Robin's Birthday. with Arnold Brooks
Hello, Agnes, and hello, Arnold. Hi. It's my birthday today. Happy birthday to
me. And what I wanted for my birthday was to talk about something that is just
important to me, but maybe less interesting or important to you guys, but
you're indulging me by talking about it. So thank you.
Um, the topic is my obsession with these ideas for better agents. And I just
want us to try to follow that route, you know, rabbit hole down and see where
it goes. So I've got a couple of related concepts. The general idea is just
the idea that in our world, we have these people we rely on, uh, but we don't
know them very well. and we're at great risk for them not doing what we want
because we fight it hard to see if they're giving us what we want. And so we
make many choices on the basis of advice that we should be doubting because
it's hard to see how they could have good incentives to give us good advice.
And so as a response to that, there are ways to set up a bit more complicated
relationships with them such that they would be more trustworthy. That is,
they would have much stronger incentives to give us what we want. And that
seems to be an enormous promise. And I've got some ways to do that. I'm going
to explain in a minute. And the issues are that it's hard to generate much
interest in them, uh, to get people to think about them or to try to do trials
of them, uh, et cetera. And so, you know, the, the, the, the metal level
question is, you know, what is it that makes people not interested in these?
And is there a way to vary the pitch or the context or something that to get
people interested, or is there a way that lack of interest is showing us some
way in which the analysis is broken or wrong? That is, we're just misjudging
what's the key issues here or what people want or something like that. So
that's, that's the framing. And so now the particulars are better agency
relations, ways that you could have agents who better serve your interests. So
I think I'll start with a medical agent. So the two standard ways to pay for
medicine are, on the one hand, you pay fee for service. Each thing they do for
you, you pay for. And the other is competition, where you just pay them so
much a month and they do everything. And these have not great incentives, So
under fee for service, plausibly, they would just do everything that possibly
could, even if it wasn't very helpful. Because they get paid each time and
under capitation, they would want to do as little as possible because they
have to pay for everything. And they're going to get the same payment from you
regardless. And so, you know, we might say, how is it that we can trust these
people? And supposedly we have professional licensing and regulation and some
sort of maybe word of mouth reputation or something that would help you judge
who you could trust. But they seem pretty weak mechanisms compared to the
problem of, you know, how do you know if what they're doing is any good?
whether they're doing too much or too little. So in this context, my
suggestion is first to merge health and life insurance. That is to make sure
that if bad things happen to you, bad things happen to them, and then they
want to avoid those. And so when they're making a trade-off of whether to do
some treatment for you, they will weigh off the cost of this treatment.
against the benefit, which is to them avoiding the bad thing if you die. And
we could also have disability insurance where they'd want to avoid the bad
thing if you become disabled or even pain insurance. And so that's the first
step would be to merge those things and then they would get a stronger
incentive. And you'd of course be advertising this to customers saying you can
trust us more because And then the second step would be to increase the amount
of insurance past what you'd want for your private reasons in order to get
their incentive strong enough. And then we need a third party financial
relationships to to make the budget work. But that's the key idea of of better
medical agents. And so I also have, you know, examples we could discuss for
career agents and for crime, avoiding crime agents and some other things, but
let's start with that one. So that's the core idea. So what do you think are,
does, does it appeal to you? Do you even feel motivated by the problem it's
solving? Do you, are you worried about something going wrong in it or. If I
just misdescribed the problem people care about such that why should people
care about this? What do you think?
So I wanted to raise a general mechanism that I think stands in your way for
all such agents, which is The status quo requires trust. That is, we, the way
that medicine works or that law or all these things work is that we have some
amount of trust in our doctors and in lawyers and in the system. And of
course, your system would also require trust. But, you know, The way that
trust works is that like there's, um, uh, There's a certain kind of
inhibition. You inhibit certain forms of suspicion with the collective
understanding that we kind of can't check those things or can't figure them
out. Or we agreed not to think about it or something like that. I don't know
what the right analysis of trust is. The point is, given that situation,
trying to check those things looks distrustful. Right. And so, um, it looks
like you're full of suspicion and distrust. And so the attempt to improve
above the status quo is going to be associated with, um, distrustful
attitudes. Yeah, that's, um, that's my thought.
So, so before Arnold weighs in, which I hope he will do in a moment, Just note
the context. We have lots of other relationships where we are distrusting. And
so there's a question of why we are so more trusting here as such that would
be an obstacle when in many other contexts we are quite willing to be
I don't think that's true. I think in all of our relationships with anyone,
pretty much we are trustful. We're not always trusting about the same things,
but for any relationship you have with anyone, there are certain like trust
boundaries where there are certain things you're not supposed to be checking.
I think it's true of all your relationships. It's true with you and someone
trying to sell you something in a store.
Right. But, but the relationships you don't have, then in some sense you
don't, for example, most people don't use matchmakers. Right. In part because
they wouldn't trust them. So then the question is, why don't we have
matchmakers? So your same argument would be if we had the matchmakers, we'd
say, I just have to trust them, and then we would. Yes. Right. But we're not
doing that, right? And similarly with a king, we could have a king and just
decide to trust them, and we wouldn't need to elect them because we wouldn't
need to distrust them through elections. We would just trust the king, and
people have at gates. So why do we have a king? Why do we have matchmakers?
There's a question. Why trust in some places?
