The Cathy's Dream Edition

Summary

Slate Money – The Cathy's Dream Edition (February 4, 2017)

Main Theme

This episode of Slate Money — dubbed "The Cathy's Dream Edition" — serves as a semi-farewell episode for co-host Cathy O'Neill. The discussion is centered on three major topics: the role of big data and psychometric profiling in political campaigns (specifically Cambridge Analytica and the Trump election), the growing role of artificial intelligence (AI) in law firms, and the threats facing government statistics in the early Trump administration. The episode is lively, candid, and full of insider insights, particularly from O'Neill, a data scientist and author of Weapons of Math Destruction.

---

Key Discussion Points & Insights

1. Cambridge Analytica, Trump, and Psychometric Profiling

[00:50–13:11]

• Context: The hosts dissect viral conspiracy theories about Cambridge Analytica's alleged role in the Trump campaign and Brexit using massive troves of big data (notably psychometric profiles derived from social media, especially Facebook).
• Explaining the Hype: Cathy clarifies the realities and exaggerations of psychometric profiling, debunking the narrative that Trump's victory was primarily engineered by secretive data analytics.
  • “All of this is somewhat true, but mostly bullshit.” — Cathy O’Neill [08:15]
• Comparison to Obama/Clinton Campaigns: Both sides used extensive data; Clinton inherited Obama’s, and their data science teams were possibly more sophisticated.
• The ‘Big Five/OCEAN’ Personality Profiles: Widely used in corporate settings, not unique to Cambridge Analytica.
• Voter Suppression Tactics: The Trump campaign controversially targeted Democratic voters with demotivating content (e.g., highlighting Clinton’s “super predator” comment).
  • “The thing that Trump did, that Trump's campaign did, that was really offensive... was this voter suppression campaign.” — Cathy O’Neill [11:05]
• Moral and Political Implications: Debate over whether targeting opposition voters to suppress turnout is merely negative advertising or a more sinister manipulation.
  • “Morally, they're very different.” — Jordan Weissmann [12:32]
• Key Takeaway: While data targeting is real and will shape future politics, the narrative that Trump “stole” the election through big data is overblown.

2. Artificial Intelligence in Law Firms

[13:13–25:49]

• Anthropomorphizing AI: Discussion on IBM’s “Watson” and common misconceptions about unified artificial intelligence.
  • “It really is kind of a cheat to call... this constellation of programs Watson.” — Cathy O’Neill [16:03]
• AI’s Strengths and Limitations: Great at pattern matching in closed domains like chess or poker, but struggles with ambiguity, judgment, or ethics.
• AI in Legal Practice:
  • E-discovery: Automated document review started by replacing young associates' labor with software capable of searching and clustering relevant documents.
    • “It replaced a shit ton of people, right?” — Cathy O’Neill [18:16]
  • Predictive Legal Tools: Highlighting Tax Foresight, which claims to predict case outcomes with high accuracy.
    • Debate ensues over whether computers should be making legal predictions; potential for both efficiency and ethical concerns.
    • “There's way too much trust being given to this algorithm... The algorithm, of course, depends crucially on what kind of attributes it asks for...” — Cathy O’Neill [23:06]
• Ethical Considerations: Will AI-driven risk analysis discourage important but low-probability cases (potentially inhibiting legal innovation)?
  • “It's a tool that could lend itself to a practical rather than ethical approach to law.” — Cathy O’Neill [25:13]
• Speculative Futures: Hedge funds might use AI to bet on litigation outcomes, further financializing the law.

3. The War on Public Statistics in the Trump Era

[26:43–37:56]

• Threat to Statistics: The Trump administration’s aversion to objective data and budget cuts to statistical agencies like the Bureau of Labor Statistics (BLS) and census.
  • “There's a threat here, both on the funding level and on the ideology level...” — Felix Salmon [27:26]
• Two Main Threats:
  1. Budget Cuts: Real risk to less-visible surveys (e.g., American Community Survey) crucial for understanding inequality and poverty.
  2. Suppression or Privatization: Fear that inconvenient statistics could be withheld, or outsourced to private firms (with non-transparent methodologies).
  • “Privatization of important information... statistics, like the census itself, started as an understanding of the public...” — Cathy O’Neill [34:26]
• Loss of Trust: Public skepticism in government data is rising, partly due to political campaigns undermining the credibility of statistics.
  • “68% of Trump supporters distrust federal data.” — Cathy O’Neill [35:23]
• Broader Economic Risks: If the integrity of U.S. economic data is questioned, it could harm markets and governance (as happened in Argentina).
  • “If you don't have a solid ground of like we believe what the facts on the ground are, no one wants to touch you.” — Felix Salmon [37:11]

4. Numbers Round & Final Reflections

[38:03–43:39]

• Cathy O’Neill’s Number: 81%
  • 81% of NYC taxi medallion loans ($690 million worth) are at risk of default due to Uber/Lyft disruption. [38:19]
• Jordan Weissmann’s Number: $24 million
  • Donations to the ACLU in one weekend after Trump’s “Muslim Ban” – six times their annual norm. [39:58]
• Felix Salmon’s Number: 12
  • The number of days since Trump’s inauguration (as of recording) – feels much longer to the hosts. [41:04]
• Final Thought: The administration is possibly leaving the Council of Economic Advisors vacant, reflecting disregard for expert economic policy advice.

