99% Invisible - 421- You've Got Enron Mail!

Episode Date: November 11, 2020

Enron collapsed nearly 20 years ago, but chances are something you use today was affected by emails sent by 150 of the company's top employees. These emails — about meetings and energy markets but also affairs, divorces, and fraud — have helped create new technologies, fight terrorism, and added to our understanding of how we communicate. You’ve Got Enron Mail! Learn more and subscribe to Brought To You By…

This is 99% invisible. I'm Roman Mars. Your online activity creates tons and tons of data, and that data is valuable. Entire sectors of the economy are built on this principle. But there is one data set that has proven to be so incredibly valuable because it was available for free for anyone to use, and this data was gathered during the investigation of a crime. This is the story of the Enron emails that were collected as the government was building the case to prosecute the company that became the poster child for corporate crime. But since then, these emails have helped create new technologies,

fight terrorism, and added to our understanding of how we communicate. It's an incredible story. And when I heard it on business insiders brought to you by podcast, formerly called Household Name, I was like, damn, I wish we reported that. And since they did such a sell our job, I thought we'd just present their story to you. It is fascinating, you're going to love it.

To guide you through this twisted tale of crime and data, here's Dan Bobgoth. Imagine you're working for love it. To guide you through this twisted tale of crime and data, here's Dan Bobgoth. Imagine you're working for a big company, like say number seven on the Fortune 500 list. Oh, and it's around the year 2000. There's no Facebook, no Gmail. We're not thinking that much about privacy. And then the company you work for

goes bust in spectacular fashion. Well, it was the corporate fraud case that since shockwaves across Houston and the entire country the fall of Enron? Congressional hearings begin this morning in the Enron investigation. And then some regulator in Washington releases your work emails, all of them.

So all of a sudden, a lot of things in your life just became public, like? July 28, 1999, 142 pm. Attached are the above reference documents. Hard copies will follow. But also, some perhaps less routine business dealings. Subject, re, dark star. To further insulate the coal group

and you from any claim that Enron misused the information, I suggest that you transfer the information to me and I will hold it for safekeeping. And some cliche, bad workplace behavior. No subject. I'm heading to New Orleans this weekend to do some partying. No Europa, just sluts in the quarter. And let's not forget the classic 90s chain emails.

Hope you're having a pleasant first week of 1999, thought it would forward this on. Top 22 signs that you've had too much of the 90s. 22 cleaning up the dining area means getting it. There were even some exchanges with co-workers that really shouldn't have been in your work in box. No subject. Let me know when you're leaving. I'll be leaving probably in about 30 to 45 minutes.

Walk me out. I'll give you a big wet kiss. But I want more. I'll give you all you want. The emails you just heard read by actors are real emails. There are just a few of the hundreds of thousands sent around the year 2000 by some of Enron's highest ranking employees. few of the hundreds of thousands sent around the year 2000 by some of Enron's highest ranking employees.

And when these emails became public, for the first time, there was a database of thousands of real emails sent by real people that was available to the public and researchers, and at least one podcast host. But these emails aren't just a curiosity, they're not just a time capsule. I bet something you use today was touched by these emails aren't just a curiosity. They're not just a time capsule. I bet something you use today was touched by these emails. They've become a huge part of all of our lives. From business inside our institution, this is household name. Brands you can trust.

Brands you know stories you don't. I'm Dan Bobcoff. Enron collapsed because of greed and corruption and fraud. But the emails Enron employees sent and received have had an astounding afterlife. They were used to create Siri and develop spam filtering and artificial intelligence. They've helped us understand gender and power. But at what cost? What happens when so much of our technology is based on the writings of some fallen energy tycoons?

And should the emails have been released in the first place? This is the first time we've ever seen a video about how to use the internet. This is the first time we've ever seen a video about how to use the internet. These days, Enron is pretty much shorthand for corporate scandal. But back in the 90s, Enron was an energy trading company based in Houston. It bought, sold, and traded natural gas and electricity, and also apparently pulp and paper.

