 Welcome to economics and beyond. I'm Rob Johnson president of the Institute for New Economic Thinking. I'm here today with Kathy O'Neill Who has founded an organization called O'Neill risk consulting and Algorithmic audit She is author of the book weapons of Math destruction that's MATH not MA double S and And she's working on a book about shame She has looked through her Blog called math babe at many many different phenomena and trying to understand how technology quantitative analysis can be used for both good and evil and I want to thank you for joining me today Kathy here at the Institute for New Economic Thinking The religion about evidence-based analysis needs to be explored and I think you can help us do that I'm really glad to be here So we're talking here in June of 2020 Where the COVID-19 panic is all around The pandemic has unleashed and unmasked many things and many concerns about society From the legitimacy of our health systems quite obviously to the legitimacy of how we're governed To trust in expertise and science And I'm curious What you see as you look out at the world as a mother of three sons What what causes you heartache? What do you See that should be done that isn't being done and How I guess in the end how are we gonna get out of this place Eric burden and the animals have a song We got to get out of this place If it's the last thing we ever do Girl, there's a better life for me and you so why don't we start with What where are you concerned in and how do we get out of this place? Wow? I like how you start with the softball questions um Yeah, I mean I have I actually wrote a list of things. I'm worried about the other day um And as he said I have three kids so they're top of mind My oldest is in college. My second is entering college, you know, they're they're They're gonna they're gonna get their degrees and what kind of economic environment. I have no idea um at the same time They're very very privileged. They've had very good educations So I worry about um deepening inequality. I know that the jobs report Even though it looked better last week Did not look better for people of color um, I think I I wrote an article from bloomberg the other day that Makes the causal argument not approved because you can almost never prove uh causal claims with Data, I mean you all you can make the argument and I I did that inequality mass incarceration and racism are actually Causing the epidemic and I know that sounds weird because we're so used to thinking of Uh, co vid as a biological phenomenon, but it's not only that it's also It's also a social. It's a function of our social systems and to the extent that we We need certain people vulnerable whether it's The the immigrant women of color working in multiple nursing homes because they can't get enough Pay to work in only one job And they are not giving enough PPE. So that's one of the reasons that our nursing homes are so vulnerable to co vid or whether it's because um The service workers are all living in much more cramped Circumstances multigenerally generationally And in the context of places that have a lot of mass incarceration. There's a lot going on there But the point is that those Those aspects of our culture make us more vulnerable to pandemics So I think about that another story outside of our country is singapore. I wrote early on how um I was I was interested in the singapore story because it was about contact tracing that that was how it was framed in the western media That they had this wonderfully successful technology that was tracing people It turns out that that's not the story heard an interview by the um prime minister or the president I forget What it's called of singapore which explained that it was in fact No, we interviewed people who were sick and we found out who they talked to and we called those people It was just old-fashioned detective work. I mean use the technology to tell us what it wasn't It wasn't bluetooth technology like we're trying to make it out to be And because of that Unbelievable effort they really tamped down on the on early cases of co vid and that was a success story Flash forward to like two months later, which was like a few weeks ago And singapore had a major problem of co vid cases that were uncontrolled Not in that original situation contacts but in their like guest worker dormitories where they have 20 guest worker men to a room from Bangladesh Um, and they don't have they don't have social distancing. They don't have contact tracing. They don't you know It was a completely different system And it caused an outbreak So there's there's lots of evidence to this effect But to answer your question to to circle back to your question What do I worry about I worry That we are so short-sighted that we won't realize that we are actually all in this together That that pandemics are going to happen more and more. We're more and more interconnected It's just one example of things that are going wrong But it is a sort of a powerful one this idea of the next pandemic And if we wanted to immunize ourselves as it were If we wanted to make ourselves less vulnerable as a country What we should be doing is addressing these things like inequality like mass incarceration like racism That seem impossibly difficult and they are very difficult But are actually 100 required in order to make ourselves healthier Yeah, I I worked for a little bit on a film As a producer it's called amazing grace about aretha franklin And I got involved in visiting prisons Uh in particular went to sing sing and I talked to a number of clergymen who were working there and Just by by heightening My awareness of that environment I was filled with what I'll call like a haunted dream When the COVID virus came out because I can't imagine how people Who are incarcerated are going to be treated like humans And that I I just can imagine COVID going like wildfire in the prisons I saw you had written a piece on racism and mass