---

Notable Quotes & Memorable Moments (with Timestamps)

• On Cambridge Analytica Hype:
  “All of this is somewhat true, but mostly bullshit.” — Cathy O’Neill [08:15]
• On Psychometric Profiling:
  “It’s an old idea… companies… have been using for decades.” — Cathy O’Neill [05:45]
• On Voter Suppression Ads:
  “They sent to African American Democratic voters... anti-Hillary videos that reminded them of the super predator comments from the 1980s.” — Cathy O’Neill [11:05]
• On AI in Law:
  “It's a tool that could lend itself to a practical rather than ethical approach to law.” — Cathy O’Neill [25:13]
• On Threats to Government Statistics:
  “If you don't have a solid ground of like we believe what the facts on the ground are, no one wants to touch you.” — Felix Salmon [37:11]
• Cathy's Parting Sentiment:
  “Thank you for all the kind letters to me because I love you guys. I really do.” — Cathy O’Neill [44:23]

---

Timestamps for Major Segments

• 00:50–13:11: Cambridge Analytica & Big Data in Political Campaigns
• 13:13–25:49: Artificial Intelligence & Law Firms
• 26:43–37:56: Statistics and the Trump Administration
• 38:03–43:39: Numbers Round & Reflections

---

Conclusion

This episode weaves together themes of data, technology, politics, and ethics with a signature Slate Money blend of wit and candor. Cathy O'Neill—on her farewell tour—anchors the discussion in both skepticism and clarity, challenging hype around data analytics while sounding alarms about real threats to public knowledge and democracy. The rapport among the hosts makes for a deeply engaging listen, with practical, philosophical, and personal takes on the forces shaping finance, law, and society in 2017.

Transcript

[Felix Salmon]

The following podcast contains explicit language.

[Jordan Weissmann]

Hello, and welcome to Kathy's Dream episode of Slate Money, your guide to the business and finance news of the week, brought to you this week by Cathy o', Neill, the data scientist, the principal and CEO of orca, which we don't entirely know what ORCA is yet, but we're sure it's going to be a unicorn, the author of Weapons of Math destruction, and the general all round excellent person who is not going to be on this show much more. In fact, we have one more special edition after this one where Kathy is going to weigh in on creative industries. And that's it.

[Cathy O'Neill]

Yeah. I'm gonna miss you guys.

[Jordan Weissmann]

We're gonna miss you, Kathy. And so as part of your farewell tour, we have given you the lay of the.

[Cathy O'Neill]

The reins in the saddle.

[Jordan Weissmann]

The reins in the saddle. So I'm Felix Hammond of Fusion. Jordan Weissman of Slate is also here. But this is your show.

[Cathy O'Neill]

Thank you.

[Jordan Weissmann]

You're taking the reins.

[Cathy O'Neill]

Wait, can I say hello?

[Jordan Weissmann]

You can say hello.

[Cathy O'Neill]

We already missed that. Okay.

[Felix Salmon]

Whoa. You've been waiting three years to do that?

[Cathy O'Neill]

I know, I know. It was as satisfying as I'd hoped.

[Felix Salmon]

It was a lot of.

[Cathy O'Neill]

Glad to be here.

[Felix Salmon]

Good climax.

[Jordan Weissmann]

So we're going to be talking about artificial intelligence in law firms. We're going to be talking about statistics in the age of Trump, which is very Cathy kind of thing, but the most Cathy thing that I think we're going to talk about because it manages to combine not only big data, but also right wing conspiracy theories, is it does Cambridge Analytica.

[Cathy O'Neill]

Yeah. There was this like really, you know, widely shared piece originally written by a Swiss magazine, Das magazine, last week, or I guess it was written a couple weeks ago, but it was translated into English and passed around like it, it ran around like wildfire, especially among people that like getting offended by Trump. So like along among the left. And it was basically about.

[Jordan Weissmann]

And it did eventually appear on Vice after. After appearing on like some weird samistat German website. That's right. It appeared like in sort of badly translated English.

[Cathy O'Neill]

Somehow it added to its authenticity that it was like just, you know, barely legible. And anyway, it was real conspiracy theory, except it was kind of based on facts. Except it pissed me off because it kind of wasn't anything after all.

[Felix Salmon]

Yeah. I just want to say, and I'll let you continue after this, but it really present. It was the closest I've ever seen a journalist try to present like a big data firm as straight up Bond villains like that.

[Jordan Weissmann]

There's actually A point in there which. Where one of the, like, big, evil, psychometric, you know, evil types moves to Singapore and changes his name to Dr. Spectre.

[Felix Salmon]

Yeah. I mean, okay, there was just a link, too. It was like. And we'll let you research that more on your own.

[Jordan Weissmann]

I clicked on the link and I got a 403 error, so I have no idea.

[Felix Salmon]

There's so much more to look know.

[Cathy O'Neill]

Quite a few empty links.

[Felix Salmon]

Anyway, we need to. We need to. Kathy.

[Cathy O'Neill]

Yeah. What's it about? It's about. Well, it's about the idea that Trump's digital campaign, his ad campaign that he used on Facebook in the hours before the election, was based on these psychometric profiles of every single voter in the United States. And, you know, it's true. It is actually true, but what they did was they took the guy who sort of invented this technology from their perspective, and they profiled him and they just sort of walked through his life where his technology, this invention, the secret sauce that he'd invented, was being misused and abused and, like, all the remorse he was feeling about it. Yeah.

[Jordan Weissmann]

Because apparently this was used not only to elect Trump, but also to persuade Britain to leave the eu.

[Cathy O'Neill]

That's right.

[Felix Salmon]

That's right. And it was. It was done by a company called.

[Jordan Weissmann]

Cambridge Analytica, which happens to be owned by the secretive, evil billionaire Robert Mercer, whose daughter is Trump's closest friend or something.

[Cathy O'Neill]

Yeah.

[Felix Salmon]

Rebecca Mercer, who played a role. Who has played a role in actually picking people in the administration.

[Cathy O'Neill]

Yeah, it's very much tied into the. And we're going to get to that later as well.

[Jordan Weissmann]

But, but wait, can I just say, because we're probably not going to get to this bit anymore, that Robert Mercer made his billions by being the CEO of rentech.

[Cathy O'Neill]

Yes.

[Jordan Weissmann]

So Jim Simons, who's the founder of rentech, is one of our sort of slate money heroes, and we love him. But his CEO, Robert Mercer, turns out to be, you know, Dr. Spector.

[Cathy O'Neill]

I wouldn't exactly say love from my. From where I'm coming from, but yes, Rentech, Renaissance Technology on Long Island, Stony Brook, they have these really, really rich guys. And Jim Simons does sort of like philanthropic. Philanthropic and scientific things with his money, and Mercer does Trumpian things with his money.

[Felix Salmon]

But anyway, so we get. We. Let's, let's talk about exact. Exactly what Cambridge Analytica did. Like, what is this psychometric?