But the thing Enron is really famous for is how it collapsed. In 2001, it was the biggest bankruptcy ever at that point. It looked suspicious, because the company was telling everyone it was profitable and successful. And then out of nowhere, it went bust. The SEC investigated, prosecutors pounced, a number of top executives went to prison for fraud. Guilty verdicts in the biggest case of corporate fraud in history. Lawyers for Jeffrey Skilling and Kathlay threw around complicated notions about margin calls and short selling.

But the reason we have access to thousands of Enron's emails is because of something else. Something Enron did in California. California became one of the first states to deregulate its energy markets in the mid-90s. The idea was to introduce market forces and competition, you know lower prices and such, Enron had lobbied hard for this, and then after deregulation came into play, mysteriously, California started experiencing serious energy shortages. And whenever that happened, Enron and some other companies just coincidentally raked in a whole bunch of cash.

The worst one was physical withholding, so you just say, I'm not going to run my powerplant tomorrow. Pat Wood was an energy regulator with FERC, the Federal Energy Regulatory Commission. It was his job to help investigate this. Now, no one was dumb enough to say I'm not going to run it so I can make money off of a scarcity price, but that's what happens. I mean, when you don't run your power plant, which they were obligated to do, would says power suppliers would overplay maintenance issues, exaggerate problems so they could shut down their plans.

What kind of smart engineers can actually go question that? I mean, you really had to get people under oath to really figure all this out. There were other tricks. And Ron would buy power in California, move it to Nevada, and then sell it back at a profit. And Ron was making bank, but by the second half of 2000, the tricks turned into a full-on energy crisis. Electricity prices in California shot up 800% at one point. There were rolling blackouts affecting more than a million people.

Now Pat Wood is a free market sky. He was all for deregulating California's energy market. But what Enron did with that freedom is not what he had in mind when he was promoting free markets. I'm a big passionate defender of competition, but I'm ruthless against people who, you know, want to eff it up. So, for investigated Enron and the market manipulators, and through the investigation got tons of documents,

like these memos describing in detail how Enron planned to manipulate the California markets to make lots of money. And they were really pretty shocking. I mean, you know, for one who loves competition and markets, it just kind of made me nauseated because I thought, man, is this the enterprise that I help create?

Was this house of cards? I mean, this is ridiculous. So I walked down the hall and showed it to my other three commissioners, and I said, you know, I got a problem with this. During the investigation, Pat Wood and Ferck were mostly focused on the memos, but they'd also gotten a whole lot of emails from Endrun.

It was huge, it was thousands of, I don't remember if it's terabytes or whatever the word after teratism, but it was huge. So what were you finding in these emails that you wanted people to see? You know, 95% of them were about nothing that we were interested in. But it's hard to read. I mean, it was a huge amount of emails. So the other question was, you know, there might be something that somebody finds here

from reviewing this stuff that we might have missed. ["Must Have Been A Hard Decision"] Firk had to decide what to do with all this data. In 2003, Pat Wood was pissed about what Enron had done. So he kept thinking about transparency. Just put it all out there and let the public see what the company did.

But I'm sure behind the scenes, it must have been a hard decision to decide whether to release all these emails with all this personal information and irrelevant. Yeah, but I will tell you, honestly, then, that issue did not with not front and center as much as I, as much as it would be today, for example. So Pat Wood and the FERC commissioners made a fateful decision.

They just dumped the entire email archive on the internet. All the end run emails about gas trading and meeting scheduling and also the divorces and affairs and talk of parties, it was all there. I've heard different versions of what happened next. Some people say the emails were cleaned up that someone went through and got rid of social security numbers and bank account info, and stuff like that.