Right. And there might be, for instance, you might be able to think at a meta
level about what are the better forms of trust. I'm just saying this is a
general theory as to why modifications of the kind that you are proposing and
even like even testing or whatever experiments are that are modifications of
the status quo, sort of no matter how this trust is distributed in the status
quo, those are always going to be distrustful. And that's always going to be
something that you're up against. One of your questions was, what am I up
against? And what I'm saying is, one thing that you're up against is you are
sowing seeds of distrust. And that makes people panic and be fearful because
trust is like a really big part of what makes society work on any arrangement,
So I mean, I had a question that I guess may be related, that's more specific
to this medical case. So it seems like one way of understanding your point
would be that life insurance providers have an interest in keeping you alive.
And we should just give them more agency over the parts of our lives that are
related to and they'll behave well because they have this nice, strong
financial interest. And as far as identifying obstacles to people adopting
this, the two that jump out at me are one, that is, this sounds like the kind
of idea where if people could try it, they might really like it. It seems like
the kind of thing that would just work well if we could try it. And so one
question to ask is, why is it that life insurance companies are so dormant and
inactive? In fact, I have this question about insurance companies generally.
Why do they do so little for people as proactively in trying to get them to be
healthier or to live longer? In general, insurance companies seem to just be
very, very passive. And that is a little puzzling. And then the merger would
require a health insurance company or a life insurance company to see the
other side of the coin as valuable to them. which they could do if there was a
great market demand for it very naturally. But if there isn't, then we have to
sort of try to understand why they would refer this to the current situation.
And then I guess finally, a worry that is sort of on the consumer side would
be that, as you've described in some of your work, Doctors are among the most
sacred individuals in our society, and the advice of doctors is regarded as
very sacred. And one of the things that pisses people off more than anything
else in the whole world is when insurance companies make decisions that they
regard as the purview of a doctor. And this idea that you're proposing is an
idea about insurance company agents. and allowing them to be more significant
agents in our lives, where people in general, I think, really don't like that.
They see insurance companies as mean, money-grubbing bureaucrats and doctors
as helpful, caring people who want to save them. And so they're going to be
just very resistant to allowing an insurance company to have any of
So, so just for context, property insurers and automobile insurers do
actually, you know, pay attention to the details of your world and give you
incentives to change your behavior. So your fire insurance for your house,
they will want to know whether you have fire, you know, uh, extinguishers and
you know, whether you're have a fireplace and you know that you don't burn
trash in your backyard and you know, there's various things they will be
watching for as things that will affect your rates and they'll happy to tell
you that. Similarly, automobile drivers, they will want to know your accident
rate, what kind of car you're driving, how often you drive and you know, they
are willing to get more involved in your choices for those things. You know,
maybe it's about, they're heavily regulated, so there are obstacles to getting
the thing. And in fact, in the United States, the insurance regulation for
health insurance won't allow you to charge people more who have higher risks.
And that's an obstacle because the life insurance market does. And so it's
harder to merge these things than it used to be, but not impossible. There are
firms that do in fact offer both of these to customers, but it's more
presented as a, you know, paperwork convenience that you could fill out both
paperworks from the same people at the same person at the same time. And
they're not really highlighting an incentive effect that are treating you
A question arose for me with the thing Arnold said, which is just maybe one
problem here is that we just don't want really large agents. That is like, we
think we don't like big business. We're suspicious of large agents in general.
Maybe we're least suspicious of government as a large agent, but people are
still pretty suspicious of government. And so maybe we think big agents can't
really help us. The people that can really help us are like the little guys
like the doctors. And what we mostly want is for these big agents to get out
of the way. And part of why we think the big agents can't help us. is that a
bureaucracy is designed to minimize the role of human emotion in decision.
That's just part of what bureaucracy is. It's not emotional. And we want the
people who help us to be emotional, to care about us.
So we're all associated with universities who are agents in many ways for
their students. They are large organizations. So that's somewhat in conflict
with the claim that we aren't willing to trust large organizations. Churches,
of course, are large organizations that people trust. Sporting leagues are
large organizations that act as agents for people that people trust. There are
many large organizations that people trust.
Do notice that the large organizations that people trust have some common
features in the way that they interface with the consumer, right? and a church
have in common, interestingly enough, the idea of a local agent who has a
personal relationship with you and interacts with you through that, whereas a
corporation, like a life insurance company, I mean, I'll have a life insurance
agent, and maybe that agent will make some efforts to get to know me or
something, but we just don't see our relationship to firms like that in those
terms. I think that there's something really right about Agnes' thoughts
Hospitals are also large organizations of which doctors are part.
Right, but people don't trust hospitals. They trust doctors. And they don't. I
think that they don't trust universities. They trust professors and they
don't. I don't know. Maybe people trust.
OK, but there are individual insurance agents who sell you the insurance plans
and they do like visit you and come to your door and try to have a personal
relationship with you over years and decades. I mean, that is a thing.
Yeah, that is a thing. It's just maybe that isn't the usual experience of
What used to be, that is people moved away from it because it cost more to
have this personal guy and they'd rather get cheaper insurance. But it used to
be that most insurance was through a particular person who you knew. In fact,
the rise of insurance was a famous business story in the 1800s because
initially insurance was gambling and it was seen as this illicit thing. And
the order to a salesperson had to go around and convince people that no, you
were irresponsible to your family if you didn't buy life insurance to protect
you when you died. And it was this personal relationship of a door to door
salesman that convinced people to buy life insurance a long ago.
