Episode 14: Disinformation and economic data, with Dr. Elise Gould

Synopsis

Dr. Elise Gould, senior economist at the Economic Policy Institute, joins Rachel and Dan to pull back the curtain on how economic data is collected, revised, and communicated — and how that process is vulnerable to political manipulation. We talk about the federal statistical agencies that produce employment and wage data, the role of transparency and revision in maintaining trust, and the pressure those institutions are under today. Rachel and Dan close with two lenses: Human vs. System Integrity and Maybe.

Interview

• Dr. Elise Gould — senior economist, Economic Policy Institute
• Economic Policy Institute — nonpartisan think tank focused on employment, wages, and inequality
• EPI Microdata — EPI's publicly accessible economic microdata site
• State of Working America Data Library — EPI's public database of wages, employment, and inequality data
• EPI Data Accountability Dashboard — EPI's tool for tracking parallel measures and monitoring changes to federal data
• Bureau of Labor Statistics — federal agency that produces jobs and unemployment data
• Philosophy Minis — Jonny Thomson's Instagram account featuring short philosophical vignettes (referenced by Dan in the lenses segment)

Lenses

Lens 1: Human vs. System Integrity

Information systems rely on some combination of built-in mechanisms and individual actors — whistleblowers, researchers, editors, external stakeholders — to maintain the integrity of their data. This lens asks where that responsibility actually lives.

• What mechanisms are built into the system for detecting anomalies or integrity issues?
• How much does the system rely on humans to address failures of integrity or reliability?
• When there are gaps, inconsistencies, or suspicious patterns in the data, whose job is it to surface them?
• What would it take for an integrity failure to go unnoticed — and how much of that risk has the system actually designed against?

Lens 2: Maybe

Drawn from the parable of the lost horse, this lens challenges the impulse to frame information as inherently good or bad news. Data systems — like dashboards and reports — routinely signal conclusions for users, even with incomplete context.

• When presenting data as either good or bad, does the system also present the sufficient context to explain why?
• How does the system signal urgency or alarm — and how are users empowered to specify the rules of urgency?
• What tools does the system offer users to reach their own conclusions without pushing them to a predetermined frame?

Edited by Jared Landis (https://www.landispodcastediting.com/)

_____________________________________________________

Personnel

• Dan Brown, Host
• Rachel Price, Host

Music

• Turtle Up Fool, by Elliot

_____________________________________________________

Unchecked is a production of Curious Squid

Curious Squid is a digital design consulting firm specializing in information architecture, user experience, and product design

Transcript

Elise Gould 0:00

When we're talking about whether or not we can trust data, I think it's very important to look at the data sources. And the less transparent a commentator or a politician is, it calls more into question what it is that they might be saying.

Sarah 0:18

You're listening to Unchecked, the podcast about the architecture of disinformation with Dan Brown and Rachel Price.

Rachel 0:31

My sister and her husband are bringing their new doggy down for me to meet.

Dan 0:36

That's almost better than getting a new dog yourself. It is actually. What kind of dog did they get?

Rachel 0:41

It's a mutt. It's yellow and lanky. Amazing. And his name is Quill, which I think is such a good name.

Dan 0:48

That is a good name. Are they Marvel fans?

Rachel 0:51

They did not name the dogs. They adopted a seven-month-old that came with the name Quill. Is that a Marvel thing?

Dan 0:58

Peter Quill is a character in Guardians of the Galaxy. Yes. Oh great. Love that. You should tell them they will be asked.

Rachel 1:05

Well, I'm sure that my brother-in-law knows that, and I just didn't know that. This seems like something he definitely knows.

Dan 1:13

Okay.

Rachel 1:13

Yeah, so we're gonna meet Quill and just gonna have we've had a rough two and a half months. So we're looking forward to this weekend. Just having some good family time.

Dan 1:26

That's great.

Rachel 1:27

No real agenda, hopefully plenty of sleep.

Dan 1:30

Good.

Rachel 1:30

No new germs. I can't say that. I've just jinxed it right there. My desk is made of wood. I'm pretty sure we'll all have a cough forever, but people are sleeping now.

Dan 1:42

That's really important.

Rachel 1:43

So that I feel like we've crossed that bridge at least. What are you up to this weekend?

Dan 1:49

Well, it's a weekend, so we're going to a frisbee game. Their kids are playing, I think their main rival high school. So that'll be Sunday morning. I think I'm meeting my best friend for coffee tomorrow morning.

Rachel 2:00

Like you haven't met them yet, but you think they're gonna be your best friend?

Dan 2:03

Or like your actual best friend. I keep I keep I keep hoping. I keep hoping this is the one. You know, I mean, I am very lucky, man. This guy, uh he and I have known each other for almost 40 years.

Rachel 2:21

Wow.

Dan 2:21

Now we met when we were in high school, we went to college together, and then we both ended up here.

Rachel 2:26

Wow. That's cool.

Dan 2:28

Yeah. So we live like a 15-minute drive apart. We find a coffee place in the middle and we meet up. He's such a good friend. He was my best man at my wedding, godfather to my children.

Rachel 2:38

So that's great.

Dan 2:41

So that's it.

Rachel 2:41

And I got what are you cooking this weekend?

Dan 2:44

That's a good question. Tonight, I don't usually know what I'm cooking until it's happening.

Rachel 2:49

Oh, I forget. You don't have a four-year-old in your house.

Dan 2:51

No.

Rachel 2:52

Okay. There's a little bit more margin.

Dan 2:54

Also, I tell the people at my dinner table there's only one person at this table I need to impress. And it's usually not a person co no. No. I eat whatever I make.

Rachel 3:07

Who do you have to impress?

Dan 3:09

My lovely wife.

Rachel 3:10

Oh. I thought you were gonna say your mother-in-law.

Dan 3:14

Oh god. First of all, Merle is very generous with the compliments. And if she doesn't like something, which is very rare, she'll say this was interesting. And that is literally the extent of the feedback that I get. Yeah. Anyway, uh, but tonight I am making cod. Cod what? Sarah found this great recipe that we've now made a couple times that's like cod in like a it's braised in sort of a coconut Thai flavored.

Rachel 3:44

Ooh, delightful.

Dan 3:46

It's great.

Rachel 3:46

I love that. We don't eat cod in my house. We eat salmon. But that's because we live in Seattle. And so if you don't eat salmon, you pay more taxes. It's a really regressive. It's a really regressive tax system.

Dan 4:00

That's amazing.

Rachel 4:01

Yeah.

