Marina Krakovsky: "The Middleman Economy" | Talks at Google - VoiceTube: Learn English through videos!

Subtitles section Play video

MALE SPEAKER: Welcome to "Talks at Google."
Please join me today in welcoming Bay Area journalist
Marina Krakovsky, the author of "The Middleman Economy."
Just published.
Marina's writing has appeared in "The New York Times Magazine,"
"The Washington Post," the "Stanford Magazine,"
"Scientific American," "Discover," "Psychology Today,"
and "Slate."
Marina was also a researcher for Eric Schmidt and Jonathan's
book, "How Google Works."
Marina has earned a BA in English
from Stanford University.
Thank you for coming to Google today.
MARINA KRAKOVSKY: Thank you.
All right.
So I'm here to tell you that we are all middlemen.
I'm a middleman.
You guys are middlemen.
And what I will say is why I say that.
And also, I'll give you ways to think about how you
could be a better middleman.
So if we think back to the day when the internet was still
in its early years and people were all
trying to figure out what's this going to mean for all of us,
a lot of people were saying that the internet's radical ability
to connect everybody would make middlemen obsolete.
Bill Gates actually wrote in his 1995 book, "The Road Ahead,"
that the internet itself would become the ultimate go-between,
the universal middle man.
He foresaw a time in the future when we wouldn't
need to turn to middlemen.
We would be able to do our own buying and selling directly.
But if you look around 20 years later,
you see, very clearly, that that actually has not happened.
In fact, I'll argue that the opposite has happened.
That the internet has made middlemen more prevalent
and more valuable.
So look at the biggest players on the internet.
eBay and Amazon are middleman businesses.
Uber and Airbnb are middleman businesses.
Google is a middleman business.
Right?
And I'm not just talking about tech giants.
I'm also talking about individual middlemen
who are thriving thanks to the tools of the internet.
So think about eBay.
Anybody can buy and sell directly
through eBay being a middleman but most of the trade on eBay
actually flows through resellers,
through trusted middlemen on eBay.
And on LinkedIn, something very similar happens.
It's all about the recruiters.
Recruiters bring in more of LinkedIn's revenue
then all the other LinkedIn users combined.
So the internet clearly didn't do away with middlemen.
It's actually made them more efficient, and often,
more effective at what they do.
So I've been thinking for a very long time
about what is the deal with middlemen, basically.
I mean, it's so strange.
Middlemen have been a part of every society that I know of.
Throughout history we've had middlemen.
And yet, people have often had very strong negative feelings
about middlemen, feelings of resentment,
anger, contempt, sometimes even disgust.
And, at the same time, middlemen are still there.
Or to put it another way, if people
are so eager to eliminate the middlemen,
why have we never been able to do so?
These are the driving questions behind my book
and that those are the questions that I really tried to answer.
So I talked to economists and psychologists and sociologists
who think about such issues.
But also with ordinary-- or you might
say extraordinary-- middlemen, people
who I think do a great job and are doing just fine--
thank you very much-- in the age of the internet.
So it's people that you might expect, like real estate agents
and travel agents and car dealers.
But also people that it might take you
a moment to realize that they are middlemen, too.
So, for example, gosh, a wedding planner is a middleman.
If you pause to think about that, right?
A venture capitalist is a middleman.
A person who started a two-sided market, like Open Table,
that's a middleman.
There are middlemen all over.
Gallerists are middlemen.
Anytime you're dealing with any kind of broker, dealer, anybody
who's connecting two groups of people, that's a middleman.
And what's interesting from talking
to all these different people is that even though they're
from different industries, you start to see patterns.
You start to see what they have in common
that is making them as effective and successful as they are.
And that's sort of what I started to do in my book,
is to tease out those commonalities that
are true everywhere.
And although all the examples that I use throughout the book
and throughout my presentation will
be of professional middlemen-- people who
identify as middlemen in their jobs--
I do believe that all of us are middlemen,
at least in some area of our work or in our personal lives.