incarceration For bloomberg recently. I think it was about the fourth of june. I'm curious what What have you learned as you study? What was the group measure of justice I had in my notes? Uh, what what is what is being illuminated there? Yeah, they did a study and that was the piece I was referring to the causal argument That mass incarceration causes pandemics. So measures for justice did a study in the Milwaukee area And they looked at neighborhoods and the extent to which that neighborhood had been affected by mass incarceration So like the rated which um people in that neighborhood were Were incarcerated um, and what they found was that that Statistically was a better signal for COVID cases than any other signal that they could come up with like including race including poverty including Unemployment rates. So if you're looking for the single strongest indicator That people in this neighborhood aren't going to get sick. It is how how much are they imprisoned? And it's not simply and there's another by the way, there's there's other studies that I've been Made aware of since I wrote that there's many multiple studies that are studying this kind of thing some of them do simulations Of like the actual interaction between people who are leaving prison entering prison the prison guards sort of trying to track and estimate synthetically But relatively realistically how fast COVID can will spread in those conditions This is not what the Milwaukee study was doing. Milwaukee study was was making the case that that the neighborhoods in question have been made fragile and weak After for from years and years of of mass incarceration that it's not something that just happened because oh people who left prison came home and got us sick That's of of course also happening. That's a secondary effect though What we're talking about is that the neighborhoods themselves can be seen as healthy or sick And that the COVID has exposed if you will the the the underlying level of health of our society And so when we see Certain neighborhoods getting decimated You know not only because they're the high-risk patients as in the the nursing homes, but for other reasons We should think of that not not as like oh those people had Cobra morbidities and they should have taken care of themselves better No, we should see that as we should have taken care of those people That that is on us Yeah, yeah, yeah, there's an old uh Kind of thing that Is done in economics Called economic justice and without going on too long If you're paid more than your productivity, you're being subsidized If you're paid less than your productivity You're being exploited if they're right in line It's fair and that's called economic justice But there's a problem with this which is Let's say all that stuff is true. I'm using Uh, I'll I'll call it a maintained hypothesis But what if your productivity Doesn't produce the kind of compensation that allows you to have health and nutrition And a family and shelter and all the basic things Well the tendency in economics is to say well It's that person's fault Yeah, but when things like health systems when things like educational systems Are our social products? We are all Complicit In economic justice not being human justice You can't you can't just attribute it to the individual And so yeah, I just I think I think it's a dreadful escape hatch For people who don't want to Address collective responsibility You uh, I had a very interesting, uh There's a group up at mit that was my undergraduate, uh college was working on uh Basically biosecurity and One of the things they've been very concerned about Is what is it going to mean in terms of health risk? When the colleges are short on cash flow and want to reopen Which creates How would I say less Distancing and I don't like the word social distancing. I I think it's physical distancing We can remain socially integrated so I'll always Challenge that phrase. But but my point my point is if you have everybody go back to college In order to serve what I'll call the cash flow machine that sustains the university That's got to be dangerous For those people meaning both the faculty and the students. Oh, I've been thinking about this very very consistently Thank you for bringing it up. I mean Number one. I have two college sons um number two My mother is a professor My mother is a 78 year old computer science professor at umas boston Wow, and you could be sure that she's not willing to go in person to teach at umas boston anytime soon um And so the question becomes very clearly are well, are you going to fire her? Are you going to force her to retire? Are you going to pay her without having her teach? There's a certain like web Web programming course that only she ever teaches. She's very very nerdy by the way. Um, she's wonderful. You should have her on too um, and uh at the same time like I think the way I think of it is that there's kind of three tiers of colleges And there's like that top tier which is to say like harvard Where they have well first of all, they have an endowment that they could be spending on like making the dorms much less dense If they really wanted to they could just you know rent out all of cambridge if they felt like it They claim that their endowments are tied up in other in other Other funds which which is kind of ironic since you know, you think of you think of an endowment as a buffer for hard times But I guess not Um, but really I would say that beyond their endowment which is an asset of theirs And of course the real estate that they own which is a huge asset of theirs Their faculty is their largest asset And they really cannot afford to put their faculty at risk And their faculty is old by the way. I mean and harvard is kind of an extreme outlier on the spectrum But it's not really in the sense that what we've had over the last 40 years is we've had The new faculty coming in at at older ages, but you know young people but you know