Dan 4:02

We've just found a salmon that we can eat. Normally we would get Pacific salmon or Alaskan salmon. But yeah, there's a new new There's a new salmon in town? There is a new salmon in town. It's this farmed Atlantic salmon that's I know kind of sacrilege saying that to a Seattleite, but it works. It does what we need it to do. Great.

Rachel 4:21

Dan, why are we talking about economics on this podcast?

Dan 4:26

I started thinking about this when the tariffs came around. And I was really frustrated with well, with the tariffs, but also the way the administration was talking about it. And it just felt like they were taking advantage of the fact that the US economy, the world economy, is one of these hyper objects that is so massive it's difficult to understand. We've seen this time and again that the stock market is a proxy for how well the economy is doing. And we know that's not true either.

Rachel 4:58

Not true at all.

Dan 5:00

So I have the very good fortune of living with two economists. My wife has a master's degree in economics. My mother-in-law, who lives with us, has a PhD in economics. So we talk about this a lot at the dinner table. I just felt like there was a lot of misinformation about the economy, and it was verging on disinformation. That the way the administration was talking about the economy was being manipulative. And so I wanted to talk to an economist, and I did a bit of research and uh found Elise. She and I had happened to go to the same college, so that's always a good in. So that's that's why I wanted to bring on someone to talk about economics for that reason.

Rachel 5:44

Aaron Ross Powell One of the things I enjoy about learning about economics and listening to economists is the more you dig in, the more you realize how much the economy and theories of economics touch every aspect of our lives. Yeah. And like you said, it's a hyper object. So it's like it's so big you can't even really comprehend it.

Dan 6:06

Right.

Rachel 6:07

And you just keep running into it. Yes. You know, no matter which way you turn, you just keep running into it. And I think especially in our current uh situation, there are a lot of weird ass levers being pulled in our economy that are getting a lot of attention. And so it's helpful to try and wrap your head around at least a little bit of it or at least develop some awareness. Right. I have this probably naive belief that economists understand the economy somewhat completely. I think a lot of the more humble ones would argue they probably, you know, like no one can understand the entire economy completely.

Dan 6:44

Right.

Rachel 6:44

But I think it's a really good thing to develop some awareness and at least try to like pick apart the parts that interest you and that you care about. And so for me right now, thank you to Dan and Elise, I'm starting to really look at data and the economy and misinformation around economic data as just one little way for me to grasp what's going on here.

Dan 7:06

Well, we should set listeners' expectations. The the intent that I had was to talk about the economy kind of in these big terms, sort of the messaging terms. And it turns out Elise is really just a masterful data analyst too. And we ended up talking a lot about economic data, which was as soon as we started talking about it, the three of us just got so much more excited. So I'm excited for folks to listen.

Rachel 7:31

Let's go check it out.

Dan 7:38

Today, Rachel and I are excited to talk to Elise Gould. She is a senior economist at the Economic Policy Institute, a nonprofit, nonpartisan think tank in Washington, D.C. She conducts research on employment, wages, and economic inequality. She frequently speaks to the media and occasionally testifies before Congress. Elise holds a master's in public affairs from the University of Texas at Austin and a PhD in economics from the University of Wisconsin at Madison. Elise, thanks for joining us today. Actually, I was gonna ask you a slightly different first question. Of course.

Elise Gould 8:11

You said there was no bait and switch, and look. There it is. Here's the bait and switch right now.

Dan 8:16

There was a bait and switch on the bait and switch. My household was all abuzz about numbers that came out earlier this week. Why should we care about these numbers that came out? Oh, right, your household this week. I forgot. Yeah. Lots of economists. Your two economist household. Very normal. It's a guys, it's a lot. It's just a lot. I married into a family of economists. It's a lot.

Elise Gould 8:38

I married an economist, so you know, there we are.

Dan 8:40

Why do we care? What what yeah, what's all the buzz?

Elise Gould 8:43

So this week we got the jobs numbers. That's the data that comes out. Usually it's on the first Friday of every month. This week it came out on a Wednesday because of the shutdown, but it was just a few days delayed. Not a big deal, not like what we saw in October. And so the jobs data is made up of two main surveys. One is a survey of households, and another is a survey of employers. And they produce information that moves markets that employers care about, that policymakers care about. And there are a number of different data points. The top line numbers that people tend to talk about are the change in payroll employment. So how many jobs were added or lost in the economy in a given month compared to the month before. And the other number that people care a lot about is the unemployment rate. So that's the share of people who want a job or have a job, the share of that, which is the labor force, that do not have a job but have been actively searching for a job. So I was actually going to talk to you about the unemployment rate.

Dan 9:43

Yeah, let's do it.

Elise Gould 9:44

Because I think it's an interesting measure when we think about okay, how do you come up with that unemployment rate? So the Bureau of Labor Statistics, they have a survey and they survey about 60,000 households every month. Households get into the survey because they're trying to find a nationally representative sample of people, you know, according to where they live and their race and their gender and their education and all sorts of different information about them. And then they they ask them questions. And one of the questions that they ask is whether or not they have a job, whether or not they're looking for a job, along with all the other demographic questions. So the unemployment rate is this particular, very specific piece of information that you're looking at everybody who either has a job or has actively looked for a job in the last four weeks. It's a very specific measure. And those two populations make up the labor force. And the unemployment rate is the number of people who actively looked for a job but didn't have one, divided by that labor force. And that's how we get the unemployment rate. And so they're measuring this very specific thing. So there could be workers that are discouraged. Maybe they didn't look in the last four weeks, but they looked in the last year, they're not counted. People who didn't actively look at all, they're not counted. People who are working a part-time job but want a full-time job, they're not counted in the official unemployment rate, which is also called the U3. There are actually six measures of unemployment labor market under utilization. Some people call the underemployment rate. That's the U6 that includes some others. So there's these very specific measures. And these households that are in the survey, they're asked these questions for four months in a row, and then they're out of the survey for eight months, and they're brought back in to ask the same questions for four more months. So you can look at a panel of people. But oftentimes when we look at the data, you're looking at a cross-section. So in a given month, in this case, January, you know, what was the unemployment rate? And that's how it's measured.

Dan 11:34

You said that they swap households in and out of this survey. Is there sort of a methodological reason why they swap households in and out of the survey?