So what do I mean by a middleman?
I think we all have an intuition about this.
And we think of a middle man is the person connecting
two groups of people or two individuals, the person
in-between.
But I want to give you a more useful definition that explains
what middlemen actually do.
And this definition comes from a man
named Mike Maples, Jr. He's a venture capitalist here
in the valley.
Started a fund called Floodgate that was an early investor
in Lyft and TaskRabbit-- middleman businesses--
as well as, of course, other start-ups.
And he says this.
A middleman is the person in a network who
connects nodes in the network to increase
the value of the network.
I think it's a useful definition,
and I'm going to talk a lot more about what that means.
But I also want to acknowledge that middleman
is a sexist word.
I realize it's a sexist word.
And I'm using it despite that because I really
think no other word will capture what I'm trying to do,
which is really push against these cultural ideas
about middlemen.
You know, that word has these strong connotations.
We just don't hear people saying,
ah, let's cut out the intermediary.
It's really about the middleman.
So forgive me for using that word.
There's a reason.
So going back to this idea that middlemen
are people who we increase the value of a network
by connecting nodes.
I do think it's a really helpful start,
but it's not a complete definition.
And the reason it's not complete is that not all connectors
are of equal quality.
Right.
We know that.
And if you think about Metcalfe's law--
I would guess everybody in this room knows what that is.
Metcalfe's law, this provocative claim
that the value of a network grows quadratically
with the number of connected users.
Well, that may not be true.
It may be true sometimes.
It's certainly not true for every additional user
you add to a network.
And if you think about it in terms of middlemen--
or if you think of it in terms of any kind of user,
actually-- if you have a network that has a certain value
and you take out some of those nodes,
sometimes taking out a single node
will just decimate the network.
It'll radically decrease the value of the network.
And sometimes taking out a node will
have almost no effect at all.
And sometimes taking out a node will actually
increase the value of the network
because some of these connectors are actually
having a net negative effect on the value of a network.
So if you think about spammers, that
would be an example of something like that.
So it is really important to think
about connectors' quality.
But how do you think about connectors' quality?
What is it that defines the quality of a connector
or of a middleman?
And the way I find it useful to think about it
is in terms of these two dimensions.
One is warmth, and the other is competence.
Warmth and competence.
And these are specialized terms that social psychologists
use when they're talking about how we judge other people.
So this isn't just true about middlemen.
This is true about human judgment in general.
When we meet somebody, we quickly
evaluate them based on our perceptions
of the person's warmth, which is all about their intentions.
Are they friend or foe?
Are they on my side or not?
And competence, which is a separate issue.
And that's really all about the ability
to deliver on those intentions.
Those are totally separate.
And if you combine them, you quickly
generate this nice two-by-two matrix
with four possible combinations.
And I want to talk through each of these
so you know what I'm talking about.
I think that these labels that I've come up
with really capture some of the emotion
and some of the ambivalence that we
have about middlemen because middlemen are not all created
equal.
They really vary in quality across these two dimensions.
So clearly, the worst type is the one
I call the parasite person, who is
low in warmth and competence.
After the financial crisis of 2008, a lot of people
put the bankers on Wall Street in that quadrant.
They've actually done studies on this,
and companies like Goldman Sachs and AIG
landed squarely in that cold, incompetent quadrant.
We don't need to say much about that.
What's more interesting is these ambivalent cases.
So, the predator-- the person who is competent, but really
doesn't have your interests at heart--
this is the person that we tend to resent.
And I don't know about you guys, but this is how
I feel about my cable company.
They're competent.
They're able to deliver high speed internet and video
to my home.
But clearly, they don't have my best interests at heart.
They seem to be taking advantage of their monopolistic power
for their personal gain, not for us.
They have high prices, terrible customer service.
All of that makes me think that they're not warm at all, even
if they're competent.
Now the pet is maybe a little bit better because they're
at least on your side.
But unfortunately, they can't really
deliver on those good intentions.
