted the man to suffer, feel the same pain that she was feeling. So she talked to this kid she knew.

SIMMONS: You know, he had a friend that was, like, in a gang. And so they kept telling me, like, get him somewhere, and we'll show him. Anyway, so I called the guy up, and he came over to my house. And they violently assaulted him.

SPIEGEL: The teens beat the man with a baseball bat and beer bottles.

SIMMONS: And so I was charged with conspiracy to commit assault.

SPIEGEL: Tarra was sentenced to eight months in jail. But that was just the beginning of her problems. When she got out, she had another child, Dominic (ph), and briefly re-established stability. But she had trouble getting work as a nurse because she was a felon and to cope, started using meth, doing more and more until finally she was dancing the exact pattern of steps that she'd learned in her childhood, of addiction and lies and eventually crime.

Tarra left her family and moved in with a new boyfriend. She says they sold drugs together out of his apartment. Sometimes Davon and Dominic would come over and watch television, sitting quietly as meth heads milled around them.

SIMMONS: Yeah it was - it was bad.

SPIEGEL: Tarra was eventually arrested and got 2 1/2 years in prison, and she says her time away from her kids took a big toll on them. But looking back, she says, she's not sorry she went to prison.

SIMMONS: It's almost a good thing that that happened.

SPIEGEL: Because it was during this stint in prison that Tarra decided that this time, she was going to break free for real. She started therapy, was taught how to recognize and interrupt the bad thoughts and behaviors she learned as a kid. She still yearned to be a nurse but knew it was next to impossible. So once she got out, she started searching for a new career. She talked to everyone, constantly asking for ideas until finally she happened on one.

SIMMONS: Somebody said, you know, you could be a lawyer. You - there is a guy named Shon Hopwood who, you know, robbed five banks and is a lawyer. And I was like, wow, really? And she said, yeah, he has a book called the "Law Man."

(Reading) Chapter 1. A wild storm was building over Oklahoma City, our final destination.

I read it all over the course of, like, two days because I was so excited. I couldn't put it down once I started reading it.

SPIEGEL: Here was someone like Tarra, a felon incarcerated for the serious crime of armed robbery. And yet he'd found a way to prosper.

SIMMONS: He didn't let his past define him, and he didn't let the conviction be the end of his story. That's when I really started to believe that my life could change, too. I thought if he can do it, why can't I? And I will give 100 percent to make it happen.

(SOUNDBITE OF MUSIC)

SPIEGEL: Tarra decided to model her life on Shon, become a lawyer just like him. She reached out on Facebook, and they became friends. And truly, through law school, Tarra worked insanely hard to prove to everyone that her old patterns were dead and gone. In addition to her schoolwork, she volunteered and networked continuously.

SIMMONS: Presenting at judicial conferences, starting the nonprofit Civil Survival. The governor appointed me to two different boards, co-chaired the Statewide Reentry Council, graduated magna cum laude, and I got the Skadden Fellowship.

SPIEGEL: But then came the big day. See, before Tarra could take the bar, she needed to appear in front of the Character and Fitness Board for the Washington State Bar, a panel of lawyers who would decide whether Tarra was a safe bet. Tarra was absolutely convinced that she would succeed, that she'd be seen as someone who had finally escaped a terrible history.

SIMMONS: I did a Facebook post, and I was like, today is going to be an amazing day. Today is a day about redemption.

SPIEGEL: But during the actual hearing, when witnesses came to testify on her behalf, she noticed the committee seemed really skeptical.

SIMMONS: How long have they known me? They've only known me two or three years. How can they be so sure?

SPIEGEL: After five hours of testimony, the character and fitness committee adjourned to deliberate. Tarra could hear them arguing behind the door.

SIMMONS: And the chair of the board came out. She was very gentle in breaking the news to me and said, I'm very sorry.

(SOUNDBITE OF MUSIC)

SPIEGEL: So what would you do?