[Cathy O'Neill]

Okay, so. And by the way, there's also another thing that's floating around the web that people love sharing, which is it. Which is like a basically a 10 minute video of the Cambridge Analytica CEO Nix. His name is Nix.

[Jordan Weissmann]

He's very English.

[Cathy O'Neill]

Very English, talking about the sort of psychometric profiles of every voter in the country and how we use these profiles to manipulate people.

[Jordan Weissmann]

Okay, so first, but let's start three and a half minutes into this segment, Kathy with what is a psychometric profile?

[Cathy O'Neill]

Well, basically it's a, it's an old idea that's being used for decades by companies to, to sort of profile somebody, like, understand them, their, their personality in five different ways. And one of them is like their extrovertedness, one of them is their neuroticism. There's three other things and there's all sorts of personality tests that do this, that measure people's first five. It's called the big five, or it's sometimes called the ocean. Yeah, because there's ocean.

[Felix Salmon]

They're all, they all have odinous extroversion. Yeah, I forget, conscientiousness, something like that.

[Cathy O'Neill]

Yeah.

[Jordan Weissmann]

So you put those all together and you get an ocean profile and you can have your O and your C and your E. And each one can be measured on a scale of one to however much. And that makes everyone uniquely susceptible to certain messaging in a way that someone who doesn't have your exact psychometric profile would not be. And the, the kind of conspiracy theory here is that what the Trump campaign did was find a way to put together literally 175,000 different ads and depending on your exact psychometric profile, target you with exactly the ad which they, which was targeted to you. And this was not only like, so for instance, when Barack Obama was using social media to get out the vote and stuff, he would be like, we need to target this kind of person and we'll hone our ads to get these kind of people. But number one, those distinctions were mostly demographic. It's like, we want to get black people on the south side of Chicago or something like that. And then it wasn't psychographic. And then the other important part of that was that he was nearly always only targeting Democrats who might vote for him. And one of the interesting things that Trump was allegedly doing was targeting Democrats too. And he reckoned that any Democrat they could persuade not to vote for Clinton was just as good as finding a Republican Party.

[Felix Salmon]

Not even allegedly. I mean, they admitted this. Okay, I'm forgetting one really key thing because we keep the tangents. The really important big data part of all this is that these psychometric profiles are put together with Facebook and other social media data that's A huge or originally. And then they started incorporating things like consumer purchases and stuff. So it's all the big veins of big consumer data that Kathy's been worrying about for like, literally three years now on this show, according to this theory, were used to elect Donald Trump. So that is the bottom line.

[Cathy O'Neill]

Now I'm going to put like the, you know, the ruler to this dick, because all of this is somewhat true, but mostly bullshit. Okay.

[Jordan Weissmann]

And this is why Cathy o' Neill cannot leave Slate money, because he's the only person who can actually come out and say this.

[Cathy O'Neill]

Well, first of all, it's really important to know that Hillary Clinton inherited all of Obama's information about voters. And I'm absolutely convinced that that information was better than what Trump got. Second of all, yeah, they didn't, in the Clinton campaign, they claimed they didn't do this big five ocean profiling, but. But they did a lot of things. And like, marketing profiling is almost the same thing. It's just phrased differently, it's framed differently. So I. And a lot of these kinds of measurements of people are very redundant. So you're like, okay, one of them is called neuroticism, but then the other one is like, you know, the likelihood of purchase. You know, it's, it's almost the same thing. There's very, very consistent redundancies among these kinds of profiles. So that's, number one, is that Hillary just had lots of data and there. And by the way, she had an enormous team of data scientists. And the most important critical thing in here with this narrative of this guy whose secret sauce was stolen is there. There was a publicly available paper on how to do this using Twitter data in 2011. This is not secret sauce. This stuff was available to data scientists. I am absolutely convinced that someone in the Clinton campaign tried this out.

[Felix Salmon]

Yeah, I think this is also one of those stories where hindsight bias really is playing a big with how we interpret it because like, so this is a, you know, this company seems really impressive because, ooh, it was part of the Brexit campaign and then it was part of the Trump campaign. And so it seems like they must be doing something right. But if you think about it, well, they were part of the Brexit campaign and Brexit worked for a million and a half different reasons. They happen to be owned by in part or their major investor was Robert Mercer, who was big on the Trump campaign. So they ended up getting involved with the Trump campaign, too. Trump won for a million and a half different reasons. There is plenty of room for coincidence here, but hindsight bias is gonna make us read into a narrative here.

[Cathy O'Neill]

Yes, thank you. And that's exactly right. And I didn't get quite. Get finished with my description of the video of Nix, which is like, if you see that video and it's very easy to find, the guy is sociopathic. I mean, he's just so hateable.

[Felix Salmon]

Okay.

[Cathy O'Neill]

And he's really creepy. And he's like, and we are going to manipulate all of these voters according to how they are most vulnerable. I mean, he doesn't say exactly that, but you, that video, and you're like, this guy is nasty. And so, like, on the left, it's really easy to think, oh, Trump stole this election with creepy data analytics. And I'm here to say the, the, you know, we are doing. We started it. Obama started it. Clinton was doing it as much as she could. It didn't work for her. But. And by the way, one last thing, Felix, you're right. Like the thing that Trump did, that Trump's campaign did, that was really offensive. That took the step past. What I think Clinton was capable of doing was he did this voter suppression campaign stuff where they actually sent. They sent to African American Democratic voters these videos, anti Hillary videos that reminded them of the super predator comments from the 1980s.

[Felix Salmon]

I wasn't that offended by that. I just want to say right now, a lot of liberals did get offended. And this is a little bit of tangent, but why not? We're doing tangents. This episode, which is. That was just a. That was just a trollish way of the Trump campaign saying we're doing negative advertising online. You know, it was negative advertising.

[Cathy O'Neill]

I have a chapter of this stuff in my book, right. I talk about the, the anti Democratic nature of this asymmetric targeting to voters from the campaigns. The campaigns know way more than the voters do. And I had this little. I had a paragraph or two about what I was afraid of happening with voter suppression. Like, you have get out the vote, which is established for campaigns. But what about don't get out the vote? That's what I was suggesting. My editor made me take it out. She said, that's just, it's, it's too. You know, nobody's actually doing that yet. You're going to sound like a crazy person. And I was like, okay, I guess I can't say that, but that's what's happening now. Yeah, I think a lot of people are offended by that.