And, to my mind, a good example would
be the stereotypical mom-and-pop shop.
So the service might be really slow.
You might have to place a special order and wait for it
to be delivered.
The prices are high.
But it's not because they're trying to gouge you.
They're not trying to take advantage of you.
They really mean well.
They just can't do any better, unfortunately.
And this sort of person evokes feelings of pity.
So you might patronize a store like that
because you feel bad for them.
But if the difference in price, or quality,
speed, and all those sorts of things--
if that difference grows to a certain point,
you're going to give up on them.
You're still going to go with the retail giant instead.
So clearly, the best type of middleman to deal with
is the partner.
Both warm and competent, they have your best interests
at heart.
And they're really able to deliver
on those good intentions.
So I think everybody wants to deal with a partner.
And everybody should want to be a partner.
And way to think about being a partner
is going back to this idea of increasing
the value of a network by connecting nodes.
What partners are able to do is they're
able to increase the size of the pie.
And then take as their slice-- you know,
their markup, their commission, their fee.
And after they take that slice for creating that extra value,
you still feel like you are way better off than if you
hadn't partner with them.
And so being a partner puts you in a very strong position
because people really want to do business with you.
They're not just doing business with you
because they feel sorry for you or because they currently
don't have any alternatives, but because you really
are providing a great value to them.
So this is my foundation for everything
I'm going to say today.
We've got this idea of a good middleman, the admirable kind
of middleman as a partner.
They create value.
You appreciate them.
And then the question becomes, well,
how exactly do they create value?
This is a very fuzzy notion of creating value
by connecting nodes in a network.
What is it that they're doing exactly?
And this is really what most of my book is about.
I distill it into six problems that middlemen solve--
that good middlemen solve.
And I think that you will all find yourself in a middleman
position at some point.
And I hope that as I continue discussing
what these middlemen do, you'll see, yeah,
sometimes I do that in my work.
Or I see somebody else doing that.
And maybe I should be doing more of this.
I hadn't thought of that.
And that's the idea anyway.
So what do middlemen do to create value,
to be a valued partner?
Well, the very first problem that they solve
is what I call the problem of distance.
Distance can take many forms-- well,
three forms that I talk about anyway.
There's physical, you know, geographic distance.
That's pretty obvious.
But you can also have temporal distance and social distance.
So with physical distance, it's two people
who would both benefit from doing business together--
some kind of trade-- but they're just too far apart.
And the merchant steps in and connects them that way,
indirectly.
With temporal distance, we don't usually
think of middlemen this way.
But that's actually what a lot of them do.
So my favorite example of this is an appliance flipper
on Craigslist.
I'm sure you guys have all used Craigslist.
What this guy does is he's like a super user of Craigslist,
a power user of Craigslist.
He's just constantly on the site.
He knows all about the prices for the kinds of products
that he buys and sells because he has to.
He lives off of that difference, that markup between what he
buys for and what he sells for.
So he'll buy a used washer or dryer for $50.
That's the price he's aiming for.
He'll do that on Saturday, take the appliance off
of somebody's hands.
He'll quickly turn around and post that same listing
on Craigslist and then sit and wait and does his other work,
and sits and waits until somebody
can match his asking price, which
is typically $250 or $300.
And then when that person comes back--
it might be a couple days later-- he
sells it to them for that higher price.
So they're a bridge in time.
They're a bridge between the seller on a Saturday
and the buyer on a Monday or Tuesday.
Social distance, this is another problem.
And this is really the biggest problem of all.
And it presents the most opportunity.
Because, what you often have-- this
is a very typical pattern for social networks--
is that people tend to be with other people like them.
Birds of a feather flock together is the saying.
And so if you map the social structure,
what you end up having is a cluster of people
who all know each other here.
And then another cluster here and another cluster there.
But then you have these gaps between the clusters.
The structural holes is what sociologists call them.
And what structural holes do is they create problems
with transferring information, but they also
create opportunities for middlemen
who can act as bridges.