Yeah. So maybe, maybe a way to get this to work and maybe this is something
that, um, you know, Zoom makes relatively more possible, is you need to, a
company would need to structure itself such that people saw their insurance
agents and they came to think of their insurance agents with whom they have a
personal relationship as the decision maker in questions about their health
Well, it could just be the doctor, you see, if the doctor has a strong enough
relationship with their larger corporations. So Kaiser is my medical provider.
My doctor at Kaiser, their practice is greatly influenced by Kaiser's
corporate policies about particular medical treatments. So I don't believe
that my Kaiser doctor is making a bunch of independent choices about
treatment. They are very much influenced by Kaiser's overall incentives, which
I appreciate because I tend to think medicine does too much and this corporate
structure is pulling them back to do less.
So maybe the solution then is we need to merge not just life and health
insurance, but life health insurance and hospitals. That is doctors need to
become employees of insurance
And Kaiser, the doctors, are employees of the insurance company which has its
own hospitals. So it seems like instead of Kaiser as the per capita payment,
that we could have this life insurance merging and it would look the same.
So have you suggested this to Kaiser?
I mean, not for like, I don't have access to Kaiser management per se. But
when I talk to ordinary people, they aren't very excited by this. So I don't
think we necessarily need to talk to Kaiser Management to get a sense that
ordinary people are not very eager for this. And that's something you guys
should be able to introspect to perhaps see what's going on there. So I've
taught like healthy economics for many years. I'm going to teach it again next
spring. And after many weeks of being professor of them, then I explained this
idea and they nod and they say, yeah, that would work, but there's very little
emotional enthusiasm or energy about it. Even though people can get very
emotional about many other aspects of medicine, this just doesn't appeal.
in my observation. Yeah. Maybe, I don't know, maybe the thing that would need
to happen is that you'd need to sit down with somebody in a room and say,
listen, you should buy health and life insurance from us because we sell them
to you together, which means that if you die, we won't just be sad. We'll have
to pay out a huge amount of money. And so when we're making health insurance
decisions for you, the fact that you might die as a result of a bad decision
is going to be salient to us. That's a really strange conversation to have
with somebody. It's sort of unsettling in the way, I went to a grocery store
once and I saw a chocolate bar that was advertising itself as slavery free
chocolate. And it was sort of unsettling because of course then that suggests
that all the other ones. It would be strange to walk into a room and be told
that the special feature of this insurance company is that they care whether
you die. Because it is true that if you just get hit by a bus, that's actually
quite good news for a health insurance company.
The new little in the sense that, you know, you know, you all get their
premiums anymore, but they will have to do more for you either. Right.
They ended up not having to pay out medical. That is, if you're old and that
happens, then that's not so bad for them.
Right. But like, imagine if you could choose between you know, two different
sets of parents. Like imagine if before you come into this world, you can have
a conversation. The first set of parents is just the regular kind of parents.
And the second set are ones where there's a, they have a bunch of financial
incentives for you to be happy or something. And you're like, so which kind of
parents do you want? The ones that are incentivized towards your happiness or
the other kind? I can imagine that people aren't going to jump into the arms
of the parents with the financial incentive and they'll just feel
uncomfortable about this whole thing because what they want is for their
parents to care about them. That is that's how they want this to go. We want
our doctors to just want to keep us alive because they'll be sad if we're
And will the money stop them caring? But the fact that with our money with
that, I mean, if you have the same doctor who cared and we add the monetary
would somehow carry.
No, but it really will feel like that is it'll feel like that's the if if
money becomes is seen as a consideration. I think people really do infer from
that that other things aren't.
So lawyers are an example where sometimes we do give them incentives like
this. So as you may know, one way to hire a lawyer is per hour. Your friend
had a lawyer and they like them and you go visit their office, they've got
books on the shelves and the wood paneling and everything looks very fancy and
they say, sure, I'll help you. And he looks friendly and he smiles. And then
you can just pay them per hour, or you could have a contingency fee. So this
works best in the case where you're going to sue Solon. And then if you win,
they're going to get a percentage of the win. And now, you know, you have a
choice of two lawyers, one of whom you can just pay per hour and they'll do
whatever he does per hour to try to win your case. And the other, he gets 40%
of your win. And now you don't have us pay very much per hour. And you even
have the benefit that if he thinks you can't win the case, he'll just reject
and say, no, I don't want to take your case on a 40% contingency because I
don't think you have a case. And then you would save the money and go home
instead of paying this other guy so much of an hour, because he'll tell you
you have a case when you don't, because he likes to get the dollar per hour.
So which lawyer do you want? I mean, you know... First guy says he cares,
Again, we're, you know, in Supervised, we're trying to explain the fact that,
though it does seem rational to prefer the lawyer who is paid the contingency,
It's, in fact, rare to get a case on contingency. Most corporations don't hire
on contingency. Most individuals don't. And in many places, lawyers have made
it illegal to pay them. Because they don't want to take that risk.
But one thing to pay attention to might be TV shows and movies in which
lawyers are presented as uh, uh, especially as supporting characters, but in a
somewhat heroic role. Um, so I'm thinking like Molly's game or gone girl where
these lawyers, the lawyers who are in these movies are like, like good guys.