[Jordan Weissmann]

And I do think there is something very just like anti Democratic about trying to persuade the other team to not vote.

[Cathy O'Neill]

Yes.

[Jordan Weissmann]

And I think that it's perfectly legitimate to go out to your own team and say, go out and vote. And it's much less legitimate to go out to the other team and say, don't vote. And I. And that's tactically, has exactly the same effect, but morally, they're very different.

[Felix Salmon]

I just. I think the distance between this and negative advertising isn't as great as some people made it out to be. I will say their attitude about it and actively calling it voter suppression does imply a sort of grossness or it doesn't play. There is a grossness about that.

[Cathy O'Neill]

Well, listen, all I can say is we should be offended. We shouldn't think Trump did anything special.

[Felix Salmon]

Yeah.

[Cathy O'Neill]

And we should absolutely gird ourselves for this to happen more and more in the future.

[Felix Salmon]

Yeah.

[Jordan Weissmann]

So, Kathy.

[Cathy O'Neill]

Yeah.

[Jordan Weissmann]

This is your episode.

[Cathy O'Neill]

Okay.

[Jordan Weissmann]

And yet somehow, because you're a generous soul, you said that you have allowed Jordan to wonk out about his favorite. Because we know that Jordan is basically a lawyer monkey.

[Cathy O'Neill]

I know he wants. He's like a wannabe.

[Felix Salmon]

Deep down, I chose not to.

[Cathy O'Neill]

He married a lawyer.

[Felix Salmon]

I married a lawyer.

[Jordan Weissmann]

He married a lawyer. He, like, deep down, he wants to pull a Shane Farrow and just quit journalism and go to law school.

[Felix Salmon]

I made the choice not to do that. I came this close.

[Cathy O'Neill]

But you do like talking about law, so I threw you a bone.

[Felix Salmon]

Thank you. Well, it's law and big data. So, yeah, for a long time, people have been talking about whether, you know, computers are basically going to replace lawyers, like, whether Watson's going to come and take every lawyer's job in America. And I guess we're taking another step in that direction right now, and a much more interesting step than some of the previous advances.

[Jordan Weissmann]

And before we go on to this, I want. Because I am hoarding all my Cathy time as much as I can, I need to ask Kathy this question, which has been so about a month ago, when I was, you know, in a certain Swiss town, I had an interesting conversation about artificial intelligence and about Watson with an AI professor from, I think it was Dartmouth. And he told me that there is really no such thing as Watson. That Watson does a whole bunch of different things. It does, like, translations, it plays chess, it plays Jeopardy, it does your hr, you know, payroll. It can do a million different things. But what he told me was basically, yeah, there's no one machine which can do all of those things. The IBM is essentially branding a bunch of different AI machines with one brand name Watson, to make it very easy to, like, you know, to make it seem very, very clever and powerful. But in fact, all of those things are so different in terms of the AI that's required, that they're basically, they, they may as well just be different machines.

[Cathy O'Neill]

I mean, I'm going to just throw. That wasn't a question.

[Jordan Weissmann]

It's a question like, is he right?

[Cathy O'Neill]

Well, yes. So I actually wrote a post, and it's kind of philosophical issue, but like, I wrote a post a few months ago about anthropomorphizing AI and people, you know, basically people complain about that. They say, oh, you shouldn't anthropomorphize AI because it gives people the impression that there's real learning going on, which there isn't. Right.

[Jordan Weissmann]

Like, well, that's machine learning.

[Cathy O'Neill]

Yes, well, that, that, that itself is anthropomorphizing. What's actually happening. Right.

[Jordan Weissmann]

But like, what's in just pattern matching going on. But Watson, you know, recently won a game of Texas hold', Em, which is, you know, the latest thing that everyone got impressed by. Yeah, well, and, and clearly got better at hold' em over.

[Cathy O'Neill]

Sure, sure. And that's what AI is really good at. I mean, AI is really good at mastering a very finite universe where the rules are very clear and there's no judgments or ethical calls. Chess, go, poker. What AI is really bad at. And what Watson, he's bad at is anything else. Right. Like any kind of judgment call, anything like trying to, like, model the world and understand whether a given statement is true or not, if it's at all ambiguous. And so it's, it really is kind of a, it's, it's kind of a cheat to call him Watson, to call this collection, this constellation of programs Watson. But, oh, by the way, one last thing, which if you, if you Google search image search AI, which I do every now and then, every single picture you're going to see is a male brain.

[Jordan Weissmann]

It's like a human brain, the difference between the male brain and the female.

[Cathy O'Neill]

Because you, you sort of like the, you see a face, it's like it's clearly a male. And it's really interesting to me. Like, so we have a tendency, or we have an absolute. We've made the decision as a culture that AI will be anthropomorphized no matter what. And the, what the question is, how do we grapple with.

[Jordan Weissmann]

When I do a mental Google Image search on AI do it, do it, I always get like, a combination of, like, aqua greens and blues and a bunch of digits and like white digits basically coming straight out of the Matrix movies. And somehow people are so unimaginative. They took this movie that came out 15 years ago and they're like. That's the universal shorthand for AI.

[Cathy O'Neill]

That's right.

[Felix Salmon]

Anyway, lawyers, Lawyers. So we're gonna talk about lawyers.

[Jordan Weissmann]

No, actually we don't digress on this show, you know, so this is actually.

[Felix Salmon]

I'm tie it back. So you're talking about ambiguity. Ye can accomplish. So for a long time, you know, the thing that law firms were using, you know, robots or artificial intelligence for was discovery.

[Cathy O'Neill]

Discovery.

[Felix Salmon]

Which is the really. The most basic, like doc review. Like really the most basic stuff is.

[Jordan Weissmann]

Can you please just. It's like control F on a bunch of emails. It is essentially look for this. Can you, can you find the word muppets? It is anywhere in these golden emails? Absolutely.

[Felix Salmon]

The grunt work. I mean, that was.