Elise Gould 11:44

Well, they're trying to get a nationally representative sample. And so they're gonna sample people and they don't always get responses, right? That's one of the issues with not having an agency that has enough funding. They can't go after people and try to get more responses. Or they might be in the survey for one month and then they fall off. They have attrition and they're they don't stay in the sample. And so you have a fair amount of attrition that happens. So if you again, you don't you're under resourced, you can't send people out and try to get those answers from them. And so they're gonna have a rotating group of people in the sample, in any of the household surveys that the Census Bureau or the Bureau of Labor Statistics runs, there's gonna be a rotating sample of people.

Dan 12:20

If you're not always asking the same people the questions uh month after month, we zoomed right into data, but I was interested in sort of the bigger picture of your work of like what do you do with this data once you have it? What is Elisa's role with respect to this data?

Elise Gould 12:36

So there's the data that's published, let's say, on jobs day, right? So there is an unemployment rate or or whatever there might be, there's gonna be some kind of a data point that comes out. And we're gonna sometimes just download that data directly from the Bureau of Labor Statistics, and I'm going to spend some time looking at the published data and see whether or not there's patterns, right? So any one number doesn't really mean anything to anyone. You know, what is 4.3% unemployment? Is that high? Is that low? And so try to put in some kind of perspective, you know, the number of jobs added. Is anything above zero good? Well, no, you know, you want to have job growth that's going to keep up with population growth, right? So, and and expanding. So, having some perspective on what these numbers mean in historical context or compared to other groups, let's say the black unemployment rate compared to the white unemployment rate, right? So that's an important comparison to have in mind or what is happening with young adults versus prime working age adults. So having 20 plus years experience looking at these data, have a sense of like what is good news, what is bad news, what are we concerned about, what are we gonna keep an eye on? And I think also there's a lot of volatility in the data. So not making too big of a deal about one given number any month because it could change. And so really paying attention to longer-term trends and not be alarmist about any of it. Okay, so that's one thing. You take the data as given, it comes out in a printed report. I look at that report, it's like a PDF, but I also am gonna download the flat files and make my own graphics about those data. So then I might have some analysis, I might put out some kind of a social media thread, talk to reporters, do that. So that's one way to think about it. We also take those data and maybe and post them with our own graphics. We have our own pages like Jobs Day analysis page. There's also microdata behind the data that produces the official unemployment rate. The microdata is the current population survey for that particular survey. Obviously, there's lots of surveys for that particular one. That's sort of our bread and butter. My work is based a lot on that. And when you download the microdata, then you can analyze it in all the cuts that you might want to do and all the time periods. And you know, maybe it's not just the published, you know, black unemployment rate. Maybe you actually want to do, you know, black non-Hispanic, or maybe you want to do, you know, any combination of interest. So the BLS does, let's say, college plus, but we want to separate out college from those with, you know, a graduate degree. And we would be able to do that in the microdata. What we have at EPI is we've created our own download site, and anyone can grab it and we show you how you can use the data. And we've created a consistent series of different variables. And so the variables' names sometimes change over time or the information that gets asked. There wasn't always, let's say, on race and ethnicity, there wasn't always an Asian American Pacific Islander series. And so sometimes the questions change. There's been changes in industries or occupations over time as the economy has shifted. So, to the best of our ability, we've created consistent series going back as far as we can, usually about 1973. And so people can grab our data. We have these new consistent series that we've used from the microdata that the Bureau of Labor Statistics produces that is publicly available for anyone. So microdata.epi.org, and you can download the data yourself and look at it. We've also done things with the data that is our own kind of special sauce on it. So the Bureau of Labor Statistics puts out average wages. Average wages is a really important, useful measure, but it's also important to know what's happening at the lower end of the wage distribution, or at the very middle, or at the top, or anywhere in between. So we create a decile measure. Again, that can be replicated, and we have the code, a GitHub page that has a code that also lets people know how to do that. And so then we can say, well, what is happening across the wage distribution in a way that other people might not be able to. And if you don't want to do the microdata, we also have a site where we publish all of the data, like these, the like the wage data that I just said. And that is our state of working American data library. Because just like the federal government, we believe that transparency is important. And so to be as transparent as possible, let people replicate what we have. And then if they just want to grab what we've done, and we make changes, right? Just like the Bureau of Labor Statistics or the Census, right? They make changes to how they want to measure things because the world changes, or we have more information, or we get better at things. In fact, the jobs report that you just mentioned that happened this week, they're doing something new. They're doing a new birth and death model that is going to get updated every month, which is a new procedure. They're trying to get more information. And so they make changes. Another big deal that happened this week was that they did the annual benchmarking process. So a lot is said about, oh, these revisions that happened, you know, they're correcting mistakes. They're not correcting mistakes. They're just getting the best data possible as soon as it's available to update the data. So there were these huge downward revisions that showed that 2025 job growth was actually much weaker than originally reported. And that was, you know, it's a pretty big deal. That made some headlines. And so what happened? Was there a mistake? No, there wasn't a mistake. They go back and benchmark payroll job growth against the official unemployment insurance tax records. That is the universe of job reporting. So it's not just a sample of some employers that are trying to make it now. This is the universe. They go back and benchmark it against the universe, and then they say, okay, we now have better information, and we're going to tell you what it is, and we're going to explain how that changes all the different measures for different industries or, you know, and every. And so that goes back a few years with that correction. So now you have a whole new series that's amazing. Anyway, I can the federal statistical agencies are the gold standard, and we rely on their data to be able to say almost anything. There is just no replacement for it.

Dan 18:13

That was great. You were talking about two strategies that are sort of essential and almost feel antithetical. The second one you mentioned was this sort of idea of revisions, too. And it I guess to some folks that might seem antithetical. You have the data, you reported the data, why would you ever go and change it? The first one you mentioned was transparency, and I feel like these two things are going hand in hand. We're as transparent as we can initially, but we can be even more transparent by offering these revisions. Can you talk to us a little bit about what role these revisions play in your process? Like when the government issues a revision like that, how does that affect the work that you are doing?