They're portrayed as being somewhat mercenary, but the way that the movie
signals that they're a good guy is that they'll do something like take a case
on contingency, but without even believing that they'll really win. That is,
it's sort of like the way that you have to present a lawyer in order to
present them as a good person, as a heroic figure, or as part of the hero's
team, is you have to make it clear at some point in the movie that the
financial incentives aren't their driving And then they form a close personal
relationship with the clients and the movie gets to go on and it works out.
But what that sort of suggests is, and you can look at the way doctors are
portrayed too in heroic ways. that those are the kinds of motives we want
people to have when they're a part of our lives. And it may be that more
effective motives like money are nevertheless dispreferred just because they
I mean, we know the usual ones do care about the money, right? The doctors
don't say, oh, I'll forego my salary in your case. Lawyers don't say that
either. The politician doesn't say I'll forego. I mean, none of these people
are saying they're going to forego their salary. So you know they have a
financial interest that's not at all concealed.
But so I, I think that, um, I do think that the, the comparison is not between
like the one who works on contingency and the one who, um, you know, you pay
per hour, because I think what we're imagining is that in the world where you
combine these things, a being that is otherwise recedes into the background,
namely the insurance company, would creep into the foreground and start to
occupy the role that, like, until that point was occupied by your doctor. And
that's sort of underscored by Arnold's thought, yeah, we have to integrate the
doctors into the system. And we just, we don't want the insurance company,
like, having that much of a role in our lives, we want them to recede into the
background. And that's because we more trust someone in the space of the
doctor who, yes, they have financial incentives, but they're not very direct,
and that's important. They're going to get paid their salary whether we live
or die. They're going to get paid their salary whether or not they do this
particular surgery, and so there's enough of a decoupling of their financial
incentives and what happens in this particular case with us. They're indirect,
and so then we can still think, yeah, but they care, and that's motivating
But notice that almost all our relationships in a capitalist economy are with
organizations where employees are primarily constrained by their organization
in their relationship with us. And that's what we seem to prefer in the vast
majority of industries that we interact with, grocery stores, restaurants,
automobiles, manufacturers, automobile repair, bus companies. housing
companies, all of those other industries we're involved in, which provide us
services that we are at risk for if they do a bad job. We are, you know, the
pilot and the steward on a plane, you're worried about the plane crashing or
something, but you're trusting the incentives of the company who makes the
plane, who runs the airline to do well by you. And you know that if they crash
that plane, then there will be hell to pay or them and for all their future
customers and they don't want that to happen. And you're not trusting your
personal relationship with the pilot or the steward or the person at the
ticket counter for an airline. You're trusting the company overall.
I wonder how flights would go if there were no flight attendants though. That
is, I think that having there be someone who wanders, they barely do anything,
right? Flight attendants. Right. That would be so scary. Yeah, like isn't that
weird that it would be scary if there's not someone there who's the
representative of the plane?
Well, I think even a restaurant would be a little scary if you never saw a
person and just the food popped out of a hole in the wall on the side or
But it'd be a lot worse on a plane. I feel like maybe we, since we're halfway
through, maybe we should just introduce another kind of agent.
Yes, let's try another one. Because maybe you'll, by considering multiple
ones, be able to triangulate some overall pattern that you don't see in just
one. So my second example is the example which seems to have the least
obstacles of anybody objecting to it in terms of losing from it, and that
would be what I call a tax career agent. So at the moment, people in music or
acting or athletics or some other careers have agents who take 10 or 15% of
their income and who advise and promote them. Most people don't have that sort
of thing and most people seem to make actually pretty bad choices often about
jobs, career planning, which school programs to go to, things like that. And
the idea is we could give everybody an agent. who gets on average 20% of their
income and then has a pretty strong incentive to advise and promote them
without any cost to anyone. We can do this for free and you don't have to have
one if you don't want it. It's just, if you want one, one will be created for
you. And the key idea is you, the government is already such an agent for you.
That is the government takes 20% of your income. But they do a bad job. They
don't actually advise and promote you. So all we have to do is transfer this
role to somebody else at no cost to anyone. So the key idea is that the
government at the moment usually spends more than it takes in. It makes up the
difference by borrowing money, which is a way of taking future tax revenue and
converting it into current money. Another way to do that conversion is to
auction off the right to get all the money you send to the government every
year. So, you know, we take you particular Agnes Callard and we say, do you
want to tax career agent? If you say yes, then we hold an auction. And in the
future, all the checks you will send to the government will then be bounced
and said to somebody else. and they'll get the money. And they don't have any
control over your taxes or what you're allowed to do. And they don't have any
other power but to advise and promote you, but they'll spend on average maybe
half a million dollars to win this auction. Government get that money in front
instead of the money later. And then you will get this person who you could
have listened to because now they will want to advise and promote you because
they're the auction winner. Therefore, they're the one who most expected you
might listen to them. And that they might be able to give you good advice and
improve the value of this asset they bought, which is a percentage of your
income. So now you win because you asked for it and you, they can't do
anything, but give you advice. They win because they decided to be in the
auction. Otherwise they wouldn't have done it. The government is fine because
they get what they were going to get anyway, in terms of borrowing. So
everybody wins. And, um, this would seem to be a big win, a big gain. Now
you've got this person, you could get advice about your career and you don't
have to listen. They've got no powers over you other than to give you advice.