[Cathy O'Neill]

By the way, I mean, it's really an interesting thing already because number one, it replaced a shit ton of people, right?

[Felix Salmon]

Yeah. So what was interesting about that was you used to have at law firms, you would have a hoard of young associates come in and just dig through boxes for hours and hours and hours and then they would bill their time and that was the discovery process. And at high priced law firms, you had high priced lawyers doing all of this. And then eventually you had what was called ediscovery and that got adopted. And you had kind of very rudimentary versions of it where it was basically control F. And then you got less rudimentary versions where they were kind of grouping together phrases and finding things that may have been related based on natural language.

[Cathy O'Neill]

I'm going to jump in here and say, because I know a little bit about ediscovery, that it's actually pretty subtle because there are, for example, different ways to spell certain words. Like if you're looking for the word muppet, but people know they're being watched, they'll spell it with three p's or something like that. So there's this kind of. There's this whole kind of oe of like methodology around E discovery.

[Felix Salmon]

And extremely, it's extremely lucrative. And one thing saves time, can potentially save money that people discovered was actually more.

[Jordan Weissmann]

What is it lucrative for?

[Felix Salmon]

Well, it's lucrative for the firms doing it. It could be lucrative if they're.

[Jordan Weissmann]

But they're not billing out as much. Right. Because they can't bill out.

[Felix Salmon]

No, no, no, sorry. For E discovery companies, it's lucrative for them. For law firms, it can save you on labor. So to some degree you might be able to get away with fewer associates. But this is getting really deep into a slight tangent. What it wasn't really doing was actually like, you know, helping lawyers figure out how to build a case. Right. It wasn't really making. Helping them make legal decisions. It was helping them just like gather evidence. Right.

[Jordan Weissmann]

It wasn't running ocean profiles on the jury box and saying this is the best argument to use for this particular jury.

[Felix Salmon]

Exactly. And so now there's this company out of Toronto apparently, and we found this in a Maclean's or Kathy found this in McLean's article. It's called Tax Foresight. Your client comes, he tells you the case, he writes up the details for you, you input it into this program and then it does a search through all the case law and then it gives some prediction about your percentage chance of winning. Is this going to prevail upon a tax judge? Again, he says it can be like 90% accurate. That's the company's word for it. But the idea is that people are really now working towards this where it's the next step in legal technology, where it's not just about shaving off labor costs for filing through documents. It's really about doing the job lawyers sort of the same way people talk about Watson doing the job of doctors or assisting them. And that's a big question.

[Jordan Weissmann]

This is something where it is quite obvious to me that the computers are going to be a little bit like they are in meteorology. The computers are obviously much better at certain kind of weather forecast, Y type things than humans are. And just like they're better at playing chess than humans are. But at all points, a human operated computer where the human comes up with the final judgment is better than either. And a very good chess player working in conjunction with a chess playing computer is always going to be better than any computer. And that's still the case. And I feel like this is a really useful tool for any lawyer. They can get the computer to do like 90% of the work and then they, if they know how to, to make it even better, they can do that.

[Felix Salmon]

Yeah, I think there is a lot.

[Cathy O'Neill]

I was just gonna. One of the things that was startling to me about this article was how completely gung ho it was about this technology. It went to the right.

[Jordan Weissmann]

I mean, it means fewer lawyers. So that's gotta be good, right?

[Felix Salmon]

Yeah, I think that's part of the reason.

[Cathy O'Neill]

But, but I mean it was really by the end of the article is saying, well, now that we've gotten that law part out of the way. What are judges gonna do with their time? I mean, basically.

[Felix Salmon]

Right, Yeah.

[Cathy O'Neill]

I think we're really jumping the gun if we think that's what's gonna happen in the next few years, number one, because just what Felix said, I think what this tool could be really good at and could be saving a lot of time and really making lawyers lives better and more interesting is if you give enough sort of attributes of a given case, it could look through the corpus of old cases and find related cases very, very quickly. And then the lawyer who's in charge can say, oh, that case was won, that case was won, that case was lost. And they can sort of. And this is probably what lawyers actually do, right?

[Felix Salmon]

There's, I mean, speeding up legal research, and there have been people who've tried, but yeah, I mean, this is. That, that is a. I mean, legal research is a cost center for law firms. Right. Because you have to pay an associate to sit around and go and just look through Westlaw and find more, you know, some court decision that's relevant. And this does, you know, this makes it almost instantaneous, it seems, or close to it. I think another interesting thing here is how a lawyer could try to input different details about the case and different versions of his case as he understands it and play the facts and see if there's a way to massage it to get a better potential outcome. Well, you can imagine point I was.

[Cathy O'Neill]

Going to make is that, like, there's way too much trust being given to this algorithm. And the algorithm, of course, depends crucially on what kind of attributes it asks for and what kind of attributes it's been trained to think are important. There's always going to be cases where the attributes that they ask for aren't actually the ones that ended up being important in the decision.

[Felix Salmon]

Yeah, I mean, and, you know, it says a 90% chance of being correct. And I mean, those 10% of chances might just be the hard cases. I mean, that's the other thing here, is that maybe it's good on the really, really easy cases that you're just like, you basically need to tell your potential client that, like, buddy, you don't have a shot in hell. But, like, the 10% where this gets it wrong are the ones where you really actually need to hire a lawyer to tell you.

[Cathy O'Neill]

And beyond the, like, maybe, maybe poor accuracy, maybe not so poor accuracy, maybe that could be addressed. The really interesting question to me is, like, the ethics of it. Like, this is an unethical algorithm, right? It's not trying to say whether this should have won or should have lost. It's basically just saying this has a 95% of chance of winning or losing. And if, if you start making decisions on which cases to take based on that number rather than whether it's a good idea whether the law should, we need new precedent or something along those lines, that's pretty scary.

[Felix Salmon]

Well, lawyers aren't taking it based on whether or not an outcome is good or not. Lawyers are taking cases based on, you know, is it a client walking the door will pay me?

[Jordan Weissmann]

In fact, the ethics of lawyering are such that you should be like relatively agnostic, you know, as to the ethics of the case and you should argue sides as strongly as you can and then just see which one wins.

[Felix Salmon]

Yeah.