Would you clear a lawyer who, since adolescence, has had trouble on and off with the law? If Tarra reverted to her old ways and failed a vulnerable client in a high-stakes case, wouldn't the public rightly ask how the profession could let someone with that pattern of behavior become a lawyer? Do you think Tarra will leave her old patterns behind, or will they catch up to her again? What's your prediction?

Tarra firmly believed that she wouldn't revert, and Shon agreed. So they decided to appeal. Shon would argue in front of the state Supreme Court that if you see people only as their past patterns, you condemn them to become the patterns you see, that believing too much in patterns was a bad idea - for Tarra, sure, but also for the rest of us.

(SOUNDBITE OF ARCHIVED RECORDING)

UNIDENTIFIED PERSON #3: The Supreme Court of the state of Washington is now in session.

(SOUNDBITE OF MUSIC)

ROSIN: INVISIBILIA will be right back with an innovative scientific study on the predictive power of patterns.

(SOUNDBITE OF MUSIC)

ROSIN: This is INVISIBILIA. I'm Hanna Rosin. When we left off, Tarra Simmons was standing in front of a panel of judges. And their job was to predict her future, to look through her past - where she'd come from and what she'd already done - and see if she would repeat those patterns.

Now, as it so happens, two Princeton professors were also obsessing about patterns. They were organizing this insanely grand experiment to see just how good patterns were at predicting an individual life. Alix continues the story.

SPIEGEL: Because we live in the age of computers, we live in the age of patterns. Computers are genius at patterns. In the blink of an eye, they can scan more data than you or I could sift in a lifetime. And in that data, they see things we could never see - beautiful things and terrible things and even, we're told, the future.

MATTHEW SALGANIK: Yeah, pretty much, with enough data, everything becomes predictable. That idea definitely exists now. So big tech companies like Google and Facebook have tons and tons of data, and they can make a lot of predictions about what you as an individual will do.

SPIEGEL: Matthew Salganik is a traditional academic, a professor of sociology at Princeton University. But he's also someone who loves to code - the puzzle of it, the way that lines of simple directions typed into a computer can multiply your ability to understand the world. And even though modern life is full of Wired magazine articles explaining how machine learning is transforming every inch of our lives, to Matt, there's still a lot of potential in computers that hasn't been fully explored.

SALGANIK: Yeah, absolutely. It could really change things. It is really changing things, and it will really change things.

SPIEGEL: Specifically, Matt thought that it would be interesting to take the techniques and strategies developed in Silicon Valley to predict things in the commercial world and use them to look at the things that he and his colleagues in sociology thought were important, which brings us to that black hole of American attention that relentlessly sucks our time without pause or pity, Netflix.

(SOUNDBITE OF NETFLIX INTRO SOUND EFFECT)

SPIEGEL: In 2006, the company Netflix did something really interesting. It staged this massive competition where it made tons of its customer data available to anyone interested in building a computer model that improved its you-might-also-like predictions, then declared that the best model would win a lot of money.

(SOUNDBITE OF ARCHIVED RECORDING)

UNIDENTIFIED PERSON #4: What's so special about Napoleon Dynamite? For AT&T Labs researchers in the race for the Netflix prize, that was a million-dollar math problem.

SPIEGEL: The competition attracted thousands of teams from all over the world, and it totally worked out for Netflix. Predictions improved, and since that time, small children have been left to fend for themselves while their parents drool over hour 90 of "Orange Is The New Black."

(SOUNDBITE OF NETFLIX INTRO SOUND EFFECT)

SPIEGEL: So wouldn't it be amazing, Matt wondered, if you did something similar except instead of using a huge competition with thousands of teams to protect, you know, movie choice, you staged a massive computer competition to predict the actual important stuff that sociologists worked on, like what really predicted a child's GPA, which child would succeed? Matt thought it could work. He just needed the right data set, a huge study with tons and tons of intimate details about tons and tons of people.