Right now you don't have anyone, you know, they don't care about everything
about you. They just care about your income. But most people advise you only
care about some part of your world, not about everything. So, but many people
I talk to, they just find this a disturbing image. even though you don't have
to have one if you don't want one, why not let somebody else have one if they
want one, but they're not so sure anybody should be allowed to have one.
I have a question about, well, I kind of have a question. So, you know,
presumably these would be like, someone would have to be betting on you in
effect, like looking towards the future. And it's like a big gamble. And what
information would they use? How would they decide how much to spend on me,
So it's like, say, buying a piece of land. I mean, investors buy pieces of
land all the time and what to decide which pieces of land to buy on. They have
to, you know, look up things about the land. They could go visit it. They
could look statistics about its previous use. They could look at business
activity in the area and land use in the area and what it's, you know, what
income it's got in the past. So similarly for you, then some information about
you would be revealed to these investors and they would go in that
information. Now, if some of them suspect that other investors know more than
they do, that will make them shy. And you might pick out particular people
you'd like to be your tax career agent and tell them, hey, let me tell you a
lot about you and I would like you to be a bit of this auction. I want you to
win. And you could influence the auction result by selectively pitching to
particular people. Because remember, part of the big value is will you listen
to them? So if you could say to someone, I would listen to you, that would
convince them that this asset is worth more to them than somebody else. And
they would then be willing to invest more. But just like with a piece of land,
if you want to sell a piece of land to people, often you might pick someone
that you want to sell your land to and say, I'd like to sell the land to you.
Here's why I think it's valuable. Here's why I think I'd rather you have the
land than somebody else. And if you're a land buyer, you have to worry if
other people have that advantage over you, maybe you should be a little more
reluctant to buy the land.
So so I mean one thought that I haven't had before about this proposal is that
it creates a bunch of incentives for in effect spying on people that. where we
might think a big part of what sustains privacy is just the absence of those
incentives. That is, people don't know that much about us, mostly because
people don't care and they don't have a reason to find out. And we don't have
a lot of ways, ultimately, especially in an online world, of preventing them
once they have big incentives. And so there's reason just not to create those
But the consumer could just say how much interference they want or how much
input they want to give. And it might affect the amount that people are
willing to bid on you. But if you said ahead of time, the only thing this
company, this agent should know about me is the kinds of things that I would
put on my tax forms. I want no closer examination of my life than that. And
they can only advise me on the basis of that. People might be less willing to
bid for you. Presumably you could just draw those limits yourself. And you'd
have an incentive not to. That is, you'd have an incentive to, say, give them
more information if you thought that the advice that you were going to get was
one of the main complaints about the modern world compared to older worlds,
that is, the story is that in the older worlds, you were embedded in a
community of people who knew you well and had many close relationships to you,
and that was important to the health of you and your friends and those
communities, and that the modern world has hurt you by having all these
arm's-length-weak relationships by which people are not watching out for you
and noticing you and helping you on things. And so a complaint about the
modern world is that you are too disconnected and that nobody really cares
about you. Uh, and so for example, people, you know, get onto drugs and then
overdose because nobody in their world really cares enough to intervene and
say, Hey, uh, this is a problem, you know, so here we're making people who do
care more about you and care about you positively. Uh, and then we say, Oh no,
you don't like that. I guess you would have hated the old world even more
So it occurs to me that maybe the thin edge of the wedge here for this
particular policy would be the government would be especially willing and
eager to sell future tax revenue of people who say were over a certain age and
had never paid any taxes, right? Then you could get that cheap. Right. And
those are also going to be especially likely to be people who have been in
some way like cast out by the system. And so we could sort of we could sort of
pitch this first as a kind of almost like insurance or remedial thing to help
specifically those people where now here would be somebody who is interested
in, you know, the drug addict who's been in and out of rehab or whatever and
who has an incentive to help that person specifically.
Sure, because you can imagine helping those people more perhaps, but it's just
about anybody that the agent can help, they would be an attractive target. Now
at the opposite end, somebody who is wealthy might want to just buy their own
tax crew agent asset and just lower their marginal tax rates. So it's a way of
paying upfront to pay fewer taxes later and then have a better incentive to
earn more money later. And so there's no, there's not a problem with, do you
trust them? Cause they are you. Right. That also seems to be a win for
everybody. I mean, nobody loses that star.
Right, so it might be that maybe those are the two constituencies that you'd
first wanna pitch this proposal with reference to, that is. Because I think
that most of us, if we're doing okay in life, are a bit wary of some giant
organization that- It doesn't have to be giant, it could be just one person
who bought it.
But this doesn't have to be a huge company. One person can buy your one asset
for one other person.
Okay, but that's like even creepier, actually.
You want the friendly doctor, but you don't want him because he's creepy, and
you want the big company, but you don't want it because it's creepy.
Everything's creepy, I guess. I think that if there were a regulated industry
of the special advice school that these people went to, and they had to pass a
bunch of tests. they had a diploma and they had their leather-filled office
with the books and whatever, then yeah, that might reassure me. But otherwise,
if it's just freelance advice giver who's bought up my revenue and I don't
know why they chose me, yeah, it seems creepy either way.
Just ignore them, right? But you're threatened. It's not just that you could
just ignore them. You feel somehow they're gonna sneak up on you in the middle
of the night and hurt you.
Yeah, or just manipulate me or
Yeah, I feel like you into making more money.