Other people say, and run employees were actually given a chance to go through and flag things they thought should be redacted. But either way, Pat would have met Firk didn't try that hard to clean the emails up. And then after they were public, he just kind of forgot about the emails, until I called a few weeks ago. And so what they were public, he just kind of forgot about the emails until I called a few weeks ago. And so what did you think would happen when you put all these emails

out in the world? Never dreamed. I mean, when you told me, when we talked last week, when you told me what had gone on, I mean, I can't tell you how much I've been looking at on the web since you pointed me in that direction. There was so much on the web about this information. And I had no idea what would come of this. The emails took on a life of their own, far beyond what anyone at Enron or FUR could have imagined. Artificial intelligence, voice assistance, counterterrorism software, all have roots in the Enron emails.

After FERC released the Enron emails back in 2003, they just kind of sat there. Even though they were public, no one was really reading them, because they were a mess. Like imagine you log into your email and you need to find one specific message, except there's no search function. You can't organize by date or sender or subject line, there's spam everywhere. That's what the Enron emails were like, and there were like half a million of them. But this is where the Enron email's strange afterlife begins. First, some academics bought the emails from FERC. It became known as the Enron Corpus.

Corpus, by the way, is my new favorite word. The Enron Corpus cost 10 grand. Then the researchers got to work cleaning it up, pairing it down, and organizing it into something that could be cataloged and searched and studied. And then, they went wild. They wrote papers, ran experiments, invented whole new areas of research,

like network science. My name's PJ Lamberson, and I'm an assistant professor in the communication department at UCLA. My research focuses on social networks and collaboration. We called PJ Lamberson because he was actually our producer, Sarah Wyman's professor.

Yeah, and that is a first day of class I will never forget. What happened? It was just so cool. Like, we all came in and on the board, he had projected a network that was made up of the N-RON emails. And so what we're looking at is a bunch of dots basically arranged in a pattern.

And every dot represents an email address in the corpus. And you can tell so much about Enron just from looking at this map, Dan. Like what? Some of the dots are bigger than other dots, which means that they're getting more emails. So are those the powerful people? Yeah. And you can also tell that like some people are getting way more emails than they're sending, which is also kind of indicative of them being more important maybe. But then the really cool thing, like the reason I still remember this class four years

after the fact, is that if you take a step back and just look at the entire network, you see something that's really interesting. What's that? So the way the map is organized, you can see projects, because people at work will email back and forth when they're working on something. Like we did with this episode. Exactly.

But the thing is, at Enron, they were not making podcasts, and some people were actually doing some really sketchy illegal things. And on this map, you can actually visually see the difference. Like the projects that were totally above board and fine look completely different from the ones that were shady. I guess the way I would describe it, if you look at the network where people are talking about like an elicit or illegal project, it looks like a really tight ball with a few little spikes sticking out of it. And so what that's showing you is that like for those illegal projects, the communication

is really concentrated among a core of individuals and they're not sharing that information or dispersing information about that project with other parts of the organization. Okay, and this is the best part because a computer has identified this. It's like a magic trick. The computer just has all of the data that's in the corpus that like in and of itself doesn't really make any sense to anyone and it just looks like a huge mess. But then once you run an algorithm on that data, it's like shining a black light on all of the corruption that was happening at N-Rot. Like you can just see it laid out bear in front of you in this network. Yeah, so this is like clearly very useful to people for a lot of things.

Yeah, and there's so much cool stuff that's happening with this technology. Like, people are using it to predict how viruses spread through populations because the software can identify the people within a group who are most likely to spread something to the rest of the group fast. So like the guy who's just going around shaking lots of people's hands will like show up

in this algorithm. Right, but maybe one of the most interesting applications of this technology is that it's actually being used to identify terrorist cells. So if you have phone records from a group of people, you can run these algorithms on those phone records and they can detect these abnormal patterns of communication and you can see where the terrorists cells are. Okay, so the technology we developed using emails from Enron is now being used to fight terror.