slightly older and older as we have um, you know At least in mathematics for example had to take more and more postdocs before before we could land a professorship So you have the professors coming in they keep on staying young maybe getting a little bit older And in the meantime, you just the faculty don't retire the way they used to so you have the faculty remaining as long as possible So the average age of faculty As in the lead institute is is very high. Um, so they are very much at risk and then you have the middle tier Um colleges that I think are probably the most at risk Which are these these colleges without an endowment That really truly rely as you say on the tuition and they simply cannot function without it And so for them, this is an existential question. If they don't open they will have to close forever And there's going to be a lot of closures But there's also going to be a lot as you point out of Schools that are doing taking risk taking risk with both their students and their faculty Because they again it's essential and then there's a third tier of like Colleges that are commuter colleges. Um, which Um, which will have a little bit of an easier time of it I think because they will say hey what we are offering you is mostly Education not this sort of holistic social opportunistic type of um experience and then they they will be able to um segue into I think online classes more or smaller classes or you know And one part of that is because they they they will less they will have fewer students who actually live on campus So they'll have fewer sort of litigious Fears of of like parents who get um who sue because their kids got sick or brought home some sickness or something like that So I really think there are three different problems to solve here And my guess is that heart the the top elite schools will just um They'll do whatever they need to do. They'll have to spend some of their endowment, but they'll survive They'll protect their faculty the middle schools. I think will be decimated I know people who sit on the boards of many of what I'll call the liberal arts schools Like my son went to Pomona Friend of mine is involved with Oberlin Bard college in that tier And they seem to be under a great deal of stress My mother's school by the U. S. Boston is definitely in the third tier their commuter And she will be teaching online. I'm very sure of it But yeah, I worry my my son my son's well my yeah, my son's go to middle tier schools. So I'm worried when uh When you talk about that I've seen something particularly in the Boston globe uh about What do you mean we're all going to stay home and you're going to charge me full tuition Yeah, uh is is kind of the meta question Yeah, I actually wrote a blog uh wrote a Bloomberg call about this. I called it a game of chicken My theory is that most of these schools aren't going to really tell the truth to the parents and to the students until after the first big chunk of tuition has been paid August 1st and And or they're that's the game of chicken. They want that tuition deposit And the students want to know well, what's the situation? I don't want to come if you're not offering me this holistic experience And so I'm the question the big question for me is like are the students going to balk at paying that without knowing what they're getting into Or the parents or are is there going to be a negotiation on price? And I actually think there will be a negotiation on price for many of these schools I think they're going to figure out like we're going to lower the price to try to keep you here and it's going to suck for our like budget, but We're going to have to we're going to have to suck it up try to survive hold our breath for a year Yeah, it's I since I've seen a lot of articles in the united kingdom as well that are Portending this kind of stress and this tension Between what you might call the family's paying tuition and the services that can be provided in light of the pandemic. Yeah the What have in going back to prisons for a minute. Yeah Um Is there a movement afoot to let more people out of prison? Is there data now on on what's happening there? I'm really glad you asked that it's There are a few different suggestions and some of them are very problematic and sort of go to the very heart of my first book weapons of mass destruction Because they're trying to use what they're calling risk scores to decide who to let out This was you know, I'm not sure how many people actually got let out under this this proposed system But what I find so fascinating about it almost a philosophical level is what they mean by risk What do they mean by risk? Of course when you hear people are being let out depending on their risk level of prison because of covid you automatically or you'd be you'd be excused for automatically going to the assumption that High risk means people who will probably need hospitalization if they get sick But that's not at all what they mean what they mean is um crime risk And crime risk is of that's what they call it That's the term of art. It is not at all a crime risk. By the way, I'm going to tell you I'm going to correct the record as I always do because it's so problematic. It's so racist in particular Crime risk scores are algorithms and I'm an I'm an algorithm expert and And what I mean by that is I can actually I have like explain algorithms in a way that is understandable Whereas people who claim to be algorithmic experts Love to make make it sound like it's impossible to understand. It's not impossible to understand I'm going to explain this to your listeners really quickly algorithms Propagate the past they look for patterns in the past and then they Predict that the the pattern will repeat and in particular they look for patterns of initial conditions that led to success Where the success has to be very explicitly defined and they say people like you Were successful other people like you were not successful