Well, yeah, I mean, it seems like the the essential problem with these
proposals is that is that we we. probably for evolutionary reasons, have a
very deep sense that there are certain kinds of relationships that should be
relationships of love or friendship, because that's way more reliable than
relationships of financial.
Almost none of your professional relationships are of that form.
No, no, I know, I know, I know. And then we've gotten used to seeing some of
those relationships in sort of bland, unemotional, institutional terms, and we
haven't gotten used to others, right? So we have not gotten used to doctors,
but we have gotten used to certain kinds of insurance, say, car insurance
agents or something. We've got a partial acclimation to the world of
institutions, right? And what you're asking for is for us to acclimate more.
So another example might be real estate agents. Yeah. So it turns out, I
remember there's data that basically a real estate agent mainly wants to just
have you buy something and be done with it. Yeah. They don't actually care
that you get the best thing for you. They have much stronger incentives, just
you pick something and be done with it. So it turns out when they pick
something for the self, they wait a lot longer. They look at a lot more things
when they're picking for themselves. When they're picking for you, they're
trying to get you done. They're going to say, this is great for you. You're
not going to buy something better. Take it and go. Right. And so you might
want a real estate agent who then had stronger incentives with respect to you
liking the house. And we can imagine constructing such a thing. But would you
find that creepy?
Yeah, I mean, I basically maybe the thing to look at is how did we go from
seeing certain kinds of relationships in institutional terms or from seeing
certain kinds of relationships in personal terms to seeing them in
institutional terms. The most important of which being that we went from
accepting that we would not be ruled by a Lord who has a divine right of rule
because of his descendancy from Adam, and we've known all our lives, we're
known as parents all of our lives, and we've just come to accept that we will
be ruled by this faceless bureaucracy And we've only partially gotten there
because we still care way too much about who's president and so on. But we've
made that transition and the question is just how. And then we need to figure
out how we can get people to make that transition in the parts of their lives
where that transition would be a real improvement.
So can I connect the tax credit to us as teachers because a plausible way in
which students might better trust advice from teachers or even what we teach
is if we had a stronger incentive in the students. So, imagine universities
had a percentage of your income. Imagine, in some sense, the university became
your tax career agent. That could make sense. That is, for the university to
become your tax career agent when you're a student, and now we as professors
would you know, face rules and advice from our administration, telling us
which students to emphasize and which directions to push them, because it
looked like that would be profitable for the university. How would we feel and
how would our students feel if we had more direct financial incentives to push
our students in particular directions, because that would make them and us
I think universities are ideal candidates for this program. That is, if
governments were willing to accept it, universities would 100 percent buy
them. It's just a great way for them to guarantee donations in the future. I
mean, there's already a system whereby universities attempt to extract a
career tax on you via guilt or appeal in fear.
Well, it would seem like they would, but the question is, this postulate that
we don't like the idea that big businesses have an incentive. We want this
personal relationship. I thought we could test that out in this story of, do
we have students who trust us because they think we personally care about
them? And will this sort of larger institutional incentive cause a problem
there where they would less like the relationship with each of us?
Yeah, I mean, aside from the fact that universities are fiscally, in my
experience, pretty conservative, I think if you got a government to offer
this, to be willing to offer this to universities, universities would buy it.
And I think that the students would be happy to live in this arrangement. I
think students would be happy. universities, especially because the
universities could offer pretty immediate incentives to students to sign on to
this kind of program in the form of tuition credits.
Just to be clear, some universities have actually taken percentages of future
income as a substitute for tuition. That's happened in the world in the last
few decades. But I think Agnes disagrees with your assessment here.
Oh, okay. Agnes, what do you think?
No, I didn't disagree that universities would want to do this. I have no idea.
I think that it would be, like, look, the thing you said about Kaiser, Robin,
you're like, I like the fact that the doctors are constrained by Kaiser to
give me less medical care, because I think medicine doesn't work, and so then
I get less medicine, which is what I want. That is not how most people think,
right? So most people will take the first few premises of your argument, and
then it will lead them to, I hate Kaiser, not I love Kaiser. And so I don't
think, I think that my students, as long as I bad mouth the university enough,
it might work out even. That is, if my students got the sense that I was
disloyal, that I was not responsive to these incentives, that I was giving
them the real advice, what I thought was good for them, not what I thought was
gonna make my superiors happy, whatever, it still might be okay. But if they
thought, about me as you think about those doctors. Namely, I'm going down a
checklist that the university gave me, and I'm like, ooh, better not give this
one the advice to go to grad school. That's not the road from which they'll
make the most money. They'll make more money if they go into investment
banking. No, my students would stop trusting my advice. I mean, independently,
I don't think my students should take my advice, and I tell them that. But we
already did the podcast on advice, so we can set that aside.
Yeah, so that's maybe the core thing to try to overcome is the fact that
psychologically when people see that money is a dominant incentive for people,
they regard those people as dangerous tricksters who will engage in a zero-sum
game by which this person will lose.
Most of our relationships are with such people. That is, when you go to a
restaurant, the waiter has such incentives when they recommend the specials
for the evening. When you go to a clothing store, the person's showing you,
telling you, it looks great on you.
I don't trust them anymore. Like when the waiter comes over and they're like,
would you like drinks? My first thought is, yeah, that's going to push up the
amount that I pay. And so it's going to set your tip. I know why you're
We don't want to go to nonprofit restaurants and we don't go to nonprofit car
dealers. If we hated this enough, then nonprofit versions would show up where
they made it very clear that these people talking to us had no incentives
whatsoever, but we don't do that.