Elise Gould 18:57

So one of the things that makes Jobs Day so important is that it's extremely timely. So they're reporting the jobs day that came out this week, they're reporting on what happened in January. And it what happened with the payroll survey, it's what happened in the payroll period that includes the 12th of the month. And the same thing with the household survey. They're asking about that reference week, those two weeks, you know, what was happening in that period. And it's very current. So you're basically talking about something that happened less than a month ago. And now we're going to produce this amazing analysis that is vital for, as I said, policymakers, you know, Wall Street employers, everybody is looking at these data. There's nothing that's so timely as that. And so to be able to do that, there's a cost to maybe having to change things. And so you send the surveys out to employers, and not all of them get back to you in time for this report. And so then there's a revision that happens the next month. So in March, there'll be some revisions when they release the February. There'll be some revisions to the January data because they haven't gotten all the answers yet. And so there's a trade-off between having the quickest data and having the final data. I think it's just better to have information sooner, knowing that it might get revised somewhat. And the annual benchmarking, what are you going to wait a year and not have any up-to-date data? That trade-off, I think, is a balancing act. And I think it's really important. And to the extent it changes my work, we update the entire series. When they update the series, we pull the entire series again and we update that for the payroll survey. On the household survey side, they do an annual change as well with population controls. And those are probably going to be, you know, not small, maybe with a lot of changes in immigration policy. There could be, there could be significant changes in the population. That didn't come out this month. And I think that's probably because of staffing, but that will come out next month, and we'll have that. What they don't do on that data, which is like a Census Bureau of Labor Statistics joint project, is they don't go back in time and change the data. They just adjust it moving forwards. Um so that doesn't make the data as directly comparable um year to year, but it you can see the differences and usually that's not a big deal.

Dan 21:05

The initial round of data before it's revised, it's like everyone's sort of agreed, we know this is not quite right. Is there like a threshold for good enough in this kind of data?

Elise Gould 21:16

Well, you can look at the revisions that happen over time. It doesn't tend to move things. I mean, it did this summer, something that was positive became negative.

Dan 21:26

Right.

Elise Gould 21:26

So things can move, and people are already talking about whether or not the next month's data is going to get revised down because it was a larger payroll employment growth. But I don't have a guess as to what will happen there.

Dan 21:39

It's just interesting to me in this world where we've become so preoccupied with fact and what is true, given the circumstances that we're in, that actually a large part of the work that many people do involves working with data that we know to be not inaccurate, but not perfectly accurate.

Elise Gould 21:57

Oh, right. There's measurement error. Right. Some measurement error, right? So there could be some error in what people say, even. Let's say you had to answer a survey question. Maybe you don't know the answer. You maybe you're answering for another member of your household and you don't have the accurate information, and you're going to do the best you can. Or an employer, you know, misread something on a payroll sheet, or you know, I mean, there could be errors, right? There could be also just not answering questions. There's all sorts of item non-response that happens. And so then there's imputations that have to happen to make guesses. I'm not particularly bothered by that. I think that given the mission of these federal statistical agencies, they're trying to get the best data possible. There's nothing politically motivated behind the questions that they ask or the way that they aggregate the data.

Dan 22:42

And it also sounds like a lot of the work ends up being not just, as you point out, it's not just about a single metric or, you know, a single moment in time, but we're sort of looking over larger distances of time. Yes. And over more measures, which form a much bigger picture ultimately.

Elise Gould 22:58

Absolutely. I think the more measures is key, right? You're not going to take one measure and say, oh, that says everything. You know, the unemployment rate, again, is only capturing those people that are actively looking for work. A measure that I would prefer is something that's called the prime age employment to population ratio, because that's actually the share of the population, 2554, with a job. So it gets rid of the question of did you actively look for a job or not? It's saying, oh, how strong is the labor market? What is the share of people that you would think otherwise would be working? What's the share of them that's working?

Dan 23:29

That's interesting. Does the government not collect that?

Elise Gould 23:32

Well, they do. They do. It just doesn't wind up being the top line number. What's actually surprising right now is that that's been pretty strong. That number has been pretty strong in light of some other weaknesses. So I think that I'll be interested to see what happens with that as we get the population controls in the household survey next month.

Dan 23:47

Right, right. So there have been some changes at BLS. Can you talk to us a little bit about what's happened and maybe what the impact of those changes have been?

Elise Gould 23:56

There's two ways that we've seen some changes in the data in the last year. I mean, one is that we've had a couple of government shutdowns, right? And so there are temporary delays or permanent holes in the data. For instance, in October, we will never have data on the unemployment rate for October because there wasn't data collected for that.

Dan 24:17

Right.

Elise Gould 24:18

Right. And then I would say the the second is that there's just been insufficient funding and staffing. So uh there's been more limited collection of certain data series, uh, let's say data on inflation, both the producer price index and the consumer price index. There's just been less data. There's some data sets that are not being released and the same regularity that they had before. So job openings and labor turnover survey, state-level survey is now going to be released once a year. They're going to do one in July, and then they're not going to do one again until next year. So that is limiting the data. I know one of the things that we might be interested in looking at is what's been happening with federal employment, and they've now stripped out some of the demographic information from the federal employment data. And so you can't look at who is affected in all of the ways that you might want to in the data. There's been huge cuts to education data from the Department of Education, the National Center for Education Statistics. Again, those are huge staffing cuts. You don't have things like per pupil spending. You're not going to have the data that you've had before. That is a huge shift. And it's on top of, I would say, a Bureau of Labor Statistics that's already been under-resourced, you know, has been fighting attrition in some of the surveys, hasn't had the staffing or the resources to be able to carry out their mission in the way that I'm sure that they would like.

Dan 25:34

You said earlier, any of these changes, they're not politically motivated. Would someone like you be able to tell if there was manipulation in the data, if things were being manipulated or changed or removed for political reasons rather than just simply for staffing reasons? And I guess I should say by defunding something like BLS, by reducing the number of resources at Department of Education to a certain extent, there's political motivation behind that too. But the hope is that the numbers that they're that these places are producing remain as accurate as they can still be. Do you think that there's an opportunity to understand or or perceive any politically motivated changes to this data?

Elise Gould 26:16

I mean, I think it's a great question in this time. So the under-resourcing itself can lead to worse data because you don't have the resources. It's just a fact to be able to collect all the data, which is why they've discontinued some series to make sure that they can be as accurate as possible on the remaining series that they do have. That won't necessarily drive the results in a predictable way. I don't think that you can under resource and all of a sudden say inflation has come down because it's not that deliberate of an attempt to change the data in that way. Not that I've seen thus far, right? So this is where we are. What could change? The thing that I have, no matter how much I measure and I'll talk about, we have a data accountability dashboard that is trying to track some parallel measures to see whether or not we see changes. And we're doing that. There are some second best data sets that come out from the private sector, but uh they're not as good as what you can get from the federal statistical agencies. I have faith that individuals, civil servants will tell us, us, the general public, people who need to know, there would be whistleblowers that will tell us that the data has been compromised. People have worked their whole careers in those jobs because they believe in the integrity of the data. And if it gets compromised in a very deliberate way, not just from under-resourcing, then I think we'll know. But the under-resourcing is really important because, again, it's because they can't collect it, but it can cause political figures to convince the public that the data is not good because they're gonna say, well, look, they're changing, you know, month to month. They must not be doing a good job. Don't trust anything that they say. So you can undermine the faith in the under-resourcing, and then the message that gets sent can be very different. Right. Right. So you can have politically motivated messages on the same data. So for instance, I remember um at one point, I don't know how many years ago, whoever was president, you know, talking about how there's more jobs than ever before, making claims like that. It's like, you know, as it turns out, our population grows. That's not that interesting of a statistic to tell us that. Right. So, you know, it's how you might spin it and how you might talk about it.