They're bridging these gaps in the social fabric.
So, an example, there's an entrepreneur
I spoke with who, when she was a college student,
saw an expectant mother, a heavily pregnant woman,
putting up fliers in this young woman's dorm.
She was looking for a nanny.
And she thought, there are so many college students,
I'm bound to be able to find a nanny here.
Now, this is in the Boston area, and there are just so
many college students in the Boston area.
It would be very inefficient to go and do this.
And it's just silly for an expectant mother
to be doing this or for anybody to be doing this.
And so you've got for these people.
There's a cluster of college students, many clusters
of college students.
And then you have many clusters of families.
And the families may talk to each other.
And the college students may talk to each other.
But they're separated across social distance.
And so what the entrepreneur did is
she saw this bridging opportunity.
This was in the early days of e-commerce.
And she started a service called Sittercity
that connects these two groups.
And now, we have many, many such services.
And that doesn't seem so remarkable.
But I'm just showing you this as an example
of how, even within the same geographic area,
you might have the need for a bridge.
All right.
So this bridge in social structure
is pretty foundational to solving
a lot of the other problems that I'm going to describe.
So let's take the problem of quality.
This is another problem that middlemen solve.
Whenever you're buying something,
you're kind of at the mercy of the seller
because the seller typically knows more
about what they're selling then you, as the buyer, do.
All right.
So classic example is a used car.
If you're selling a used car, you
know how well you've taken care of the car, what
problems it has, and so on.
And the buyer doesn't know that.
And this creates a real problem for both sides, actually.
This is what the whole market for lemons is about.
It's not just that the seller is suspicious of the buyer
and is at a disadvantage.
The buyer ends up being at a disadvantage
just against other buyers because-- I
mean, sorry-- the seller.
How can the seller prove that the car is in
as good a condition as the seller knows it to be?
How can they justify that price?
So this is where a middleman can help because a middleman is
doing trade constantly.
If you're selling a car, you're doing it once every few years.
But someone who specializes in selling cars
is doing it all the time.
So a really clear demonstration of how
a middleman can solve this quality problem
is apparent on eBay.
Because on eBay, it's all right there.
You can see who is doing the selling.
You can see a record of their transactions over time.
What everybody wants is a high feedback score on eBay
because it makes buyers trust that person more.
The problem is, it's very hard to build up
that kind of positive score, that positive reputation
on eBay, if you're only selling occasionally.
You see?
This is why being a middleman gives you an advantage
because-- one of many reasons-- is that you're
buying and selling a lot.
And so you can much more quickly build up
a reputation as someone who delivers on the quality
that they're promising.
So one of the people I interviewed for the book
is a woman who is a PowerSeller on eBay.
PowerSeller is eBay's designation
for its most active and reputable sellers.
She gave up a career as an appellate lawyer
to sell luxury designer goods on eBay
because, gosh, it was just a quick way
to become an entrepreneur.
And she's an expert on this stuff.
That's one of the keys.
You have to really be good at sussing out quality.
And she's staking her reputation on everything
that she sells on the site.
She really guards that average feedback score on eBay
because she knows that all her future sales depend on it.
So this is the advantage that a middleman has.
They're an expert in quality.
And they're really able to lay their reputation
on the line with every sale.
So I call this type of middleman, the certifier.
And, by the way, we often talk about people saying things
like, we don't need gatekeepers.
Gatekeepers are evil.
Let's bypass the gatekeepers.
Let's do our own thing.
Well, gatekeepers are actually very
valuable for this very reason.
It's in their interest to close the gates when crap is coming
through, and to open them up when
they see something of quality.
And when we're buying, as consumers of goods,
of cultural products, it's valuable to use
the services of the gatekeeper because we know that there's
a stamp of approval.
You're not going to see a lot of schlock
go through a great gatekeeper.
All right.
So the third problem is the problem of accountability.
Accountability is related to the problem of quality,
except with quality, you're just dealing with goods.