Yeah, it's being used for all kinds of stuff. The Enron emails have been a huge opportunity for researchers like Sarah's Professor. They are publicly available, there's no copyright, researchers can swap them between institutions because no one owns them, but they've also been this really big deal for any researcher technology that involves language. Because these emails, this corpus, is a rare example of unfiltered, uncensored, totally organic human communication. So the bankruptcy of Enron was really the wonderful stroke

of luck for researchers interested in conversation. This is Owen Rambo. He works at an artificial intelligence company called Elemental Cognition, used to teach a Columbia. He's been a part of lots of different research projects involving the N-Run emails, and a lot of them involve using the corpus to train computers to understand human language. It's unique. There's nothing quite like it. And it's real. You know, these are real people who are conversing not in order to create data for linguists, but in order to achieve their goals, whatever it was, you know, some work related goal or just tell each other jokes or whatever.

Before the N-Rine emails, researchers like Rambo mostly had to work with stilted conversations or texts from old newspapers. One typical example is that people are students are brought into a lab and play a game against each other and engage in conversation as part of the game playing. So they're real conversations, but they're very limited and they're not naturally occurring. But these N-RON emails were what people really say to one another, especially when they don't think anyone is reading over their shoulder. And they've taught Rambo and computers a lot about how humans communicate.

Like based on syntax and word choice, you can predict if an email sender is male or female, a boss or an underling, bosses write shorter emails, male bosses tend to write direct emails, like give me the report by Monday, female bosses tend to say things like, would you be able to finish the report by Monday? I can say something like it's hot in here,

and it can either be a speech act to inform you of this fact or it can be a speech act to request a view that you turn on the air conditioning. It is hot in here. I'm sorry. Yeah. It's actually okay. These are the kinds of problems that cause bugs in artificial intelligence. Machines aren't great at interpreting nuance or tone or intent. They need practice and the end run Corpus is like one giant perfect training ground for developing those skills. It's helped train spam filters. Hey, the Enron emails had a lot of spam. We can connect you to the world's rich and famous, capture the attention of millionaires.

The emails played a role in the development of Siri. Google reportedly used them when it was developing smart compose and Gmail. If you've used Gmail in the last year or so, you'll know what I'm talking about. This is that thing where it predicts what it thinks you want to say next. Sometimes it's actually really helpful.

But early versions have this bad habit of suggesting the phrase, I love you a little too often. If you're a researcher, you could spend hours sifting, organizing, studying these emails, and come to think of them purely as data. And then you might come across one like this. November 26, 2001, 723 PM, no subject. So you were looking for a one night stand after all? Whoever wrote that email probably didn't want it to have a long legacy.

Did you feel any ethical qualms using the N-Rion emails? I relied on the process having worked. The process being that people were given the chance to withdraw emails. This said, I had my doubts, because in one of the releases, the very first email you saw was a very personal email, which probably the sender didn't or the receiver more likely didn't want to spread around. A few years ago, Owen Rambo was on a train in Texas. didn't or the receiver more likely didn't want to spread around. A few years ago, Owen Rambo was on a train in Texas. He and his husband were sitting in

the dining car. And we started talking to the people who were added to our table and therefore were from Houston and I was working on Enron Day in and day out so I just said, oh did you work for Enron just like that. And the guy said yes. It was kind of like meeting a celebrity. This guy was one of the 150 in the corpus. And that was just sort of such a fascinating, weird coincidence. And it reminded me that this corpus,

which we give to our computers and run through algorithms and reduced to numbers and correlations, there really are real people at the other end. And you can meet them in Amtrak trains in Texas. And we then gossiped a little bit about other people who were mentioned in Run Corpus, who sort of almost seemed like people I know. So much of what we know about the world and how it works was somehow learned through this

corpus. So much of our technology was developed using the corpus. But Owen Rambo is right. These aren't just data points. These aren't just emails. They're real people. At one energy company, at one period of time, right before it went bust. And that raises all sorts of red flags. More about that after this.

There are two really obvious ethical issues with using the N-RON emails for anything. First of all, the people who wrote them did not sign up to be part of an academic study. They did not give researchers or robots permission to comb through all of their old conversations, and we'll get to that. But first, there's another problem here. The biases of the people writing the emails could become the biases of the AI system that's trained on them.