[Cathy O'Neill]

So you don't think there's any ethics involved here? Didn't we talk at some point about law firms that or about financiers who like decide on whether what are the chances of something winning and therefore finance them or not?

[Jordan Weissmann]

Well, that's like Peter Thiel funding a lawsuit against Gorka.

[Felix Salmon]

But yeah, I think the lawyers themselves, I mean, the idea of legal representation in this country is it's supposed to be kind of mercenaries, like everyone on the criminal side, everyone's entitled to representation on the civil side, I think everyone should be entitled to representation on the civil side for the most part, if they want it. But you know, that's, that's another.

[Cathy O'Neill]

I'm just saying it's a tool that could lend itself to a practical rather than.

[Felix Salmon]

That's true. You can imagine a really crazy version of this where financiers are using it to pick which cases they are. I mean like, this is like way down beyond, way beyond tax law. But if this, something like this in a sci fi world could be built up to take to deal with like more difficult kind of lucrative litigation, I'm down, man.

[Jordan Weissmann]

I'm going to start up a hedge fund which uses this thing to predict the outcome of lawsuits. And then I'm just going to buy a stake in the outcome of all those lawsuits and it'll make me lots of money on the ones which win and I'll retire a billionaire.

[Cathy O'Neill]

You can have like high frequency law trading.

[Jordan Weissmann]

I mean, it doesn't even need to be high frequency. One of my, one of the hedge funds I know the best is this hedge fund called Elliot Associates, which is basically its entire existence is down to litigation strategies. And they're like, we have found this litigation which we think is going to win and it's going to be very profitable. And that's how they make money. If you could do that in a kind of more hands off way without actually having to do the litigation yourself. But just being able to look at all of the litigation which is going on on the planet and then make bets on the ones which are going to win, I don't know. And that's. That. That's kind of what bill ackman did with. With herbalife. Right. He kind of made a bet that various herbalife litigation was going to come down and the FTC was going to close it down. And that bet lost.

[Felix Salmon]

He really could have used a better algorithm.

[Cathy O'Neill]

He also tried to affect the outcome. Felix.

[Jordan Weissmann]

Yes.

[Cathy O'Neill]

Tell me about statistics.

[Jordan Weissmann]

Okay. You know what I'd like to tell you about? Tell me is that one of my favorite financial journalist is this wonderful is a compatriot of mine, another English person called Mona Chalabi, who used to work for five, 30 years.

[Cathy O'Neill]

She's actually my favorite person.

[Jordan Weissmann]

She's awesome.

[Cathy O'Neill]

Oh, my God. The vagina dispatches are amazing. The vagina dispatches are an amazing series about everything you'd ever want to know and more about your vagina. And they're on the guardian. Everyone should check them out.

[Jordan Weissmann]

She now works for the Guardian.

[Cathy O'Neill]

Yes.

[Jordan Weissmann]

And she wrote a really good piece this week about basically the risk of the Trump administration to our favorite thing, which is statistics.

[Cathy O'Neill]

Yes.

[Jordan Weissmann]

And so I want to give you a statistic about statistics. This is a statistic which appeared in Monon's piece. The bureau of labor statistics, they're just one of the many statistical arms of the U.S. government. But they do a lot of very important statistics, including our payrolls report that we try not to talk about every month. The bureau of labor statistics has a budget of $600 million a year. This is big money. This is the kind of money that only government can do. And we now have an administration which is explicitly opposed to almost all statistics, or what you might call objective reality. They don't like objective yardsticks for what they're doing. They want to just be able to send out a tweet saying we won. And then everyone should just report that they won without trying to, you know, look at statistics to find out if it's true. And so the. There's a threat here, both on the funding level and on the ideology level that the completely nonpartisan. All we care about is accuracy. Multi decade tradition of the BLS and various other government agencies is threatened now like it has never been threatened before. And that we might lose some of the time series which so much value is in. And if you stop a time series, you can't just like pick it up again. It's, it's very hard.

[Felix Salmon]

I, so I think the, and I've talked to some former government officials about this too, because I think it's something that's in the air. Pretty much every econ reporter is a little bit worried about this right now. And I think the, the two, they're two separate threats really. There, like you said, there's the funding level, which is I think a very serious threat and we can get deep into that. But like the, what Trump will do to the budgets of these agencies and what his people might try to make them discontinue. Those are real deep concerns. The separate threat, the more conspiratorial they might fudge the numbers. I think that's less of a real concern. And part of that is because of this deep tradition of nonpartisanship that exists at these bureaus like the bls. If someone were to actually walk in. I was talking to a former chief economist over there, right. He was telling me that this is a group of people who literally, when they are doing the jobs report, they put paper, newspapers over the windows of their offices so that no one from the Department of Labor can even see in to see what they are doing. These are incredibly secretive people who take their job very seriously. You know, and if there was an official coming in and saying you need to change, you know, the denominator here to make it look better, the chances of that getting leaked are very, very high. People would find out very quickly and it would cause chaos, absolute chaos.

[Jordan Weissmann]

That is not.

[Felix Salmon]

Yeah, that's about.

[Jordan Weissmann]

But there's other things which they could do which are much more realistic. Like just say, oh, that statistic you've just come up with doesn't fit our narrative. So you're not allowed to release it.

[Felix Salmon]

I mean, that, that itself, that is, I mean that is even that is ham handed enough though that that would cause that would come out.

[Cathy O'Neill]

That's, that's already kind of happening, right? There's like gag orders on various kinds of.

[Felix Salmon]

Yeah.

[Jordan Weissmann]

And Jordan, like no one is saying that they're going to be able to gag these statistics without anyone knowing about it. But if they gag the statistic and they refuse to let it come out and it's not publicly available in public data sets and they, they stop the release of all of these time series, that's bad. Even if they do it with great publicity.

[Felix Salmon]

So I Think so.

[Cathy O'Neill]

By the way, I want to, I wanted to jump in, like, I talked to my FOIA expert friend, the Freedom of Information act, and like all of the stuff that he said that Trump has declared is unreleasable by the EPA and potentially the Bureau of Labor Statistics would still be discoverable under foia. But then the question wouldn't be like, would you ever get it? You wouldn't get it on time. You would get it at some later date, which would make it relatively useless and stale, I think.