With accountability, you're dealing with services.
So let's say you need to hire somebody.
How do you know, regardless of that person's track
record or underlying quality, how
do you know that they're going to deliver
for you, that they're going to do a good job for you?
And again, you have the same problem
that the infrequent seller or the infrequent buyer
does in the certifier scenario.
If you're only selling something,
a service, occasionally, how can you
prove that you really are an honest, accountable person who
works hard and so on?
And, as the buyer, how can you ascertain that as well?
So again, middlemen to the rescue.
I'll give you an example.
A modeling agent, an agent who represents
models for photographers or brands
that need to hire models.
This is a person who's keeping both sides honest.
I'm using that as an umbrella term,
keeping both sides honest, to capture
that whole notion of accountability that
can take many forms.
So in the case of a model, being accountable
means that you're going to show up.
That's the main thing.
You're going to show up on time-- which, by the way,
in the modeling world means 15 minutes-- or at least
15 minutes-- before the shoot starts.
And for the other side, the photographer
or the brand who's taking pictures,
that they stick to the letter of the contract,
that they uphold the contract terms.
So if they are allowed to use these images in a catalog
because that's what they paid for,
the agency, the middleman in this transaction,
is going to make sure that they don't
slap on those pictures on a big billboard.
So the middleman is able to keep both sides accountable.
A model might still flake out, but she'll
be less likely to flake out when she's being represented
by the agency.
And the reason is that the agency's
in a position to have her brought in again
and again and again.
So she's got a longer future horizon
when working through the agency than if she's doing business
directly with a photographer.
Then she can say, uh, it's not worth it for me.
Something-- a better gig came up.
I'll just cancel on you or whatever.
So that's an example of accountability.
And it happens everywhere.
So, OK.
We've talked about these several problems.
I'll also present the problem of risk.
And, of course, all of these things
that I've been talking about-- the problem of quality
and the problem of accountability--
do have to do with risk.
And that is the risk that what you're buying
isn't what you thought you would get.
But, when I talk about the middleman as a risk bearer,
I'm really talking about a different kind of risk.
And that is the kind of risk that neither side can predict.
A lot of services in the on-demand economy
actually deal with this kind of risk.
It's those unpredictable fluctuations in demand.
When are you going to need to call up a car?
When is a car going to be in this neighborhood?
None of these services know when each of these events
is going to happen, but because they're
pooling all these resources-- they're
kind of building a portfolio, you
might say-- then in the aggregate,
they do know that at this time you're
going to have somebody who's going to need
a car in this neighborhood.
And you're going to have a driver that's
within a certain radius of this neighborhood.
So that's risk pooling.
That's the risk-bearing role.
And often, it has to work hand-and-hand
with these other types of risk, the quality problem
and the accountability problem.
The clear demonstration of this is
with a service called ZocDoc.
It's another one of these two-sided marketplaces.
It's sort of the OpenTable for doctor's appointments.
So if you need to see a doctor on short notice,
you can go to this website and find someone in your area,
in the specialty that you're looking for.
And for the doctors-- who are the ones who
are paying for this service because it's
free to the patients-- for the doctors,
this is a great service just like OpenTable
is for restaurants because they often
have last-minute cancellations.
And they'd rather fill that appointment slot.
So for each doctor, it's unpredictable
when that opening is going to show up.
But for the site, for the service as a whole, for ZocDoc,
it's quite predictable that, yes,
we can really count on there being
a certain number of openings.
And we can count on a certain amount of demand from patients.
And you have to keep those in balance.
But here's the interesting part is that it really
has to work hand-in-hand with the work of accountability
and the work of sussing out quality.
Because you're going to run into this problem.
Who do you think will be the doctors with the most openings?
That's the kind of risk.
Doctors who are not very good, who have not
been able to build up a very good reputation
in their community.
So everybody's going to have, you know, these flukes.
Like, you know, very honest reasons for having an opening.