and what they do when they are Actually trying to predict for a new person they look at that person They say oh We're going to look for people who are close to you and there has to be some definition of what it means to be like Someone else or so so what does it mean to be close? And then we're going to look at examples of people that were like you were like these initial conditions Um and what happened to them? Were they successful or not if you know seven out of ten with them were successful Then we're going to give you a sign you a probability of seventy percent that you will be successful But just i'm just thinking it i'm just repeating myself to say that what all an algorithm does Is it looks for Similar type of examples in the past what which of these led to success or not? And so now we're going to tell you what a crime risk score actually does a crime risk score predicts who will be arrested People like you will be arrested other people like you will not be arrested Okay, so just thought experiment. What do you think is predictive of being arrested in this country? And the answers are Poverty mental health status Addiction status and race Maybe race first they don't actually ask the race In the questionnaires that these crime risk scores are using They only use proxies for race. They use things like did you graduate from high school? Were you suspended in high school? Do you have a job? Have you been married? Do you have gang friends and by the way gang friends? Is is a construction that is completely Defined by the police department of who used who's a gang member and who's not and you might expect and you will Be right that it is a very racialized Then they do ask the question. Do you have a mental health problem? Have you ever been addicted to drugs? Basically in all sorts of poverty proxies, so they're they're basically profiling somebody For their likelihood of being profiled by the police or otherwise arrested by the police So that's what a crime risk score is measuring and I just wanted Reiterate that this is not Actually measuring a criminality the inherent criminality of a human like unless you actually think that somebody who has untreated mental health problems is a criminal Then okay that we'll have another discussion But what we're really measuring and we're doing a pretty good job of it actually this is not an inaccurate algorithm we're doing a pretty good job of predicting what the police will do But not what the criminality of a human being is And yet those crime risk scores are used to sentence people to longer in prison or not To give people parole or to deny parole and to keep people incarcerated pre-trial So these crime risk scores are are used as if they are in fact measuring the criminality of somebody Let somebody with a higher score deserves to be longer in prison Whereas I would argue that somebody with a higher score deserves more support Like for example mental health support or addiction treatment It's really a reflection of All of our problems and all of the problems that we've been seeing in protests this the last couple weeks All of those problems that we have been just shifting on to the police force as if they would be better Dealing with mental health problems than like social workers or actual therapists We've just been giving that problem to the police because we don't want to deal with it as a society What's happened is that they've been embedded in these algorithms and these algorithms Believe it or not are the ones that are deciding who gets to go home during covid from prison Who gets to go home as people who are low risk? Not the people that are high risk because that again, there's a double There's a double meaning here and the high risk scores Are interpreted as people who are dangerous to society Um, so those people who are blacker poorer at worse mental health and worse addiction status Are being kept in prison Um, even if they're old by the way, one of the things that absolutely Enrages me about this all of it enrages me, but one of the things that enrages me is that they're static So a 17 year old who had an impulsive episode it goes to prison for it and is kept in prison for 40 years When they're when they're 57 their score is the same as when they were 17 Even though we we have much better information And know that 57 year olds just do not commit crimes at the same rate as 17 year olds It's just completely ridiculous, but the the point i'm trying to make is It is in fact not at all measuring who is at higher risk for covid if we were to do that If i if i were to ask by the a prison system to develop a covid risk score to decide who gets to leave prison It would actually be almost the opposite Of what i just described because as we know people of color are at much higher risk Of covid and have higher mortality rates at least that's what the data is telling us Although we can talk another conversation about covid data and how much we can't trust it Um, which is a lot, but the point is that these risk scores. It's like it's a philosophical nightmare to say to hear that oh, we're we're letting people out of prison based on risk and i'm like what risk The wrong risk the risk that they will be they will be harassed And arrested by cops yeah, the uh, I i'm taking notes here listening to your uh Your different perspectives and try to relate it to uh The first is the notion What sti sti now i understand you have a phd in mathematics from harvard yourself So you're not outside looking in you you've been down into The boiler room of how mathematics works in and What economists or econometricians call ergodic stability? Meaning when you analyze something With data you presume a stable structure whereby the past Is prologue That you can count on that And I think in some of the natural sciences One can employ such a thing like when you drop a tennis ball off a 15-story building Adjusting for the wind you get how many seconds it takes to hit