[Felix Salmon]

So here's the thing. You have to think about who's in Trump's administration, right? There are a bunch of financiers. There are a bunch of, you know, there's guys like Wilbur Ross, right. Andy Puzder, who's, you know, a former fast food executive is probably going to get confirmed in the end. I think he's going to be heading the Department of Labor. These are guys who understand the value of specifically the BLS numbers, and they understand that if there was any question about their integrity of what was going on at the BLS specifically, there would be a riot in the markets, among other places that you, but you cannot.

[Jordan Weissmann]

There are BLS numbers which the markets care about, and there are BLS numbers which the markets don't care about. So they have something called the supplemental poverty measure, which is really important.

[Felix Salmon]

That's the census, that's cps. So that's why I'm talking about the BLS number. And this is very, this is actually an important distinction to make.

[Jordan Weissmann]

Okay. So, but the point is there are a bunch of statistical agencies who measure things like racial inequality, poverty rates of abortion stuff, which the Trump administration has lots of reason not to want to measure. And the idea that, like, we can be reassured in any way by the fact there are a relatively small number of data series which the markets really care about, I think is deeply wrong.

[Felix Salmon]

No, no, what I'm saying. So I, again, I'm, I am talking specifically about the conspiracies about them playing with, like, the unemployment figures, right, where there is a big concern. And actually Mona does kind of talk about this in her article, but the big fear is that they're going to go after the Census Bureau's funding, particularly for something called the American Community Survey, which is sort of this rolling sample survey, right? The, you have the, the regular census that comes out every 10 years, and that's what determines things like congressional districting, things like that. And then you have the American Community Survey, which gives us a lot more specific detail about things like income poverty. You also have something called the cps. But anyway, you have these specific, smaller surveys that require money to do. And that's. Right. That's where we get a lot of our information about things like poverty, about inequality.

[Jordan Weissmann]

And you're saying that we should be worried about.

[Felix Salmon]

And we should be worried about those. Not because someone is going to quietly go and say, hey, you need to cover up this result, or you need to stop, you know, because. Or you need to change this number, but because Republicans have explicitly talked about eliminating funding for some of these things. The acs in particular, has been targeted by Mick Mulvaney, who is currently nominated to be the head Donald Trump's budget director. Mick Mulvaney thinks in his heart of hearts that basically all the data in the ACS could be replicated by the private sector. He has talked about this at public hearings. This is, I think, the real fear.

[Jordan Weissmann]

And this is one other thing as well. Well, is that it is very much in line with Trump administration ideals to take a lot of the statistical data gathering and privatize it.

[Cathy O'Neill]

So that's that. Thank you, guys. I haven't been saying anything because you guys have said all the things I want to say. I think we're all scared about similar types of things. Although I want to say that Trump didn't seem to be afraid of a trade war, which businesses don't like, but he did it anyway. Right. So I actually think that all statistics are up for grabs, but the. The major thing is this privatization of important information. And this, you know, statistics, like the census itself, started as kind of like an understanding of the public. And it was meant, in a large part, not completely, because there was Japanese internment issues as well, but like, in large part for the public good. And now what we're seeing is more and more data is being privately owned. And the, The. I think the sort of perfect example of this is Steve Bannon. Steve Bannon, who, you know, tries to undermine people's faith in statistics at all times. I have a statistic.

[Jordan Weissmann]

You have a faith in statistics. Statistics, yeah.

[Cathy O'Neill]

So 68% of Trump supporters distrust federal data.

[Felix Salmon]

Yeah.

[Cathy O'Neill]

Okay. I mean, it's insane how, how successful this has been, this campaign to stop believing. Believing federal government. But at the same time, guess where else Steve Bannon is. He's on the board of Cambridge Analytica, that place we were talking about, that uses privatized data on every single voter to manipulate votes. So he believes in statistics when he gets to own it and manipulate it, but he undermines the public's belief in statistics.

[Jordan Weissmann]

And for every single data series that the federal government puts out, you know, they spend a certain amount of money and then they get a data series in return. And in principle, you could, you could put that out to tender and you could ask a whole bunch of private companies, can you give us this data series and we'll just go with the lowest bidder. Now obviously there's a bunch of problems with that, but one of the big problems with that is that the private companies are going to bid much lower if they get to keep their methodologies secret. And private and public methodologies are key to reliable statistics. And as someone who used to write about emerging markets a lot, I can tell you that if you go to a country like Argentina where no one believes the inflation statistics, that has knock on effects throughout the entire economy and businesses, which you have no obvious reason to be affected by wonky inflation statistics wind up getting damaged in very, very real ways because they just can't raise capital anymore. Because if you don't have a solid ground of like we believe what the facts on the ground are, no one wants to touch you.

[Felix Salmon]

Yeah. And I think the subtle thing here is like it's consistency. It's that you can't even take a government data series and then change the methodology and say, okay, we were calculating inflation one way but now we're going to give it over to this company with secret sauce. Right. Like this is, and this is something that when you guys. Like I've watched Mulvaney respond to arguments like this, right? Like I've watched him in hearings, like with data, with data people, where he's asked him why can't the private sector do this? And they've given him like they've given him responses akin to what we're talking about here. And it's just blank. There is, there's no receptiveness. And so you are getting an administration that doesn't seem to understand the value of this public, consistent, transparent data.

[Jordan Weissmann]

All right, since this is the Cathy o' Neill episode, I feel like Cathy o' Neill should have the first number in the numbers round.

[Cathy O'Neill]

Yeah, my number is embarrassingly local, but I still find it really interesting. So in New York City, there's a taxi crisis.

[Jordan Weissmann]

What's the taxi crisis?

[Cathy O'Neill]

It's a taxi medallion crisis.

[Jordan Weissmann]

Wait, you have to give the number the taxi owners.

[Cathy O'Neill]

Yes.

[Jordan Weissmann]

I'm looking at getting poorer. What's your number, Gabby?

[Cathy O'Neill]

My number is 81%. Thank you, Jordan. So 81% of the 690 million do of loans for taxi Medallions are at risk of default.