But doctors who have a disproportionately high number
of openings, you have to wonder about these doctors.
And ZocDoc, as a middleman, has to protect themselves
from that kind of risk.
They have to weed out the bad doctors.
But they have to embrace the other kind of risk,
which is that nobody knows when you're
going to have an opening.
So this is where these reputation systems come in.
It's almost an industry standard now that all of these sites
have a way for users to evaluate the other side.
At least in this case, patients evaluate doctors.
And when doctors know that they're going to be evaluated,
they're going to be, presumably, on better behavior.
So they're not going to, maybe, keep you waiting as long
as they would if you were not transacting
through the middleman.
At least, that's the theory.
And it's in ZocDoc's interest to keep
that system working very well.
So that's the risk bearer role.
And again, it comes up in lots of other places.
There's another problem.
And I would call this the problem
of information overload.
I think Google certainly understands this problem.
It's like the official, you know, mission of the company
to organize all the world's information.
A lot of people don't appreciate this.
That the more information you have,
that's not necessarily a good thing by itself.
Because then you have the problem
of sifting through this information,
processing the information to make a good decision.
And Herbert Simon said, pretty famously--
you guys might have heard this quote--
that a wealth of information creates a poverty of attention.
And by attention, he just meant time.
Information consumes time.
If you've ever had to plan a complex trip,
you probably can relate to this.
Humans actually have a much harder time
than computers in processing information
because we get overwhelmed.
We get emotionally overwhelmed.
And so we just get paralyzed with indecision.
And so this is another rationale for a middleman to step in.
So I call this role the concierge role.
Just like you could do your own research on whatever city
you're going to through your iPad or whatever,
or you can go to a concierge who can just
cut through all of that and tell you exactly
where to go to meet your needs.
So that is the concierge role.
And I think computers are going to be
doing more and more of that, but we're not there yet.
We still can really benefit from a good concierge.
By the way, this is where I think real estate agents can
really step up their game.
We no longer need real estate agents
to just tell you what the listings are
if we're buying a house.
But we need them to guide us through this long, complex,
high stakes transaction.
And I think that they could be doing a better job at that
if they thought of their role that way.
All right.
Finally, the last problem that I think middlemen solve
is a problem that I'll just call self-advocacy,
the problem of self-advocacy.
Which is the problem of trying to advocate on your own behalf.
So we, just as humans, we have a hard time tooting our own horn.
Much easier if somebody else does that for us,
if somebody else introduces us, if somebody else
says how great we are.
And also the problem of pushing for what we want.
We're afraid of rocking the boat.
And one of the middlemen I interviewed for this book
really saw this problem in his industry.
And we already see this in other industries
like in sports agents, right.
They advocate on behalf of the athlete.
So not only does the athlete not have
to become an expert in negotiation,
but the athlete can just pretend that they're
just interested in the game, and let the business sign be
taken care of by the middleman.
The person that I talked to does the same thing
for young doctors.
So he's a doctor himself, a more seasoned doctor.
And he's seen both sides.
Middlemen often do have to be able to appreciate
the needs of both sides.
And he realized that young doctors
are at a disadvantage in negotiating
for their first job out of residency or medical school.
And the disadvantages-- well, there are many disadvantages.
One is they just don't even have the information.
They don't have experience negotiating.
But they're also hamstrung by the fact
that throughout medical school, it has been pounded into them
that medicine is a higher calling.
It's not about the money.
And you shouldn't show an interest in money
when you're talking to your future employer.
But hospitals and clinics take advantage of that norm.
And so they're able to get these young doctors to just sign
whatever the first offer is.
So this agent is able to step in and he's
able to say things like, well, I know
you really need an ER doctor.
And I know that the median pay in Kansas City for an ER doctor
is X. And so, let's just move toward X
with this person who's a young hire.
We don't know anything about his quality.
So let's just take the median salary as our target.
There's just no way that the doctor-- even if the doctor had
that information-- would feel comfortable saying
these things on her own behalf.