the ground and it's Reasonably stable but You worked I worked in the hedge fund industry. I worked with george soros's firm for years and and you worked With de shaw I recall The idea that statistical analysis of subjective psychological expectations embodied in price Is ergodically stable is absurd It's used as a parable So that you can say government doesn't need to supervise Analyze constrain or what have you so that people Can do what they want in the financial sector and then Which amount I'll give all kinds of campaign contributions and things to make sure they Have that unbridled and unrestricted status But the idea that it's justified because Value at risk and all these things are stable And predictable Is not science. It's scientism And what they call in economics radical uncertainty meaning The outcomes are unknowable and the probabilities are unassignable Is how would I say uh It's hard for people in the quantitative science who want to show that their methods do help To accept that And but it's a kind of form of demagoguery. It's a kind of form of Of what you might call pretending and know what you don't And I can see that In this kind of social dynamics that people At times may Yorn for certainty when they're anxious But it doesn't mean you can provide them that certainty Because the the systems are not structurally stable, particularly when they involve subjective human psychology and object and subject being How do you say the same thing? So I don't And then I you know taking this to the levels of poverty mental health addiction and race What we do in race is quite inhuman But when society Creates poverty. They're creating the risk And not wanting to pay the price of making society safer But anyway, I've been talking for a bit here, but but I just some I'm trying to bounce this off of what my economics audience is familiar with and But you're really cutting right into it And let me respond to you I mean, I I have a name for this in my book. I refer to it as the authority of the inscrutable Which we which we use as quants. I mean, I saw that during the crisis when I worked at hedge fund I did to yasha and then by the way, I have a story for you I I quit to yasha and I went to risk metrics because I was still under the impression that we could do better if we just did better math And I worked for value at risk firm the risk metrics I've invented that Was that the firm that came out of jp morgan? Yes. Yeah So we did we did the risk portfolio for all of the big banks and many of the hedge funds and And in particular, I was assigned to revamp the credit default swap risk model Um, and guess what? Um, I found that the assumed return distribution was normal Because everything was that way in that model And so I invented a An alternative assumption for credit default swaps because the credit market was far from normal It was much more like double exponential, which is to say much more fat-tailed And I saw that because by looking empirically at the data and again, like I'm not claiming that I we can really Really tell in advance what's happening, but you can do better than stupid assumptions And so I introduced this new way of looking at credit default swap risk And guess what happened rob? That they didn't want to hear it because guess what what it said It said that if you have a local measurement of volatility You cannot assume that you that the market's not going to move because this is a very very Fat-tailed distribution that credit default swaps actually follow And it's and a local measurement of small volatility doesn't keep you safe So your risk is much bigger than it seems with the old model So why don't you move to my bigger my my new model and be honest about your credit default swap risk? And as you can imagine nobody wanted to be honest about their credit default swap risk Because that's not the game that we're playing when we work in hedge funds events We want to maximize Our returns and minimize our risk because that improves our sharp ratio, which is how we get our our bonuses Long story short like I am both absolutely with you that people hide behind this authority of mathematics the authority of science It's not really science Um, we we don't like uncertainty. So we just pretend that things are explainable using these models In fact, I'll go further. I I think most of the examples of my book weapons of math destruction I could argue our Deft political moves to avoid a difficult conversation A difficult conversation about race about criminality what we why we have prisons a difficult conversation about Whether a teacher is a good teacher because it's hard to agree on what it means to be a good teacher So we don't want to have this conversation. So we replace those conversations with these sort of clean looking hygienic sanitized algorithms Where we don't explain them and we hide behind their authority Mm-hmm Yeah, there's an old article from the 1920s Called the dismal science H. L. Menken who was kind of a gadfly Wrote about it and he said in essence I think the only people I trust Less than theologians are economists Because they are not free They use their methods and they use the kind of dance or ritual of their methods to intimidate Create a sense of awe But he said the problem is The consequences of your analysis vis-a-vis powerful people and institutions are something that you could anticipate And therefore when you analyze something honestly You may be which my call silent and not express it if it has those ramifications for your career Yeah, where on the other hand if you're reinforcing The legitimacy of power etc You're you're online for promotions and chairs and high level government positions and for the ambitious we uh Who have those yearnings They're not really providing which you might call analysis for the public good And I find this uh very haunting that this man wrote this so lucidly and he was not A left-wing guy or anything. He was just saying You've