Dan 28:29

Rachel sent me a video of Justin Wolfers earlier this week, and he didn't seem too worried about the new BLS commissioner. And I wonder if like, and he was sort of telling us not to worry so much about it. Do you have thoughts on the new BLS commissioner or the nominee anyway? I think it could be much worse.

Rachel 28:55

End of episode. No, I'm just kidding.

Dan 29:01

That is fair.

Rachel 29:03

One of the points Justin made in that video that I thought was pretty interesting. I hadn't thought about was like how it's easier or sexier to report on how the data is getting potentially compromised, or has, you know, people are worried the data is getting compromised. And we hear a lot about that, which is, you know, good for it to be on our radars. And then he made this point. He's like, actually, though, there's like a real win here in this new commissioner who is like a career statistician. But we're not, but no one's gonna report on that because no one wants to talk about career statisticians.

Elise Gould 29:37

Right, right. Yeah. I mean, clearly has expertise. It's not somebody coming in who's a hack, a political hack.

Dan 29:45

Yeah. I have a question here. Maybe we talked about this already. What steps do you see BLS civil servants taking to ensure the data is as accurate as possible, even given the short staffing? Do you see them making changes to the process at all, besides releasing things on Wednesdays instead of Fridays?

Elise Gould 30:02

That's just this month. It's not gonna keep happening. I have faith that's not gonna keep happening. I don't know. I think that we, you know, I think we've talked about that somewhat. I think that they are always doing research to improve their process. I mentioned the death and birth model, the benchmarking process. They're updating, always trying to be more transparent. When they want to have a new series, they have a research series first before it becomes, you know, a public new series. You know, there's always more information and feedback that happens and transparency. And I think that that's business as usual. They just don't have as many resources as they need.

Dan 30:37

Right. The other thing that has been a little bit covered is the removal of government data. Rachel sent me a story about this earlier, and I actually I think there is work being done at our alma mater on people kind of recovering or trying to maintain the amount of government data that that exists. Has this affected you at all? Do you anticipate making any changes to your process because of this risk?

Elise Gould 31:03

Right. So there's a couple of things that I think you're talking about at the same time. The first is that there is, you know, are data going to be collected that have been collected before? And we sort of talked about that, right? So there are data series that are being discontinued. But then there's a question of removing data. And I think that there have been in many different places archival efforts to download data that existed before they were discontinued using all sorts of tools to be able to do that and create uh new databases that should be available. That doesn't help moving forward with new data collection. But there are some groups that have archived data before it was uh before it disappeared. And I think that that's those are very important efforts. I don't think they've recovered everything. We don't know. And at the beginning of this administration, there were certain series that disappeared and then they came back. And so there was lack of clarity there, you know, in terms of what was going to happen with. I think I was looking at some series of the poverty rate, right? So where were those household poverty? You know, and they disappear and they came back. And I think there was just a lot of moving parts. The real shame is that we're not going to have more data collection on some of these series moving forward. Right. And you need that. You need a comprehensive measure of what is happening, you know, like a comprehensive measure of the count of people. You need to know what the unemployment rate is because policymakers, the Federal Reserve, state and local policymakers, they rely on that information to make policy. And if you want to have evidence-based policy, this is your evidence. You can't just go do a Google survey and ask people, you know, that's not going to be nationally represented. You're not going to get any of that information. And um you can technically it's not going to line up. And even some of the best uh private sector data out there. So let's say from Indeed or from ADP, you know, these large samples of job openings or you know, employment, they're not nationally representative. They mean something different. They provide information. You know, last week Challenger said we have huge amounts of job cuts coming, all reported job cuts. And it hasn't shown up yet in the official data. But, you know, it's information to be able to track, but it doesn't necessarily translate. It's sort of like, what did they say? The plural of anecdote is not data. You actually need it to be much bigger than that.

Dan 33:24

It really speaks to the role the federal government can play in this, which is sort of a completely unbiased, nonpartisan, really just focusing on collecting the data and making it available for you know smart people all over to kind of analyze it, look into it, glean insights from it. But that data source, collecting data at the source requires a very, very neutral agenda. And that's hard to come by beyond the federal government.

Rachel 33:55

That's right. You talked a lot about rationalizing data, making sense of the series over time and helping to so that you can actually look at consistent patterns. And first of all, I just want to call out this is so much in line with the heart of like why a lot of information architects become information architects. A lot of us are from like the library and information sciences field, and rationalizing data is like this calling, right? So I just want to kind of put a highlighter around this overlap in interests between like your world and our world. And then I was thinking there's probably like standard ways of rationalizing data in normal situations, like normal amounts of data you have to rationalize or blind spots you have to fill in. Has that changed recently? Like, have you had to develop like new techniques or new guardrails? Because I'm assuming the blind spots are getting bigger right now, based on what you're talking about, like data that we're just not collecting anymore.

Elise Gould 34:52

Right. So the Bureau of Labor Statistics will impute or Census Bureau will impute data when it doesn't exist for people. So let's say they answer almost all the questions, but they don't tell you what their wage is. You've told BLS that you have a full-time job and that you work in the information sector, whatever it is. Love that. And that you um, you know, have a college degree and you live in a particular state or you know, whatever information um that you have. The state actually doesn't come into the imputation, but they do a hot tech procedure, they do an imputation based on the information that you have, how old you are, how many years of experience, your education, uh, your race, and they'll put in an imputation, educated guess as to what your wage is. And if there's fewer people responding or there's fewer more holes in the data, then there's more data that has to be imputed. And so for work that I do, for instance, let's say looking at the union wage premium, how much more you get paid when you're in a union, all else equal, you know, people in similar jobs. And the imputation procedure that BLS uses doesn't actually impute based on the union variable. So it's important to remove people from our analysis that had imputed wages, because then you're gonna get you're sort of like you're gonna bias your results. And so it matters if more people have imputed information for the analysis that you can do because you're left with a smaller sample size.