So that's a value that this role that I call the insulator
provides.
So what's interesting about all of these is they
all show-- they're all able to wield power
on behalf of their client, on behalf of their trading
partner.
Going back to this idea of being an admirable middleman,
a true partner.
But they're able to do it without any kind
of a formal authority.
It's not like anybody's reporting to this person.
But we're all gladly partnering with these people.
We're all happily using these services
because these businesses are providing
true value to both sides.
And I want to just leave you with this thought
about how this might apply to you and how can
you start to see yourself as a middleman as well.
So this diagram shows somebody's actual map
of their LinkedIn connections.
LinkedIn no longer offers this visualization service.
But if you were to manually map all your connections,
I would bet that your diagram, your own personal network
diagram, would look much like this.
And this person, right here in the middle,
is the user whose connections are being mapped here.
And all these colors, they refer to a cluster.
Remember, I was talking about a typical social network having
clusters and having bridges, and gaps between clusters
that can be bridged.
So, the blue cluster are all employees of one company.
And maybe the orange cluster is employees
of a previous employer that this person has worked for.
And, I think, the green cluster refers to some classmates
from college.
And the smallest cluster is typically your family members.
Again, all these people know each other, more or less,
because they're in the same social circle.
That's what we would say, social circle.
But they don't all know each other across clusters.
And in this case, you sometimes have people who overlap.
But the most prominent bridge is the user, him or herself.
This is the person who is a bridge across all
of these clusters.
So the idea here is that we're all separated,
but we're all part of a bunch of networks.
And if we can think of ourselves as being
that middleman, that middle person,
the person who can bridge all of these different networks,
then we can increase the value of our whole network.
And we can do it through these ways
of being not just a bridge, but also a certifier or an enforcer
of the accountability role.
Or, perhaps, an insulator when you're saying something
on behalf of a co-worker that they can't say
on behalf of themselves.
That's how you can be a more valuable middleman yourself.
[APPLAUSE]
AUDIENCE: So you mentioned that historically we
have viewed middlemen as something that we reviled.
However, clearly, like you said, most
of the services that we work with
are, essentially, middlemen.
Why do you think the social perception
has changed for middlemen?
Why don't we revile, like, Amazon or Google, as we would,
I don't know, an older middleman?
MARINA KRAKOVSKY: Yeah.
Well, there are a lot of answers to that question.
So first of all, middlemen as a group are still reviled.
I follow this word on Twitter because I
want to join conversations about middlemen.
And people still say things like, cut out the middleman.
This service cuts out the middleman.
And they're not seeing that, well, this service, actually,
is a middleman, too.
So, that's one thing.
I'd say, in general, people still
don't have this, like, negative view of middlemen.
A lot of people wouldn't see Amazon as a middleman.
Another answer is that Amazon I would
put in that golden quadrant of being a partner, at least
from the consumer's side.
If you talk to small manufacturers who
try to sell through Amazon, you might hear a different answer.
So I think that the ones that we look to as admirable ones,
we just don't call them a middleman.
AUDIENCE: So middleman is a more semantic thing then?
We just call things middlemen that we don't like middlemen,
but the middlemen that we like, we
don't realize that they are middlemen?
MARINA KRAKOVSKY: Yeah.
I do think it's largely a semantic issue.
Yes, that if we like them, we call them
a retailer or my trusted agent.
We don't use the word middleman.
I think so.
AUDIENCE: Can talk a little bit about the role
of universities as middlemen?
Because, to some extent, they're certifiers and networks and--
MARINA KRAKOVSKY: Yeah.
Universities are, definitely, middlemen for the reason
that you identified.
A lot of people would say that that's
actually their main role, is just to certify you.
You know, like, oh, you got into MIT.
Well, that's good enough for me.
I don't even care what you learn at MIT.
So that certifier role is really important.
They actually work as middlemen in other ways,
too, because there's a real-- at least, a two-sidedness.