got to look at the contextual process in which analysis is done In the incentive systems around the people doing it to understand and I I've said this a couple times one of my board members is julian tett Very fine writer for the financial times. Who's a phd in cultural anthropology? And at the onset of inet she said to me I said julian, what should I do? We had lunch and she said rob as an anthropologist. I'm gonna tell you Go out And study the silences because if you have the map of what's not said You'll understand the structure of power that your profession is grappling with I just wrote an essay today Uh, and I if my editor just sent me an edited version while we were speaking on Missing data and and I I think I called it. Well, he always changes the title, but I called it something like um dangerous Line spots because I think the danger in data sets almost always lurks in the blind spot And whether that's police accountability data like You know, we just saw Cuomo say that he'd be willing to sign a repeal of 50a which allows police personnel records and disciplinary records to be secret um Or badge numbers You know badge numbers that were being hidden during the protests like all sorts of the powerful protect their data It is the powerless whose data gets scraped Yes There's a famous old historian named e.h. Carr And he wrote a lot about the period between the first and second world war including the depression called the 20 years crisis his most famous book And in that book Oh, no, excuse me in another book that he writes about the methodology of being a historian. It's called what is history? He says facts are like sacks You got to put something in them or they don't stand up And the idea that there isn't an interpretive dimension of what facts you collect what facts you focus on Or how you interpret them Uh has this What you might call false consciousness of being value free. Yes and this is uh This is haunting to me Because we went from a time Call it prior to adam smith's wealth of nations 1770s Where the church got in trouble by being a partner with the oppressors the landed era stochracy the feudal lords and we moved away through the industrial revolution to a form of governance Where moral and ethical discourse was the currency or the language of policymaking to creating a value free scientific technocratic mode Uh How they say of expression and and I can give people a little bit of slack Which is if the church was really corrupt and the population knows it You go in and talk moral and ethical language and they think he's part of the wrong team And if the earnest people trying to do better, but I think the staleness Of this moral and ethical Discourse being absent and pretending these are all objective value free things is very very dangerous and things in your work Particularly like we were just talking about with the The prisons and who should be let out so these these are these are just immersed In human values and the tension between economic process money And human safety is another dimension that You know like the closing of the universities or not. There's they're just enormous moral and ethical issues. Yes Hiding in plain sight right now Yeah, I sometimes think that science You know often Is exactly what you just described. It's sort of like a way of pretending We don't have we don't it doesn't come down to an ethical or moral choice And I feel like I'm you know, I'm like I'm lucky. I'm very lucky Because I do have a phd in math I have learned data science like I can see that the the veneer of sophistication around a lot of these algorithms Is just that it's it's a veneer. It's hiding A much more interesting and more critical conversation And that's one of the reasons I you know I decided to quit to quit my job and and to go on this quest of sort of trying to get people to to question To to discard blind faith in in in this technology. I think it's I think by the way that has largely been settled I I think people are much less Uh, you know, they're they're much less trusting of algorithms and I'm glad to see it. Of course That begs the question Um, what what next, you know, like can we um, can we do better? and because after all like To back up a little bit the karma scores were introduced Because judges were known to be racist and classists. They were known to throw throw the book at poor and darker people and so the question We need to ask ourselves next is like well if we can't use algorithms to solve all our problems and make things fair What can can we do? We can't go back to humans Or maybe we can but I don't I don't know and so one of the things I've been working on Because I I'm a hopeful person believe it or not. I might sound like I'm only angry and sometimes I do feel that way but I always want there to be a hope and So that's that brings me to my algorithmic auditing work where I I really do what I find is that algorithms Are inherently are racist because they're trained on historical racist data But we can actually adjust for that. We don't have to accept whatever comes out hope off we can Just as just as pollsters account for the fact that they didn't really talk to enough people of a certain demographic and so they can Like reweight the answers that they have to to sort of Create a mosaic that is more realistic to the populace. We can do that as well Whereas it's not we don't particularly want a realistic picture. What we want is probably For algorithms to follow the values that we state the the stated values Our firm of our society or what have you now? What what have I just described? I've described that algorithms should follow values Which is to say that the conversation around values has to come first and that is the difficult conversation I was referring to earlier. We must actually have That conversation in order to make the algorithms work well in particular for prisons we have to actually decide what prisons are for And if we decide that prisons