But waiters, I think, have every incentive to produce a convincing illusion.
that they are not motivated by money at all, right? And so all of these places
where we do accept for-profit behaviors nevertheless involve employees that
have to convince you that they're not doing it for profit. Like, we're really
freaked out about that.
That's a really good point. That's a really good point, and maybe part of what
all of your examples do for the person who encounters them is confront them
with a situation where it's obviously going to be impossible to produce that
Yeah. But it's not clear that doing, creating better incentives makes the
problem worse. That is, so imagine a restaurant, you knew that if anybody died
from their food, they'd get a $10 million bonus to the restaurant too, for
them to redo their facilities. Okay. Would you be more comfortable with that
restaurant? Because this guy had the image that he didn't care. I mean, it
would seem like moving away from that would be better, right?
If I could choose between two restaurants, and the only difference between
them is that one of them will get shut down if I die of food poisoning at that
restaurant, I'm not inclined towards that one. That creeps me out, and I don't
want to go to that restaurant.
When they have a better incentive not to hurt you with their food, you don't
like that restaurant.
Correct. And they have a better incentive. I do not like it. And I think that
and I think the explanation is the thing that Arnold said, I think, like, I
don't want to think about the restaurant killing me. And they're making me
think about that.
You don't have to think about that while you're at the restaurant. You could
think about it ahead of time in general. I mean, sounds like- Okay, okay.
Hold on. Imagine your waiter comes out and they're ready to take your order.
And you notice that they come out with a police officer. And the police
officer has his gun drawn, trained on the waiter. And he says, listen, if this
waiter attacks you, I'm going to take them out right away. You're totally
safe. You're not going to be there by sitting. Oh, that makes me much more
comfortable with the waiter experience I'm having right now. You would be
terrified of this waiter who is apparently so lethal and so fast that he needs
to have a cop trained on him at all times. It would just freak you out. But of
course, the waiter has now a much greater incentive than a normal waiter not
to attack you. But the only thing you're going to be thinking is this is some
kind of super murderer waiter, like
So the analogy there is that you believe the situation where doctors would
have an incentive to give you a bad treatment is so rare that it's not even
worth thinking about. And telling you that we've given you better incentives
now says, oh, the problem is more common than I might've thought, and now I
don't want to be around this space where the problem is common. Because I
thought, of course, doctors always just give me the best treatment, because
it's the law.
I think that that's right. It's it's do you remember when we were at Vibe Camp
Robin and somebody and they're like, how do we get more women at Vibe Camp?
And someone's like, how about we just like pay the women to come in? You're
like, no, no, no. That's a bad signal. Well, basically, this is just the same
So there are times in which customers are willing to be skeptical, right?
Yeah. For some kinds of products. And then some kinds of products, they offer
warranties, for example, or money back guarantees, or they show you that
they're inspected by somebody, or they post a bond. in case there was a
problem. So there seem to be many businesses for which we are willing to admit
there might be a problem and that they reassure us in some ways that their
services are higher quality. So this isn't impossible. The question is just
why it would then be impossible about some of these other things we might do.
Yeah, I think it's absolutely possible to get this to work. It's just a
question of getting us to see certain kinds of institutions and economic
relations in those terms when we don't. currently see them that way. Because
you're completely right that we have absolutely done this with other kinds of
economic transactions, right? Nobody goes into a car dealership thinking
they're making a friend or thinking that they're getting a car sold to them
out of love. We just accept that it's an economic transaction, that their
incentives are in a particular direction. So we just have to figure out how,
it's just a matter of figuring out how to get people to make that transition
with regards to healthcare.
So I think if there were a car dealership that really did work that way, I
would totally shop there. The love.
Yeah. Car dealerships have every incentive and they make every effort to try
to make it.
Saturn tries to be that that was the Saturn. Oh, yeah. Saturn says, look,
there's a price. It won't vary. We can't control the price. You come in here,
you'll just get the one price. You don't have to worry about us trying to
screw you on the price. Just come buy a car from Saturn.
But I want to come back to the thing I said first, because I actually really
just think that that's where we've landed. Namely, that there's a certain way
that trust is distributed and there's a certain pattern that it takes for
relationships. There may not be that much reason why, like, what it is for me,
like, say, my trusting Arnold, right? Arnold and I are now, like, you know,
physically, like, separated, that he's in Philadelphia, I'm in Chicago. And
like, um, he tells me about stuff he's doing, and I don't think he's lying to
me. Um, but I could ask him for proof. I mean, it could be like, you know,
show me photos or show me this or that, right? Um, and I could just imagine a
form of marriage where that's how it worked. And it wouldn't necessarily be
bad or negative or whatever, but to-to say to Arna, let's transition to the
kind of marriage where we give each other a lot more proof, that's a terrible
signal. And it's-it's hard to do those transitions. So...
One theory here is that we, by habit, trust high-status prestigious people,
but not low-status understages people. And the problem is we are quite willing
to distrust car salesmen and janitors and busboys and everything else because
they're low status, but it's a problem for us to suspect high status people.