Rachel 36:14

Yeah, because you don't want to impute on imputed information on imputed like you want you want to make sure that it's clean as possible. This is what does AI call this now? Synthetic data.

Dan 36:25

That's a is that the word they're using?

Rachel 36:27

I think so.

Dan 36:29

Data that it made up that it's now using.

Elise Gould 36:31

Yes. But in general, when I think about guardrails, and this would happen anytime, right? And it's not just a this year thing. When we're doing data analysis, what we try to do is we try to benchmark. So let's say I want to come up with an unemployment rate for a group that BLS doesn't have information on. I'm gonna make sure I can benchmark the overall. And I'm gonna benchmark maybe a subgroup that I can get closest to, or look at trends over time for similar groups and make sure that the data that I'm running doesn't have coding mistakes in it, right? Isn't misrepresenting what's going on? And so that benchmarking procedure, I think, is is a really key fact. And again, we post our code, the microdata and we post are the final results that we have for a lot of the data, and we update our procedures too. So we've changed. I mentioned our death siles, we've changed how we've done that. We noticed that sometimes the 40th and the 60th percentile move in a different direction than the 50th. And that's just, you know, that could be truth, but it could also just be some idiosyncrasies in in the data. And so we changed how we were smoothing our data to look at changes in wage growth across the distribution. We also, I think, you know, like many other people, sometimes make mistakes and we try to be transparent about that or get feedback and let people know there's a mistake, and then update that and and and correct and adjust our methodology over time.

Rachel 37:54

We haven't actually directly talked about this yet, but I'm intentionally leaving this vague because I want to see where you take this. What role does your work play in kind of the economy of misinformation that that is blossoming right now?

Elise Gould 38:07

I think that when the data come out, like for Jobs Day, we try to provide context, like we've talked about, so that any one number isn't blown out of proportion. Try to be very measured in how we talk about the data. And again, try to be transparent in any analysis that we do, a sort of post-production of government data.

Rachel 38:29

I'm getting the sense from your answer that it's like we're we're trying to be like a grounding function. And we're trying to keep data in the conversation and provide some view to evidence.

Elise Gould 38:42

Yes. Yeah, exactly right. So, so like the unemployment insurance claims data that comes out every Thursday morning, and sometimes that goes up, but sometimes it goes down. That went up a ton for federal workers. And you'll see reports of people saying, Oh, it went up by this amount. And I'll look at the data and I'll think, oh, that's interesting. It always does that this week of the year. You know, in January, that's a thing that happens in the third week or whatever it is. It's like, okay, let's put some perspective on that. And I understand that people are selling newspapers.

Rachel 39:12

Yeah, you just said the word perspective. And I think that's really like the keyword here. You know, the reason I was asking is because as we talk to people in different fields with different focuses, we're starting to see how different organizations or even different individuals want to play different roles. Some organizations play grounding roles in providing perspective like yours. Some take an advocacy role for kind of quote unquote fighting back. Some take a role for working on reframing, you know, like if there's like a really funky reframing of data coming out, some organizations are like, no, my job is to actually bring this other frame that I feel is much more valuable and useful and truthful and really like get that in front of the same audience. I'm building like a Game of Thrones map in my head of like where where all the organizations are we've talked to.

Elise Gould 40:03

I I think there's overlapping, right? Because I do think that we're trying to provide grounding and translation of the data in a meaningful way. But I also think it's the questions we ask, right? So I mentioned that average wages mean something, but you kind of want to know what's happening for low at moderate wage workers. So you're going to make sure that you can cut the data in that way so you can tell a story that isn't just about the top-line number, but actually tells about people's lived experiences that aren't just uh who you might see in the average or who you might think about.

Dan 40:36

I like that question, Rachel. And I I think it gets at something that we have been observing, which is, and I guess I would sort of frame this as is your job any different if we lived in a perfect world and there was no disinformation, right? And I I get the feeling that no, you just kind of keep doing at least we keep doing what she's doing, of trying to produce this grounding.

Elise Gould 40:57

Yeah, I think that's right. And this isn't, I mean, in some ways we're in unprecedented times and the attacks on the federal data are are larger. You know, the attacks on federal workers are larger than we've seen. But for decades, people have questioned when they don't the data doesn't say what they like.

Dan 41:14

Right.

Elise Gould 41:15

That the data's cooked or whatever. That's you know, people have said that. It's just that we're in a it feels particularly acute right now.

Dan 41:22

Right. And I mean, this is in almost every expert that we've talked to, their very field is under attack in some ways. Because of the complexity of the things that you're dealing with, it's sort of easy to use that as a smokescreen for, well, clearly they get messing with the the messaging or messing with the numbers that are coming out. In these unprecedented times, what gives you hope?

Elise Gould 41:46

Again, I think that my belief in the civil servants is utmost. I believe we will learn if something is being very obviously tampered with. So I think that those civil servants are strongly motivated by their mission. I think that that has come to the fore, even though government workers are being wildly underappreciated. I think that there is some understanding of the importance of what they're doing. And obviously, it's not just about the data, it's about the services that federal workers provide. Some might think that the data does not rise to the top of that list. There are obviously a lot of other things going on in my work world. Data is pretty important. I think that the economics profession has gotten more based on the real world and has cared more about empirical testing in a way, it's become less theoretical. So I think that's an optimistic take. And I think that more data and the ability to analyze that data is more broadly shared. So more people are able to look at the data or look at the microdata themselves and being able to do their own analysis and information. And I think that that is great. And I think the public data that we still have provides the kind of evidence to inform, you know, even let's say at the local level, you know, changes in the minimum wage and see what that means for low-wage workers and being able to use the data to make those kinds of tax analysis. I think there's also federal data that's used for telling us about vast inequality. And so you have people out there telling us using the tax data, and that tax data is very hard to get a hold of, right? There's all sorts of confidentiality to sign away everything to be able to get access to that kind of data. And they do, they take that confidentiality very seriously. So I think any tampering with data or using that for nefarious purposes, you know, any concerns about that, or as a researcher, you have to be very careful. And I think that when you look at the Bureau of Labor Statistics or those statistical agencies, the sort of they have a huge return on in investment for their size. What they've done has been incredibly useful and valuable for all of us as researchers and for anyone who relies on that data.

Dan 43:53

That was great. At least, was there anything that we didn't talk about that you wanted to talk about?