SARA MCLANAHAN: So I was a divorced mother for 10 years and a single mother.

SPIEGEL: Sara McLanahan is a Princeton sociologist who works on the second floor of Matt's building. She's a well-respected academic now, but it wasn't always that way. In her 30s, she was a divorced mother of three in Houston, Texas, trying to finish her college degree when one of her new classes opened up a view of the world she'd never thought to consider.

MCLANAHAN: Sociology was all about - instead of me with my psychological problems or whatever, getting a divorce - it was all about all the people in the country were getting divorced. And all of a sudden, I realized I was in the midst of a massive transition in women's roles. But I had never thought of it that way. I'd always thought it had to do with me. Why did I want a job? Or why did - you know, it was all kind of very individualistic, to be resolved through some kind of therapy session.

(SOUNDBITE OF MUSIC)

SPIEGEL: Sara felt comforted by this idea that she was part of a larger pattern, that what she experienced as personal emotions and individual experiences were actually the result of a much broader system.

MCLANAHAN: Totally changed my life because it gave me a different perspective.

SPIEGEL: So she became a sociologist and 20 years ago, in 1998, began an enormous study, the Fragile Families study. It focuses on 5,000 children and their mothers and fathers in 20 locations across the U.S. Sara and her husband, Irv Garfinkel, have followed these people literally since the very first moment the children were born.

MCLANAHAN: OK, so the first question we said is, have you picked out a name for the baby yet?

SPIEGEL: Every few years, they went back and collected more information from the mothers, fathers, teachers, doctors, till they had a truly enormous catalog on each child. Then last year, right after Sara got her data on the 15-year-olds nice and organized, there was a knock on her door. It was Matt from the floor below with this very strange proposal.

MCLANAHAN: I don't know anything about this machine learning.

SPIEGEL: But once Matt explained, Sara agreed to contribute her data. Matt would open it to scientists from around the world and ask them to write programs that would search the data for hidden patterns that predicted what actually happened in the actual lives of the actual 15-year-olds that Sarah had studied. They would keep the names of each individual child secret.

And it wasn't like Tarra - none of the predictions would have any consequences. They were just trying to figure out if the computers could, by looking at patterns known to be important and also finding new ones, predict what actually happens in the life of a child.

SALGANIK: First question - have you picked out a name for the baby?

SPIEGEL: Matt says when he first started reviewing Sara's data, he was kind of taken aback.

SALGANIK: Like, the richness of this data is amazing. Everything that you think might be important in the life of a kid, I think they tried to measure all of that.

SPIEGEL: Other people were interested in the data, too. More than 400 teams applied. So Matt gave them all Sara's information about the kids from birth to age 9.

SALGANIK: How close do you feel to your mom? Would you say extremely close, quite close, fairly close or not very close?

SPIEGEL: And told them to predict what happened when each individual kid turned 15, stuff like grade point average and how well the kid stuck to a task.

SALGANIK: How well would you say you and your mother share ideas or talk about things that really matter?

SPIEGEL: Matt and his team already had that information, but they held a lot of it back with the idea that once all the models were in, they'd stage a big reveal where they'd match the data in Sara's study, like the real student grades, with the predictions that the computer models have made and see which model did best, just like Netflix.

Show me your spreadsheet.

SALGANIK: Oh, yeah. OK. OK. Here it is. Let's open it up.

SPIEGEL: I wanted to see what a human life looked like to a computer. So I asked Matt to open the spreadsheet with all of Sara's data in one place. It took a long time to load.

SALGANIK: There it is. It's loaded.

SPIEGEL: OK, so this is the spreadsheet.

SALGANIK: This is the spreadsheet.

SPIEGEL: In a way, it was beautiful - row after row of entries and numbers, more than 12,000 for each person in the study, translated from questions like, have you picked out a name for the baby yet, into a language computers could read.

SALGANIK: M1INTMAN (ph).