[Felix Salmon]

Oh, shit.

[Jordan Weissmann]

Well, I mean, technically, 100% of almost all loans are at risk of default.

[Felix Salmon]

High risk, I think.

[Cathy O'Neill]

Yeah, okay, you're right. You're right. That's not a very precise statement. But I feel like it's a real crisis in the world of taxi medallion loans.

[Jordan Weissmann]

There has been massively diminished liquidity in taxi medallions in general and taxi medallion loans that the taxi medallion loans used to just get rolled over, but a lot of the lenders are now insolvent, and so they're not rolling over the loans. And there's a big financing problem among taxi medallions. And the people who are getting hurt are mainly the lenders, a lot of whom are insolvent and going bust, but there's basically only like, three of those. And the people who own the medallions, who tend to be large fleet owners, who we also don't have a lot of sympathy with.

[Cathy O'Neill]

Oh, so we don't mind.

[Felix Salmon]

Wait, I don't think how much. How much is because of Uber. Like, what percentage?

[Cathy O'Neill]

Well, it's just kind of ironic for it to come in the week where there was that, like, JFK taxi strike, by the way. I went to that. It was an amazing, amazing experience. No taxi, there was no traffic to get to the airport. And then everybody hated Uber. And there's this, like, hashtag delete Uber. But, you know, I guess. I guess we're not going to worry about it because taxis are going to continue to test.

[Jordan Weissmann]

The supply of medallions in the city is fixed. There's about 14,000. And so long as there's 14,000 medallions out there, the value of those medallions is really not important to normal people.

[Cathy O'Neill]

That's right.

[Jordan Weissmann]

So talking of which, like, there were large protests over the weekend at the really unspeakable executive order banning a bunch of Muslims from entering the country. And My number is 24 million, which is the number of dollars that was donated to the ACLU in one weekend. This is absol. Unprecedented. Normally the ACLU gets about $4 million of donations per year. They managed to get 24 million in one weekend. And what fascinates me, as Slate Money listeners will remember, is that I have this theory that philanthropy always needs to work with the government in order to be effective, that you have to kind of like, find a way of getting the government to do what you want it to do, otherwise you will never really be able to scale. But the ACLU is really the exception to that rule, and it's the one place where you can Try and make a philanthropic donation against the government. And that's clearly what a lot of people want to do right now.

[Felix Salmon]

My number is 12 as of at the moment we are recording this. It's been 12 days since inauguration.

[Cathy O'Neill]

Oh, my God. That's impossible.

[Felix Salmon]

Yeah, it's over.

[Jordan Weissmann]

It feels like 12 years.

[Felix Salmon]

Yeah. Right. But each impossible. Yeah, there's like. It's like that Modest Mouse song. Right? Like the years go fast, but the days are so slow.

[Cathy O'Neill]

Absolutely.

[Felix Salmon]

But so specifically, I'm bringing this up because at this point, Donald Trump has still not named ahead of the Council of Economic Advisors. And a few episodes ago, I talked about how it was possibly going to be Larry Kudlow. He was maybe going to fill this role, but he's not actually an economist or an academic economist. So this might be an administration without economists right now. This administration has one economist in it. It's got Peter Navarro, who's like an anti trade, you know, guru, anti China trade guru who's in the White House. But the CEA is not being filled. So right now we are, it looks like, and there's talk of it never being filled right now. Political playbook kind of thing about how Trump's just like, maybe not even going to run.

[Cathy O'Neill]

Maybe he's going to appoint Steve Bannon to run that, too.

[Felix Salmon]

Well, that's a possibility.

[Jordan Weissmann]

The point is he has. He has Gary Cohn running the neck. Someone who is not a political wonk. And maybe Jordan can explain this to me, but I have never really understood the point of having a CEA and an nec. What is the point of having both with the.

[Felix Salmon]

Well, what the. Well, NEC is more of a. It's more of like a operational role. You're coordinating different economic parts of the government that deal with economics. The CEA is actually supposed to give you advice on, like, oh, we're heading into a recession. Like, maybe here's what you should do in terms of fighting it. Like, you know, Christine Romer was like, hey, Barack Obama, I think you need a big stimulus right now. Like, a couple hundred billion dollars would be a good idea. And so that was. And Barack Obama was like, oh, okay, yeah, we should do that. Like, that's the kind of thing you're supposed to have them around for and.

[Cathy O'Neill]

For giving you sage advice.

[Felix Salmon]

Yeah. Giving you, like, actually informed that he doesn't have that. Yeah, right. It's just like. But anyway, it's not shocking. It's just like, Like, I feel, in a way, vindicated that we are mostly. Not entirely, but mostly, it appears, heading for an administration without economists, and I'm.

[Jordan Weissmann]

All in favor in principle. I feel like an administration without economists, like, that's probably the best part of the Trump administration in any case. On which note, I think we need to wrap this show up. Cathy o', Neill, Yeah. Thank you for curating this show.

[Cathy O'Neill]

I hope everyone enjoyed my dream.

[Jordan Weissmann]

This is you are of course welcome to come back to curate any show in the future. We might drag you back against your will and just tie you into the chair at some point, force you to opine on certain things. But we do have you for one more show on creativity and then after that it's going to get interesting. I kind of have this secret hope that we might be able to get Mona Jalabi in one sort of way.

[Cathy O'Neill]

Oh, my God, she's the best.

[Jordan Weissmann]

So that's a possibility. Find out who we're going to be having on this show because there's a lot of great names that you've been sending in.

[Felix Salmon]

Thank you.

[Jordan Weissmann]

Thank you very much to everyone who's been nominating people to replace the irreplaceable.

[Cathy O'Neill]

And thank you for all the kind letters to me because I love you guys. I really do.

[Jordan Weissmann]

Keep the emails coming. The email address is slate moneyleep.com do keep on subscribing because we are going to have some fun stuff even without Kathy. Many thanks to Zach Dynastine, who produced this show, as well as to Steve Flicktight and Andy Bowers, the executive producers here at the Panoply Network, which [e-mail] panoply so we are going to talk to you next week with another special guest, a man called Derek on the Slate. Money in the life that we call home the years go fast and.