Elise Gould 43:58

No, not really. I mean, I think that when we're We're talking about whether or not we can trust data. I think it's very important to look at the data sources. So when anyone, this administration or other commentators talk about data and they don't tell you where they got the data from, it can be very hard to fact check and to have a good understanding. So I think it's important for everybody to always take a look, try to find the source article or make sure that the graph actually has a source at the bottom of it. So that when somebody says that, you know, incomes have never been higher, you can look and say, oh, it's that, is that really true? Where are they getting that information from? And being able to back that out. And the less transparent a commentator or a politician is, it calls more into question what it is that they might be saying. So that's, I think, on all of us to go and look and see and try to find it and know what we can trust and repeat.

Dan 44:57

When you and I first started talking to Elise, I think we both had a different idea as to what direction that conversation was going to go. I was really curious, as I said, about the messaging around economics and the economy, the US economy. And we ended up digging into what was probably a more interesting topic, which is the data behind understanding the economy. And what I liked about that was that it was very You were talking about a system, right? A system. And that's sort of what you and I think are trying to do is look at the systems of information that exist and how they are vulnerable to disinformation practices.

Rachel 45:35

Yeah. The economy is a really fun system to explore because it touches every single aspect of our lives. But we don't really talk about it that often as a thing to unpack and pull apart.

Dan 45:47

Right. And in Elisa's perspective, it's great because she is like all in on the data. Like that's her thing. What was your lens?

Rachel 45:55

I had a couple lenses, and I'm choosing the one that interests me the most, which is human versus system integrity.

Dan 46:04

Oh, love it.

Rachel 46:05

At first I was just thinking about integrity. We talked a lot about data integrity with Elise. But then she was talking about like her faith in individuals to flag where integrity is crumbling and her faith in whistleblowers. That's what brought me to this lens, which asks us, you know, how much does the system rely on individuals to report data integrity issues? Does the system have mechanisms for analyzing its own data integrity? I ask this with this like lilt at the end of my voice because I've never thought about this. For better or worse, there are lots of applications of like machine learning and AI tools. And I was sitting here thinking, like, man, what if a system could flag places where integrity seems like it's being breached or is becoming questionable rather than relying on human whistleblowers to comb through things in a dark closet and like highlight when something seems suspicious? Right. Now, the cynic in me is like, there's a reason systems don't have mechanisms for this. Yeah. But setting that aside, thinking in good faith, when you're trying to examine a system from multiple angles, I do think this is a really interesting angle for like what are the mechanisms for noticing or highlighting potential areas where there are data integrity issues.

Dan 47:25

I'm gonna ask a Rachel style question. What do you mean by integrity?

Rachel 47:30

Ooh, that's a rude question. That is Rachel style. I was really thinking about this in the context of our conversation with Elise, which is places where you realize you're having to do a lot of rationalizing or imputing and making a bunch of assumptions. That's a very specific case where I could see like there's a threshold of the amount of imputation you're having to do that could be thin ice. I could see where there's like inconsistencies that are suspicious. I'm sitting here thinking, like, man, now I want to talk to like a data analyst or someone whose like whole job is to like look for patterns and like flag like suspicious patterns.

Dan 48:08

Yes.

Rachel 48:09

Where the story maybe isn't, there's no there there.

Dan 48:12

Yes.

Rachel 48:12

This lens is probably pretty open. It could be literal, like looking at data patterns and seeing where the data seems suspicious, like something might be missing, or there's been like some sort of imputation error. Or you could take this more philosophically on the integrity side of like where is maybe it's not just about data integrity, maybe like system integrity, like where is the system upholding expectations, where is the system failing expectations, and whose job is it? Are there mechanisms in place to flag that? Are we relying on individual humans and individual whistleblowers to notice those things and bring it to attention?

Dan 48:44

This really jumped into my head when you were talking about could AI play a role in this? And I was sort of thinking, what do humans bring to the table? And it's always something like larger context, right? Which might be lost. But there's also a impetus, a motivation, a strongly held belief, I think, that sound data is important.

Rachel 49:10

Yeah.

Dan 49:10

And when I say it out loud, it sounds simple, but I think knowing what role the federal government can play in data collection across the board and why, and it is so powerful to have data collection that is honest and that strives to build a truthful picture. And that at the heart of that is yes, data collection mechanisms, but also the people who want that to be true as well, want that to be real.

Rachel 49:38

Yeah. I think another aspect of this lens that is getting clearer to me as we speak now is with any system, how much are you relying on humans to be the source of integrity, like an individual with integrity versus an entire system that has been designed to maintain its integrity?

Dan 49:58

Right.

Rachel 49:58

I think that was the original thought I had when we were talking about like whistleblowers and faith in individuals who would flag if the data were getting really stinky. I'm like, that's cool. And I'm glad that we have faith in those people. And one of the things we talked about from the beginning of this podcast is when we think about misinformation and we talk about like mainstream solutions to addressing misinformation are behaviors that we expect individuals to learn and practice rather than changing parts about the system. I think this is maybe just a different flavor of that.

Dan 50:27

Yeah. Well, it gets back to, I thought you were gonna say when we talk about resilience too. Is a system considered resilient if it doesn't have these mechanisms in place that it does everything it possibly can to try and maintain data integrity? Yes, there's gonna be other outside forces looking at that data and providing its own pressure for keeping the data integrity. But I think there's an opportunity for the system itself to do more, to not simply count on those external forces.

Rachel 50:59

Yeah.

Dan 51:00

I think a really good example of this, and we've probably used this before, not so much with quantitative data, but with community comments on things like threads.

Rachel 51:10

Yeah.

Dan 51:10

Where people can spew misinformation on threads, but then the community can kind of come together and add a note to a stupid thing that someone said or falsehood that someone said, and provide some clarity. All threads is doing is providing the means for doing that.

Rachel 51:26

Yeah.

Dan 51:27

But that feels like it doesn't even meet the threshold of the lens that you're setting out here, that threads itself, meta itself, needs to do more for data integrity.

Rachel 51:40

Yeah. And I think of a lot of these lenses as like a thing an individual designer could take to a situation that they are in and ask, why aren't we doing this? Like ask to themselves, like, why why isn't this company incentivizing this kind of behavior?

Dan 51:55

Aaron Powell You and I, I think, have developed the capacity to walk into pretty much any room and ask a hard question of the people in there. And maybe what you're saying is that these lenses are meant to empower others to do that too.

Rachel 52:12

Yeah. So in the threads example, you could say, like, yes, okay, here's one small mechanism for integrity. We don't see other more powerful system integrity mechanisms. Why not?

Dan 52:23

Yeah. Why did we stop there?