SPIEGEL: If you scrolled, they seemed to go on forever, a human life in math form. Matt truly believed that, buried in this spreadsheet, were new patterns that could help social scientists understand the path of a human life. Basically, he believed, like the panel of lawyers who had rejected Tarra, that patterns have real predictive power. And we need to pay attention to them.

SALGANIK: M1LANMIN (ph). CM1TWOC (ph). CM1FINT (ph).

ROSIN: INVISIBILIA will be right back.

(SOUNDBITE OF MUSIC)

ROSIN: This is INVISIBILIA. I'm Hanna Rosin. We've been looking at the life of Tarra Simmons and asking listeners to put themselves in the shoes of judges who are supposed to decide her fate. Is Tarra a good bet, or would she revert to the patterns that had repeatedly derailed her efforts to escape what everyone agrees was a horrible childhood? Alix continues the story.

SPIEGEL: Two days before her hearing at the state Supreme Court, I went to visit Tarra. And she was listening to a song on repeat, trying to find in its lyrics the emotional fortification that she needed to get herself through the thing she felt she was facing.

(SOUNDBITE OF SONG, "EVEN IF")

MERCYME: (Singing) They say sometimes you win some, sometimes you lose some. And right now, right now I'm losing bad.

SPIEGEL: Tarra, understandably, was worried. She'd been in a position of going in front of a panel of legal professionals whose job was to decide her fate before, and it hadn't been a very uplifting experience.

SIMMONS: I'm so afraid that it's going to be a no again because I've been here, and it didn't work out. And I was crushed, you know, because I really believed I had done enough to prove to them that I deserve this chance.

SPIEGEL: But it wasn't just Christian music that she was using to get her through. She told me she was trying to stay positive till the hearing, but after that, for the month that she would sit waiting for the state Supreme Court decision to come out, she knew she needed something more.

SIMMONS: I know that I'm probably going to have to develop, like, a safety plan because I do fear that I will feel suicidal - not that I would act on that. But I would be very depressed, and I would, you know, potentially be at risk for relapse. And so - but I think nothing can probably be as bad as April 14 - right? - because I wasn't prepared for that at all - at all.

SPIEGEL: Did you feel suicidal after that?

SIMMONS: Oh, yeah.

SPIEGEL: It wasn't a stretch to think that if the answer came back no, she might go spinning back to square one. Stress had done that to her before. So Tarra was planning to write out a long list of people to call. The idea was to carry it with her at all times.

SIMMONS: So when the decision comes out, you know, I can call people on my list who I really trust.

SPIEGEL: When I left Tarra that night, she was writing thank-you notes. Every once in a while, the silence in the kitchen was interrupted by the ping of a text, someone writing to remind her to stay strong.

(SOUNDBITE OF TEXT NOTIFICATIONS)

SIMMONS: Front row. Front row. Left row. Left row.

SPIEGEL: Two days later, about 40 minutes before the start of the hearing, Tarra was busy revising the seating chart she'd made for the court room. People respond to stress in different ways, and Tarra's preferred response appears to be extreme organization.

She'd studied the Supreme Court seating options before drafting her chart and was trying to make sure that reality conformed to the plan on the paper. But reality wasn't cooperating. A huge group of strangers had descended to watch the hearing, and she was having to regroup.

So you're just trying to figure out where everybody goes?

SIMMONS: Yes.

SPIEGEL: She finally got them settled, and then the clock struck 9.

(SOUNDBITE OF GAVEL BANGING)

UNIDENTIFIED PERSON #5: All rise.

SPIEGEL: Shon also looked anxious when he approached the bench but pretty quickly hit his stride. The core of his argument was that Tarra had already shown that the negative patterns that had defined her first 40 years were a thing of the past. She'd been clean for over six years, but Jean McElroy, the lawyer arguing against Tarra, wasn't buying it. I mean, she granted that Tarra looked OK now.