You know, the idea of a two-sided market
where the buyers want to be where the sellers are.
The sellers want to be where the buyers are.
There's a way in which universities are like that.
So if you think about a good university,
they attract good students because they
have great faculty.
And then great faculty want to work at a university that
has smart, motivated students.
And there are other effects like that.
AUDIENCE: So, do you see any problems?
I mean, all of this sounds excellent.
But I'm thinking now of the physician's metric
that you mentioned.
And do such middleman services actually create
behaviors that otherwise would not be created?
For example, some people may think,
you know, oh, this is a bad doctor.
He treated me in a certain way.
But the doctor is actually-- his cure is performing well,
but his personal skills may not be up to par.
So--
MARINA KRAKOVSKY: Yes.
AUDIENCE: --he gets, like, a bad review.
And now you have this network effect
which may be multiplying this.
And suddenly strange things happen.
MARINA KRAKOVSKY: Definitely.
Yeah.
That is not a perfect system with ZocDoc.
I mean, doctors really think-- and a lot of people
would say this-- gosh, what business do patients even
have evaluating doctors?
If you were as expert at medicine as the doctor is,
why would you even need a doctor?
And you could say that about any of these middlemen services.
To a large extent, we turn to middlemen because we cannot,
we just don't have that expertise ourselves.
And so, I would say that these services
do have-- these rating systems do
skew the ratings in a certain direction
that they measure certain things more than others.
Clearly, things like the doctor's bedside manner
and how long they kept you waiting
weigh heavily, much more heavily,
than other factors that somebody else would find important.
Yeah.
It's not a perfect system.
AUDIENCE: Yeah.
I was just trying to think about the role of technology
taking over middlemen roles.
MARINA KRAKOVSKY: Yeah.
AUDIENCE: And I just wondering if you
have thoughts on are there certain roles
or characteristics of these roles of middlemen
that are inherently human?
That technology, maybe, conceivably
could never take over?
MARINA KRAKOVSKY: I really don't.
I thought about this question as well.
And I actually posed it to Al Roth.
You guys know Al Roth.
He's an economist at Stanford.
He won the Nobel Prize a couple years ago.
And he thinks a lot about middlemen.
Anyway, he says no, that really there's nothing you can say
that is like that because there's just no limit.
We don't know what artificial intelligence
is going to be able to do.
I actually talk a little bit about this in this book.
I mean, in the same way that we turn
to Netflix with its sophisticated collaborative
filtering system to recommend movies that we're really
going to love, there is no reason, in theory,
in my mind, why we can't someday turn
to a sophisticated piece of software that
will ask us natural language questions to find out
about our tastes and ask us to rate past experiences
and on the basis of that construct
the perfect itinerary for us.
That seems like an impossible challenge,
but, in theory, I don't see why it's not possible.
I'll say one other thing about technology.
You know, part of the reason that people were saying
that middlemen we're going to go away
is that the internet would reduce transaction costs.
The cost of finding what you need,
and finding out about quality and so on.
And it actually has done that.
It's reduced transaction costs.
And that's why we're able to book our own travel so often.
If you're just flying to LA or New York,
staying at a hotel you've stayed at before,
your transaction costs for doing that on your own
is really quite low.
And so it makes no sense to turn to a travel agent to do that.
But the thing about transaction cost
is that the internet has lowered transaction costs
for everybody, including middlemen.
And so middlemen bring whatever human capital they have.
Whatever skills and knowledge they have,
they bring it to the task.
And then for them, the technology is a complement.
So their transaction costs have actually
fallen faster than our transaction costs have
for the kind of special, complex transactions
that we can't easily do ourselves even
with the aid of technology.
MALE SPEAKER: Though that they can't never
go to zero marginal cost for transaction.
MARINA KRAKOVSKY: Yeah.
I mean, yes.
Yes.
There's always going to be-- yep.
No free lunch, right?
MALE SPEAKER: Let's thank Marina for her visit to Google.
[APPLAUSE]