Amanda Levin-Dowsky teaches at NYU and studies how bias creeps into technology. And she's worried that a ton of our artificial intelligence is based at least in part on the emails written by 150 people at an energy company that went bust because of fraud. First of all, they're not geographically representative. A lot of those emails were from people

based in the Houston office. It's not going to be representative in terms of corporate culture because it was a Houston-based oil and gas company. Because it's 150 senior executives at this company, you're not going to have the kind of gender or racial diversity that you might expect at a different sort of company. And if you're looking for evidence of this bias, you don't have to look any further than the emails themselves.

Like there's this one email chain where someone sends an article about Bill Clinton's dog buddy getting hit by a car. The Enron official writes back, That is a shame for the dog. I'm very happy about Clinton's grief. There are emails about taking on the world wildlife fund. Subject, WWF Remember, this is the group that publicly announced that Enron has gotten away with murder for

years and we are going to get them. These are the emails underpinning a lot of our artificial intelligence. If there are misogynistic jokes, or shows of power in particular emails, those same implicit biases can become encoded in the AI that's trained on that corpus. Computer scientists tend to put this another way.

They call it garbage and garbage out. So, who wrote this stuff? I wanted to talk to someone who worked at NRAWN at the time, who actually wrote some of these emails. All the names are there, and I found that a lot of them list NRAWN as a former employer on their LinkedIn profiles.

So I started calling. And I hit a lot of dead ends. We're sorry you have reached a number that has been disconnected or is no longer in service. If you have reached it. While I was searching though, I met someone else who was obsessed with the emails, a guy named Sam Levine. He's an artist and educator who has used the N-Run emails in his own work. And his art deals with these questions.

It forces people, like me, to really think about why reading through the corpus makes us feel so uncomfortable. So one of my favorite series of emails to read from the archivists between two people who were married and they both worked at Enron and they're going through a divorce because she cheated on him, I think. And you can read their whole correspondence. And what do you see in these emails? You know, it's like, I saw you today from a distance and I thought about what we used to have

and I'm so sorry and I hope we can be friends again one day and stuff like that, right? Do you feel like you're watching a relationship just to integrate? Yeah, it's also something you shouldn't read really. You know, it's invasive. and stuff like that. Do you feel like you're watching a relationship disintegrate? Yeah. It's also something you shouldn't read, really. It's invasive. Yeah, I mean, there's something like I feel kind of dirty,

reading these emails, even though it's so long ago, it's been public for so long. And yet, it feels like, am I just a voyeur here? It is very voyeuristic. I tried to reach that couple Sam Levine was talking about to ask them how they felt, but I got a voicemail Hi, Dan Bopcoff calling from business insider in New York Eventually I reached the guy by email. He said he's not angry about his emails being released, but he didn't want to do an interview

Sam Levine the artist has been living with these emails for a few years now. Along with his colleague, Tiga Brain, he used the emails as the basis of an experimental art project. So, our project is called The Good Life, and in The Good Life you get the opportunity to receive all of the emails from the Enron Archive, direct-try inbox in the order that they were originally sent, and with the appropriate amount of time between a Gmail. Apparently, a few hundred people have signed up to get their actual inboxes clogged with old Enron emails.

Levin even did it himself. Do you have this on your main email account? Yeah. Is that, it's like not filtered or anything? No, no. So I read every single one. Oh my god. How much time do you spend a day on this? Not that long. And because like a lot of them are really short, you know. So it doesn't take that long to read all of them. And

it's a really interesting experience, I think. Because it's sort of like a lot of times you'll see the email come in and it'll be like meeting in room 10 and 15 minutes and you're like, oh, oh no, I've missed a meeting. I didn't know about that meeting. And then you'll open up the email and you're like, oh, right, right. This happened in 1999 actually. It's okay. I didn't actually have to go to that meeting.