JEAN MCELROY: The process that she's gone through and the efforts that she's made - she is very clearly on the right trajectory.

HOPWOOD: You'd agree it's remarkable, wouldn't you?

MCELROY: It is remarkable. And the board acknowledged that as well in making its findings. But what the board was faced with was this pattern - repeated patterns of misconduct.

SPIEGEL: The patterns were the thing. Shon responded that during Tarra's time in prison, there had been a deep psychological change. So the, quote, "patterns" no longer fit.

HOPWOOD: And that was the first time that Ms. Simmons finally got treatment for the untreated trauma from her childhood and the sexual assaults that she was a victim of...

(SOUNDBITE OF MUSIC)

SPIEGEL: Psychological growth is less tangible and harder to measure than a prison record. It's the kind of thing that a computer might have trouble seeing. But it was, Shon argued, real.

And he was asking the people judging Tarra to recognize that and encourage it because if they didn't, it would just reinforce all of the negative patterns and prisoners would have a harder time finding a way through - a little space between the patterns that they could pull into a window and climb out of. That's what Tarra needed, Shon said, the court to believe that there was a place outside of patterns for her.

UNIDENTIFIED PERSON #6: Stand between them. OK.

SPIEGEL: Which is exactly what Tarra said, when after the hearing had concluded, she stood on the steps of the courthouse.

SIMMONS: And I want to stay optimistic that the Supreme Court is going to rule in my favor, and I know that I take with me a lot of people who are yearning for that hope of a second chance.

(SOUNDBITE OF MUSIC)

SPIEGEL: It would be nice to know for sure who could be trusted with a second chance, wouldn't it? Lovely to be able to predict, with reasonable certainty, not just what would happen to this criminal, but also to this child. If I read to him every night, if I sit with him at dinner, would that ensure he went to college? Or is it some other input that's important, more predictive?

That's what computers promise to define with all of their pattern-finding, what Silicon Valley has been hinting is soon to come. But Matt's experiment is, I believe, the first to use decades of detailed personal data, facts, but also the opinions and emotions of the subjects and the people most important to them, to predict individual outcomes in this kind of ambitious way.

So I think of it as kind of a first crack at testing how much we can predict about individual lives, not just lives like Tarra's, but yours and mine.

SALGANIK: So let's see. I need to change the code then.

SPIEGEL: Matt's big reveal - the day they finally crunched the data so that he could see the winners - started at 11 a.m. on a Monday. By 5, he told me, they'd have their answer.

SALGANIK: I don't think I've ever done a project that has one moment like this.

SPIEGEL: Really?

SALGANIK: When else do you get, like, one moment where it becomes clear?

SPIEGEL: It did feel odd that by 5, we would know what truly predicts the GPA of a child and stuff like how well she'll be able, when faced with difficulty, to persevere. It was exciting. Before they could get the answers, though, Matt and the two graduate students working with him, Ian Lundberg and Alex Kindel, had to review the code to make sure that it was perfectly perfect. Unfortunately, there were a lot of lines of code.

UNIDENTIFIED PERSON #7: Make a matrix.

SPIEGEL: And once they started doing trial runs, testing the code on small batches of data to make sure that it was working right - turned out that not all of those lines of code were perfectly perfect.

It...

SALGANIK: Failed.

(LAUGHTER)

SPIEGEL: D'oh.

And so it began. There were test batches...

(YELLING)

SPIEGEL: ...And then attempts to locate the problem. Then test batches...

(YELLING)

SPIEGEL: It went on and on.

(SOUNDBITE OF MUSIC)

SPIEGEL: Since Matt had told me before I came that the big reveal would be done by 5, I didn't bring a change of clothes or anything. I was planning on taking the train home. But round about 9, I decided that Matt - the very nice man - was a serious time underestimater, so I called a hotel for a room, then headed out to the local CVS for a toothbrush.