And an anecdote just occurred to me from early in my work career I was working
at this computer lab and we just got these new computers, which were really
expensive computers. And, you know, the next morning after one had been
delivered, it was gone. It was disappeared. And we questions what happened to
it. And immediately we suspected the janitors of stealing because they weren't
one of us and they were outsiders and turned out. that the janitor had thrown
it away because it looked like trash because you left it. The rule was if you
leave something outside your office, it's trash and he takes it away and you
just follow the rule. And somebody had stupidly left this new computer. So why
all this fray? Well, it wasn't them stealing it or being illicit. It was just
they were following the rule. But the point is, as you may know, when
something is stolen in a rich household, they're willing to like, was it the
cook or was it the maid or was it the driver stole the jewels? It wasn't the
son or the visiting cousin because they're one of us, right? And so there's
just this way in which distrust often is directed toward whoever is low
status. And so you might say, well, doctors are very high status, lawyers are
very high status, politicians are even. Of course, once upon a time, we just
trusted the king. And then I guess at some point, we're willing to distrust
the king. And surprisingly, we're willing to distrust our top political
leaders, not willing to distrust our professors or our lawyers or our doctors.
But maybe soon we'll live in the utopia where we distrust everybody.
But something that's just occurred to me just as a practical solution would be
corporations in order to get more trust may just have, I'm getting this from
your H of M book, may have an interest in getting a language model who you can
talk to, right? So you can go to their website and you can say, listen,
Merrill Lynch, I want to have a conversation with you. And Merrill Lynch will
politely, firmly explain to you why it makes the decisions that it did. It
says, well, look, here's your file and here's the details of that. And here's
the kinds of considerations that we're taking in. And you sort of get a sense
of why it was making the decisions that it was making. It wouldn't just look
like a big black box. You could have it explain to you.
That already happens with humans.
No, I know, but people don't see that as, I think, particularly available to
them. That is, you call customer service, you wait six hours, it's somebody
who doesn't know what they're talking about, they're speaking in an accent you
But they're going to present an argument that looks good to them, and they'll
do it with the human or the language model either way, and you'll still have
to wonder if you can trust their explanation.
Yeah, right. But people take things like chat GPT to be high status.
Say the lawyer who tells you to go ahead with the bad case because he's
charging you per hour. If you ask him why to go ahead with the case, he'll
give you some reasons. And then a large language model will also give you the
same sort of reasons, but it will be selected. It won't be giving you the bad
reasons. It'll be giving you the good reasons because they want you to keep
paying them so much an hour. you're not overcoming the trust problem by just
having them explain things if you can't audit their explanations very well.
Well, remember that we're not trying to overcome the trust problem as in get
better epistemic relations to the decisions that these things are making.
We're trying to overcome psychological trust problem. Trust is a psychological
phenomenon, and we're trying to figure out how we can create trust in faceless
institutions, given that they might be better equipped.
I mean, it's more the opposite. I think we have too much trust. How can we
undermine the trust that we have so that we actually get effective service
because we're actually trusting ineffective service? I mean, as you know, in
fact, on average, medicine doesn't work for people. On the margin, people love
more medicine aren't healthier. So there's a huge collective problem with the
quality of medicine we're getting, which is a reason why you might want to
distrust it more. Where are we on time? About the end. Any last words?
I think that. You might be alone in your view that we have too much trust or
in any way, that's a. That's not something I hear a lot of other people
saying, so it's worth, I don't know, I'm thinking about what should be your
next, where should you push. And my advice would be, though I don't have great
incentives for producing this advice, so take that into account, but my advice
would be too much trust, how could we get less trust in our world and make the
case for that? And also that the career agents, what if we focused on the
people who have been excluded from society, drug addicts, et cetera, homeless
people, and we're offering them a service, in effect. If you pitched it that
way, that might be better. That's my final thought. Arnold, final thoughts?
I like the... That is the helping... I mean, because the government does this
in the form of social workers. That is, social workers are like government
career agents in a way. And I like the idea that that could... But they don't
have that incentive. Yeah, right, they do have that incentive, but that that
could become a better organized system, especially because they would be
willing to do things like spend some money on an initial investment with
somebody, right? Buy them a new set of clothes, get them an apartment. That
would make sense if you bought somebody's tax revenue. So I really like that
idea, but I also think that you could probably effectively put it to rich
people, and rich people might actually be in a position to get the government
to adopt buying out their own tax revenue as a way of basically loaning the
government a bunch of money, and that people would be interested in doing this
maybe even for their children, and that there you're working with levers of
power. And so it might get it instituted. But I do think that just as a last
thing, the central problem here is as regards to this question of trust, that
we should have less trust. I think we shouldn't see trust as something that we
sometimes give out to people. Trust is something where our lives fall apart if
we don't have a certain amount That is my last word, need to trust.
Well, what we want is a matching of trust and trustworthiness. Yeah. If we
trust without trustworthiness, we are vulnerable to being exploited. So the
point is we actually have low trustworthiness in our world, and that's why we
should have less trust. If we can create more trustworthiness with better
incentives, then it'll be great to have more trust. The problem isn't directly
having the wrong trust, it's having the wrong match between trust and
trustworthiness, which the key thing is to increase trustworthiness.
Right. But trust has interpersonal characteristics to it. That is, it's, I
think, probably emotionally speaking, irreducibly a matter of a relationship
And if you're in the habit of just trusting when you have a relationship
without them being trustworthy, you are in deep trouble. Yeah, that's true.
Anyway, thank you both for this conversation.
And happy birthday.