Rachel 52:25

Yeah, why did we stop there?

Dan 52:26

Yeah.

Rachel 52:27

And the answer to that tells you a lot about a lot of things.

Dan 52:30

Yeah.

Rachel 52:30

Okay, that's enough of mine. What was your lens?

Dan 52:33

I really want to talk about this, but I don't remember the context in which this came up. I follow a newsletter slash Instagram account called Philosophy Minis. And the guy will sort of post a little philosophical vignette. And I'm very inspired by this. I find them very compelling, but I'm also very inspired by the format and the approach that he takes. And he told the story of the lost horse, which I will do a very bad job of retelling right now, and I will just sort of try and make it quick. But the idea is that a farmer, you know, is in a small village and the horse runs away. His horse runs away. And the villagers come to him and say, Oh, what bad news. And the farmer says, Maybe. And the next day the horse comes back and it's got six wild horses in tow. And the villagers say, What good luck. And the farmer says, Maybe. The next day, the farmer's son is trying to tame one of the wild horses and gets thrown from the wild horse and breaks his leg. And the villagers say, Oh, what bad luck. And the farmer says, Maybe. And it keeps going like that for several iterations. And I find it very compelling, especially in this day and age where we are confronted with, yes, a lot of bad news, but also a lot of misinformation, where we do have to sort of take a moment and we can't jump to any conclusions based on things that we've heard because they may not be real.

Rachel 53:59

Yeah.

Dan 54:00

So I can't remember what it was and what Elise was saying, but I think there was this notion that the numbers themselves don't convey good news or bad news, that the system can't presume that one thing is good and one thing is bad, and that we need to help users come to that conclusion, whatever conclusion they need to come to themselves.

Rachel 54:23

Amazing. So why should we care about this lens?

Dan 54:27

I think we should care about this lens because a lot of the data systems that we design, I know in the work that I do for, say, designing dashboards or things like that, there is a real push to make these things light up, right? And sort of telegraph good news or bad news. And I think it's potentially dangerous to make some of these assumptions about what users want to see.

Rachel 54:58

This feels like a more specific version, which so I think it's good, of some of the framing lenses we've talked about.

Dan 55:05

Oh, yeah, good.

Rachel 55:06

Because good and bad news is a frame. And it's maybe not a narrative frame, but it is like a a slant. Right. Right. This lens is really interesting to me because I feel like this is a very stark reversal from what most designers are asked to do.

Dan 55:22

Right.

Rachel 55:23

I don't think we often interrupt the assumption that there is a correct narrative in terms of is this good or bad that we're trying to communicate.

Dan 55:33

I mean, I've been thinking a lot about data displays because several of my clients are dealing with it right now. And I I feel like it's important to highlight outliers. But the system itself, certainly the systems that we design, probably don't have enough context or understanding to know whether it's good news or bad news. And also, like the farmer who took a much, much longer view of these things, something that looks bad today may or not be bad kind of in the long term. Yeah. And if we frame something as bad news based on the rules that we have today, that may trigger certain actions or certain behaviors or certain reactions that are inappropriate that don't lead to the best decisions either. I want to be really clear. I think it's important for us to acknowledge that not every lens is going to be meaningful in every situation.

Rachel 56:25

Yeah.

Dan 56:25

You know what I mean?

Rachel 56:26

Before you get too far with this, now that you had a chance to talk and my brain had a chance to work, I can think of so many applications for this lens. Yeah. I think it's a reflection of that I work in tech right now that AI is very much on my mind every single day. I'm thinking about when we're designing AI interactions and trying to improve the outputs of LLMs. This is actually a hugely important lens. Yeah. And you said something about like having enough context to know whether or not this truly is good or bad. I had this really funny interaction with an AI chat in my financial planning software.

Dan 57:03

What did you just say to me?

Rachel 57:05

This was a personal situation. And I was gaming out some alternative scenarios, doing some financial planning, and the AI chat kind of landed on this outcome that it framed as like being really negative, which with my human context, I was like, oh, that thing you're pointing out is like a potential disaster is really not a big deal. And it actually was in direct conflict with this other piece of data you just told me, both of which are true. And so I got really annoyed. And I was like, you know, in the human language, I was like, how are you like, what? Like this doesn't make sense. Why are you doing this? And then it immediately was like, oh, you know what? Now that you point that out, you're right. This number isn't so weird, actually, when you do look at it in the context of this number. I thought you just really wanted to focus on this other number. And I was like, okay, we're done here. But I can think of this in all sorts of situations where like a human is trying to get a little bit of assistance thinking through a situation, and that AI not having the right context, which is usually always the problem, then frames good or bad with missing context. And so I think there's this is a very powerful lens.

Dan 58:08

Aaron Powell What we know about LLMs or certainly the people working on LLMs is that I think they're trying to crack the nut that is the lack of context, right? We know they're working on that. And as I'm talking through this and hearing you talk about it, it's almost like I want the system to give the user permission to say, maybe.

Rachel 58:28

Yes. Oh, I love that.

Dan 58:30

I don't know what that 100% looks like, but I like your example a lot, right? The AI wants to paint things as gloom and doom. And you're like, well, hold your horses, literally. Yeah. That may not be the actual case.

Rachel 58:41

Aaron Powell Or the other really annoying thing that most chatbots do, which is if you work with an AI chat in a work setting like Claude or Gemini or whatever, until you've really trained them out of it, they're so supportive and they're so like they have so much conviction that you are the best source of everything. And I think a good response to that is maybe.

Dan 59:05

Actually, now that you say that, I've been working on a project with Claude, and it's been incredibly helpful in a lot of ways. But I I had this observation today, so I put it into Claude, and it said, This observation is the most important observation you have made this entire conversation. And I was like, slow your roll. First of all, I've made a lot of really good observations just in the last 10 minutes. So F you and Yeah, it just felt like that felt like overkill. I was like, I got it.

Rachel 59:37

Extremely charming. Yeah. Yeah. So like extremely charming, extremely deprecating, those are both frames that lack a lot of context usually.

Dan 59:44

Right.

Rachel 59:45

Cool. Great lens.

Dan 59:52

And that was Unchecked. Thanks so much for listening. We really want to hear from you. If you've got ideas for topics or guests or stories, drop us a line at unchecked at curious-squid.com. If you made use of the lenses that we described today in your practice, we want to hear about that too. Hey, check the show notes for any of the links that we talked about today. And it would really mean a lot to us if you shared this episode with a friend and rated and reviewed us on your favorite podcast platform. Thank you.