are To you know to ruin people's lives and to especially ruin poor people's lives and people of color Then we're doing great But if we have a better View a better vision Then we can we can use algorithms. We can use data. We can actually theoretically at least Do that we can optimize to success where success actually is an agreed upon definition of success rather than what we're doing Which is actually extremely destructive I uh How would I say I think you just in that last passage you just spelled the notion that you're only angry very convincingly Uh, I think you have a very constructive Sensibility And I want to move a little bit now in the direction of the book On shame that you are working on because in some level Economists act like You have what are called preferences And the preferences are kind of icy and they're They're just there rational And they're rational and they're structural And then with that you Build something from preferences called demand and then the market serves Your needs And in sometimes In some circles it's kind of romanticized the market becomes deified Yeah, but what concerns me Is that I see an awful lot of energy being spent Trying to shape Your values and your preferences Exactly You know you go to business school. They don't let you not take marketing So they obviously think persuasion is in play And so I I'm I guess I'm coming at the notion of shame Yeah, which is you trying to impart to people What is right and wrong to think and How would I say that you will be rewarded or penalized For expressing Things that are what you might call a braid those structures But Tell me tell me what inspired you to write a book on shame. Okay. Well, yeah Lots and lots of responses. I'm having all at the same time. I'll try to sort them through So the first thing is I actually was inspired To write this book because Because I'm always I'm fascinated by power And what I saw was a a connection between two events that seemed very different to me But were connected by shame By the power of shame. So the first one was Um, I was talking to a teacher who had been fired based on what's called the value added model score And this is from my first book It was it was a terrible so-called teacher accountability tool I put that in air quotes, which was itself not accountable It was you should think of it as a random number generator because it really was almost a random number generator It was part of the no child left behind thing and it was awful And it was also inexplicable and high stakes because again teachers got fired and I was asking this teacher You know What well, what how does it work? And they and their answer was why I tried to find out but they told me it's math. You wouldn't understand it And I was like, oh my god like That would never work on me But what I realized was like that is math shaming that was in a very explicit Shame tactic and it was in fact the same thing we were talking about earlier It's it's an authority of the inscrutable. You're not smart enough to understand this is too complicated for you In fact, of course your job on the line So you have every right especially as a civil servant to know exactly what how it works But this line was given not just to my friend This teacher but to many many teachers. It was a common tactic. It was in fact probably Um a suggested deployed tactic. So that's the first story and the second story is I I'm an overweight woman. I'm a fat woman. I've always been fat. I Came to terms with being fat after a lots and lots of fat shaming as a child But I you know got married had three beautiful kids felt good about myself Then my my older brother was diagnosed with diabetes my father You know passed away recently and for diabetic reasons I really don't want diabetes and I don't I didn't have diabetes But I was finding myself You know with all the risk factors for diabetes and I'm I was gaining weight and having trouble exercising Which I knew was like the only real way I could avoid it. Anyway, long story short. I decided to get a A bariatric surgery, which by the way, I will be writing about in my shame book Unbelievable miraculous surgery, which I did end up getting But the the hoops I had to climb through Rob to get this surgery and the reactions I got from people Because again, I I didn't feel particularly fat shamed at that point in my life But it all reoccurred like I got fat shamed explicitly multiple times Because I wanted to get this surgery Even by fat friends of mine who are like, oh if I ever lose weight, I'm going to do it The honest way, you know, there's all I can't even tell you it's just like it's actually outrageous Anyway, so it really the power of that fat shame over me, which I thought I had been ignored to It was not I was not I was very vulnerable to it because shame is powerful And so it was like that was the connection. I was like, wait a second I thought my getting a surgery had nothing to do with algorithms But what the commonality is that the power of shame is unbelievably potent and it has nothing to do with the economic rational choice theory And so the second point I wanted to make in response to your Your references to economics is like Economists, I I think they cheat and I'm not an economist, but you know, I think they cheat I think they basically say We believe that you know people's preferences are as they are and then we decide that the value of something is how much people are willing to pay for it And like it's all like a circular argument. Um, so they never actually have to discuss value um, and to that point like Mark and you also mentioned marketing and I'm being a slightly inarticulate But I think it'll come together at the end In business school, they tell you how to market things. They tell you how to persuade But one of the things one of the most sort of historically Winning ways of persuading someone is shaming them So there was this effort for example by a soap or shaving company