At the CVS, getting toothbrush - toothbrush, toothbrush, toothbrush - hello, M&Ms. And hello, Mr. Twizzlers.

(SOUNDBITE OF MUSIC)

SPIEGEL: They were still hunched over their computers when I got back. But by 11, there appeared to be good news.

SALGANIK: Something cool is about to happen.

SPIEGEL: The code was finally fixed.

SALGANIK: I think this is - are we ready?

UNIDENTIFIED PERSON #8: Yeah.

SALGANIK: Are we ready? We really want to do this?

UNIDENTIFIED PERSON #7: OK. This batch.

SPIEGEL: Even though he was tired, you could tell that Matt felt optimistic about the competition. After all, machine learning, this way of using computers, had given us so many things - Siri and Google Translate. Why wouldn't it be good at predicting which child got a good GPA and which became homeless?

SALGANIK: It's basically the same thing. You have some features. You have some labeled outcomes. A lot of things seem to be able to be solved in this way.

SPIEGEL: Finally, around 2 in the morning, the computer finished its crunching. Matt projected the results on a white screen at the front of the room.

SALGANIK: But is this true?

(SOUNDBITE OF MUSIC)

SPIEGEL: What Matt wanted to see was a harmony between at least one of the models submitted to the contest and the actual group of 15-year-olds that Sara had studied - at least one model able to predict with reasonable accuracy the outcomes of each child in the study, the grade point average of each child, how well they stuck to a task.

But none of the models did as well as Matt expected - none. If the models had been predictive, had matched the real world outcomes 100 percent, the screen at the front of the room would have been filled with tall, colorful towers - bars stretching from the floor of the Y-axis to the top, indicating that the predictions had gotten close to 100 percent accurate.

But there were no tall, colorful towers. Instead, what you saw was a bunch of squat bars crowded around the bottom like flattened mushrooms, indicating that the predictions were a lot closer to zero percent accurate than 100. Most of the graph was a vast, white space - just emptiness.

SALGANIK: I would say this is not impressive. This is - I think this is sad or disappointing.

SPIEGEL: But was it really disappointing? Or is that just an accurate representation of how predictable individual lives, like yours or mine, are?

DUNCAN WATTS: That's what we find everywhere.

SPIEGEL: Around the time I went to see Matt, I met with this man named Duncan Watts in New York City. Duncan works at Microsoft Research, where he does computational social science, including prediction studies similar to the one that Matt was doing. Over the years, Duncan says, he's done and read tons of them. And what he explained is that when it comes to predicting stuff like what will happen in a particular human life, the outcome that Matt found is kind of the outcome.

WATTS: We find exactly the same pattern everywhere we look, which is that there's a lot of white space. There's a lot that cannot be explained, whether it's a tweet or a person. When you're talking about individual outcomes, there's a lot of randomness. That's what I think.

SPIEGEL: There's a lot of randomness. Duncan told me that he's happened on that conclusion over and over in his own career using computers to make predictions and says almost all of the studies that he's read have found the same thing.

WATTS: It's all pointing in the same direction, which is that most things are mostly random, and the other half of this conversation is that people don't like that answer. And so they keep wanting a different answer. And yet...

SPIEGEL: And why do you think that they don't like this answer?

WATTS: Well, I don't know. I - you know, I've thought about this a lot. And this is now just me speculating about, you know, human psychology. But we abhor randomness. And they say, like, nature abhors a vacuum. Humans abhor randomness. We like deterministic stories.

SPIEGEL: We like to be able to say that we know what the woman whose childhood was relentlessly dismal, who was repeatedly convicted of theft and drug possession and assault, will do. Identifying a clear pattern in the facts in that way allows us to move through the world with some confidence. So even if it's only partially accurate, we cling to it.

WATTS: People make up these stories that aren't true. But they make them feel like they have control, right? And that means they can get up in the morning and do stuff. And that's useful. If you think that you can predict things, if you think that you can control things, even if you're wrong, it means you get up in the morning. And you feel confident. And you can invest your time and energy today in things that won't pay off until tomorrow.