Ironically or perhaps fittingly, some of the Enron emails get caught in the Vien Spam filters. How do you feel about your emails getting caught in spam filters that might have been trained on the very emails that you were trying to send? I think it's really nice. It works, I guess. I think it's great. I still wanted to know how it felt to be someone

whose emails were released in the corpus. Who's every word at Enron is now parsed and dissected by researchers without their consent. Eventually, someone picked up. Hello. Hi. Is this Mitchell Taylor? Yes.

Hi, Dan Bobcoff from Business Insider in New York. How are you? Good, how are you? Good, thank you. So I'm calling you for an interesting reason. Mitch Taylor was a managing director at Enron. He's also the guy Owen Rambo ran into on a train a few years back. Owen Rambo?

I had to jog his memory. Oh, oh, oh, oh, what's his last name? Rambo, R-A-M-B-O-W. I do remember this now. Okay, so I had him on the phone. This was my big shot. I wanted to know how he felt when all these emails were dumped on the web back in 2003. Yeah, you look at thoughts, some privacy advocates would have come to our defense, but at that

point, being with Enron, there was no one coming near defense. No one gave a sh** at that point. We were the evil empire, so everyone was happy for any bad things that happened to Enron people. He told me that there was so much news about Enron, so bad press about Enron back then that the email dump didn't really register It was just another thing that happened. I've read some of Mitch Taylor's emails by the way most of them sound like work emails subject Re-project granted update

Thanks for the update with regard to further interference from the PN, and the comment that the PUCN must approve each generation. You get the idea. I don't think I've ever gone and looked at which ones they had. I certainly didn't have a mistress. I didn't have any criminal stuff going on now. Whether I've passed along an inappropriate joke, I may have it, and I tend to be rather sarcastic at times in emails. That aspect of it never come back to to bite me in any way at least

that I've seen or that I'm aware of. When I first heard about this story I remember thinking how weird and cool it is that the emails from N-RON of all companies have been so important in our lives, that the company died, but the emails live on. But the more I dug into it, the weirder I started to feel about the corpus. I really shouldn't be able to read about strangers, divorces, and affairs.

I shouldn't have access to someone's daycare scheduling, even if it happened two decades ago. But on the other hand, I use Siri. I like when Gmail suggests what I was going to say so I don't have to type it out. I think it's nice that some real good and some real human progress came out of the Enron collapse, but at what cost? I really wanted to know what Pat would think about all this now. After all, he's the guy who released them in the first place back when he was a regulator, and he told me he's lost sleep over it well it was hard because i said that there go you know it was just uh... you know the impact on all it would

and now that i've been through privacy breaches on my own and i just thought man i was a huge accomplice and doing that to lot of people might love him within the town right now live uh... there's probably about a of people whose privacy was impacted significantly by what I did who live in my area code. He said that if he could do it over again, he'd probably release a lot of the emails, but would have taken much more care to scrub out the personal stuff, to protect the people in there who were just collateral damage.

What would you say to them? I'm sorry. If you didn't do anything wrong, you probably don't have anything to be ashamed of. If you did something wrong, I got you. For all those people in the middle who just had a normal expectation and privacy of just their personal affairs, or not personally, they their business affairs and how they would be viewed. You know, I think that's, you know, I wouldn't want that. What, you know, doing to others as you'd have them doing to you, I wouldn't probably want that done to me like that. This episode was produced by a bunch of people who had email addresses in the early 2000s,

like MusicMunchkin07 at Yahoo.com, Padulaa at Comcast.net and love to swim at Verizon.net. That's Sarah Wyman, Amy Padula, Jennifer Seagull, and there's also me. I think I was critical eye at AOL.com for a bit there. That was Dan Bobcoff. The show is called Brought to you by. It is excellent. It is produced by business insider.

We'll have a link in the show notes and on our website, 99pi.org. Joe Rosenberg organized this feature for us. We'll be back with an original homegrown 99pi story next week. See you then. Radio Topia. From PRX.