to get Japanese women To be ashamed of their hair Because they weren't buying enough razors So they need to make women ashamed of the hair on their legs So that they'd buy more razors and it worked That is a persuasion But it so so then women these Japanese women were like, oh, I should go buy a lot of razors because of my shameful hair Is that Does that mean those razors were suddenly worth more because they're there's the shame had been increased into buying them They didn't need those razors in the first place, but because they felt shame they started buying them Do you see what I'm saying here? It's like a really aesthetic manufactured concept of value Which means for me that it actually has has no value Mm-hmm. Well, there's shame is sort of the stick and then there's the what you might call elixir of Temptation You know, you should smoke cigarettes. It makes you fashionable or cool or whatever at one point in time. Yes It is so I can say and there's like there's a lot of examples like that. I mean, I was just writing my Chapter I'm writing my chapter on like sex and body shame and like There's like so much for especially for it's very gendered for sex like For women if you're approved that's shameful, but if you're too sexual It's also your slut shame like it's really on both sides and in many many cases Yeah And how would I say it preys upon the vulnerable? If the structure of society makes you Feel anxious feel Like how would I say you want to turn off the fear? You can be very vulnerable to all of this pressures psychological pressures related to values and behavior. Okay, please so So potent because it makes you question your own value as you say it makes you question your value as human It is so potent that it creates a kind of cognitive dissonance and and there's nothing that Makes us do more crazy impulsive things than than cognitive I mean my theory is that most of the Conspiracy theories that are pushed around are Almost directly due to this cognitive dissonance. We need a new set of facts Going back to the question of what is what is a fact? We need a new set of facts so that It, you know, so that it wasn't in an unacceptable behavior inappropriate behavior for those cops to push down that 75 year old peace protester Let's make it that he's a member of antifa so that he was the bully You know, that's that's a cognitive dissonance because we were we're accusing the police of being Doing something shameful having shameful behavior and the people who want to protect that Assumption that police aren't shameful They have to build an entire universe of alternative facts in order to make that case the Structure you shared with me as we were preparing of the different types Of shame. I remember there were three types that you Illuminated and sort of Reacting to authority Talking down But why don't you why don't you elaborate for audience on on all three? Yeah, I well Right. Um, I by the way haven't finished the book yet. Um, so I I'm not even sure when it'll come out But like maybe we'll come back we'll come back and do another session when the book's ready to go I guarantee and I hope we still you know, I still have the same opinions about shame by by that But yeah, I have three different types of shame and called punching down Punching up and punching nowhere And punching up as you say is sort of holding powerful to account And by the way, you know, you could you know If you heard me talk about shame in the last 20 minutes You might have assumed that I think of shame is a bad thing. No shame is an essential tool But like algorithms, it's a tool that could be bad or could be good. It depends on how we use it Sometimes I when I with with respect to algorithms people say, well, how do I know if an algorithm is good or bad? You know, it's it's very contextual a given algorithm could be used for good or used for bad Um, but if it but the general way I think about it is if it makes unlucky people Unluckier and makes lucky people luckier That's probably the wrong way to do it. If it makes unlucky people luckier You're you're getting similar And similarly for shame the punching down shame is So the punching down shame is basically when you're making unlucky people unlucky or you're sort of You're shaming people for stuff that they probably can't avoid and probably cannot improve Of course shaming is supposed to be correcting behavior So you have to you have to make it sort of Reasonable shame you have to be asking someone to correct behavior that they have control over and that brings us to punching up shame Which is when you're holding powerful people to account and you're saying hey, you're doing this wrong You have the power to do something better. You have, you know, and you're in a position To do that and we want we're going to watch you and we want you to do better And so that's sort of like every civil rights movement ever That was punching up because it's holding powerful people to account And then finally there's punching nowhere Which is when you're trying to punt when you're trying to shame people and correct them with respect to a norm that they don't even agree to So when we try to shame white supremacists for racism It's punching nowhere because they are they don't actually agree to the norm that racism racism is bad That is actually the opposite of what they think I make the case that punching nowhere is not only useless. It is actually counterproductive in the book Uh Well, Kathy you've covered on an awful lot of ground today your organization It's called orca and I must say I had a whale of a good time I think you're gonna have to come back because there's so much more to explore but Everybody should read weapons of bad destruction and wait with baited breath for your book on shame Thank you very much. Let's uh, let's talk again soon and uh, but mostly thank you for today Thank you And check out more from the institute for new economic thinking at inet economics dot org