SPIEGEL: Duncan believes that being more realistic about what patterns get you, though, is critical but also really hard to pull off because it involves accepting something that feels like a contradiction - that patterns are important and predictive. So if you have a large number of people, you can see that the kids who grow up in poor, broken families will, on average, do worse than kids from stable, affluent families. So structural differences do have huge consequences.

And if you are born white and middle class, as I was, you have, in a sense, been born on second base - but also that individual lives really can shoot off in any direction. You cannot know what will happen in the life of your child or whether your career will be derailed or soar. There is too much randomness, though the stories that we tell about ourselves and other people usually don't account for the role of chance.

WATTS: We want to think that, you know, my success story was determined. Like, I controlled that. I made that happen. Like, it wasn't luck because it was determined by me.

(SOUNDBITE OF MUSIC)

SPIEGEL: But accepting the real role of randomness, when judging ourselves and the people around us, is just as important as recognizing the power of pattern, Duncan says, because it has profound implications for fairness and for justice. Though, of course, randomness doesn't get a lot of respect in our algorithmic world.

SALGANIK: (Sighing).

SPIEGEL: Back in Princeton at 3 in the morning, standing in front of the vast white space, I raised randomness with Matt. But he seemed resistant.

It's a lot more random than you think. What do you think about that in light of this?

SALGANIK: No. I think we just have more work to do.

SPIEGEL: Matt pointed out that this was the first time that they had done anything like this, that there were a million decisions they made along the way that they could do better. Most of the arguments he was making were technical. But underneath, you felt a scientist belief in an ordered world and what discovering that order might provide to people in need.

SALGANIK: Worked on it for a while. We would eventually figure out, maybe...

SPIEGEL: You just have confidence in answers.

SALGANIK: I have confidence in, like, people thinking hard and working hard, figuring stuff out. I totally believe that progress is possible on all of this stuff if we work hard enough and think hard enough and have lots of people working on it. We can do better. And we will do better. And we'll see what we learn.

SPIEGEL: Matt was going to keep going. His work trying to shrink the white space was far from done. He wanted to make it smaller so that we could make the right choices about people like Tarra and ourselves.

(SOUNDBITE OF MUSIC)

SPIEGEL: Hello, Tarra?

I left Tarra and Shon literally on the courthouse steps the morning of the hearing. I had a plane to catch. And decisions in these state Supreme Court cases always take months to come out. We just needed to sit tight. But then while I was in the air, I got a text. Something had happened.

What happened?

SIMMONS: OK. So I was laying in bed with Eric (ph).

SPIEGEL: Eric is Tarra's husband.

SIMMONS: And he was asleep. And I was thinking about trying to take a nap. And then I get a text from Prachi from the ACLU of Washington. And she said congratulations. And I said to her, thank you. I said, you know, I just really don't think there's going to be any other answer but yes. And then she texts me back. She goes, no, the order. And I said, what?

SPIEGEL: Prachi immediately called to explain, told Tara that the Washington State Supreme Court had done something it almost never does - delivered its ruling on the same day as the hearing. Every single justice thought that Tara was not at risk of repeating the patterns in her past. They believed she deserved a second chance.

SIMMONS: I just, like, started crying and freaking out (laughter). Probably thought I was a little off my rocker or something. But I was, like, literally crying. And I got on my knees, and I was sobbing. And she said, you have a unanimous decision. You're sitting for the bar exam.

SPIEGEL: For Tarra, at least, there would be a place outside the patterns she could live. It took her a lot of work to get there but also a lot of luck. She realized that most people in her position - other former felons - would have a lot of trouble getting the break she'd gotten, given our attachment to patterns. So at that moment, she just felt extremely grateful.

I can't believe it.

SIMMONS: I know.

SPIEGEL: And what were you thinking when you were crying?

SIMMONS: I was just like, I'm finally free. That's what I kept thinking. It's like, I'm finally free. Like, I finally made it out.

(SOUNDBITE OF MUSIC)

ROSIN: That's Alix Spiegel. Stay tuned for a preview of our next episode.

(SOUNDBITE OF MUSIC)

UNIDENTIFIED PERSON #9: We're trying to get the bad guys - hostile alien species. Oh, no. There's a black hole.

ROSIN: Next time on INVISIBILIA, everyone sneaks off into their own daydream world every now and then.

UNIDENTIFIED PERSON #9: Just felt good.

SPIEGEL: But what happens when you love your daydreams so much it gets harder and harder to go back to reality?

ROSIN: What world do you want to live in?

UNIDENTIFIED PERSON #9: (Laughter). Well...

ROSIN: That and other tales on the pleasures...

UNIDENTIFIED PERSON #9: It's just electric.

ROSIN: ...And dangers...

UNIDENTIFIED PERSON #10: There's just nothing more, like, viscerally scary.

ROSIN: ...Of trying to live in between worlds.

SPIEGEL: Next week on INVISIBILIA.

(SOUNDBITE OF SONG, "STUCK IN THE PAST")

AIMEE MANN: (Singing) Stuck in the past like drawing rings around Saturn. Shadow is cast, but now it follows a pattern. I don't what that arrangement was. I could never tell. But you could. I had hopes, but the hopes all fell leveled with a smell of new wood. Stuck in the past. I plan it only on paper.

SPIEGEL: INVISIBILIA is hosted by me, Alix Spiegel.

ROSIN: And me, Hanna Rosin.

SPIEGEL: Our show is edited by Anne Gudenkauf. Our executive producer is Cara Tallo. INVISIBILIA is produced by Meghan Keane, Yowei Shaw and Abby Wendle. Our project manager is Liana Simstrom. Lulu Miller is a contributing editor.

ROSIN: We've had help on this episode from Alex Chang (ph), Rebecca Ramirez, Jon Hamilton, Andy Huether, Meredith Rizzo, Nancy Shute, Amy Weiss-Meyer, Greta Pittenger, Micah Ratner, Mark Memmott, Morgan Givens and Sidney Monday (ph). And our vice president of programming - Anya Grundmann.

SPIEGEL: Special thanks to Irv Garfinkel (ph), as well as Austin Jenkins (ph) for getting us that courtroom tape.

ROSIN: Also thanks to Ramtin Arablouei for additional music and Rick Carlos (ph) for his song "Rain."

SPIEGEL: And Aimee Mann for giving INVISIBILIA permission to use her song "Stuck In The Past" for this podcast - also for the songs that got me through the hellscape of my late 20s in New York. Thanks for that.

ROSIN: To learn more about this music and see original artwork for this episode by Sara Wang (ph), go to www.npr.org/invisibilia.

(SOUNDBITE OF SONG, "STUCK IN THE PAST")

MANN: (Singing) Stuck in the past, just drawing rings around Saturn.

ROSIN: And now for a moment of nonsense.

UNIDENTIFIED PERSON #11: You know what I got for us, though. I have apricot scrub, so we can all, like, wash our faces.

(LAUGHTER)

UNIDENTIFIED PERSON #11: Aren't you glad that you're a reporter here?

UNIDENTIFIED PERSON #12: Yeah.

(SOUNDBITE OF MUSIC)

SPIEGEL: Join us next week for more - Hanna.

ROSIN: INVISIBILIA.

(SOUNDBITE OF MUSIC)

SPIEGEL: Hey. So in addition to our stories, INVISIBILIA creates all sorts of cool original digital content around each episode on npr.org. Follow us on Facebook or Twitter to see photo essays, read Q&As with our experts and learn more about the topics we discuss every week. You can also sign up for our newsletter at npr.org/newsletter/invisibilia. Transcript provided by NPR, Copyright NPR.