Tim O'Reilly on What's the Future
Oct 9 2017

WTF.jpg Author Tim O'Reilly, founder of O'Reilly Media and long-time observer and commenter on the internet and technology, talks with EconTalk host Russ Roberts about his new book, WTF? What's the Future and Why It's Up to Us. O'Reilly surveys the evolution of the internet, the key companies that have prospered from it, and how the products of those companies have changed our lives. He then turns to the future and explains why he is an optimist and what can be done to make that optimism accurate.

James Bessen on Learning by Doing
Are workers being left behind when the economy grows? Is technology making the human workforce obsolete? James Bessen, author of Learning by Doing, talks with EconTalk host Russ Roberts about the role of learning on the job in the past...
Michael Munger on the Sharing Economy
Mike Munger of Duke University talks with EconTalk host Russ Roberts about the sharing economy--companies like Uber, AirBnB, FlightApp, and DogVacay that let people share their houses, cars, or other assets with strangers in exchange for money. These companies dramatically...
Explore audio transcript, further reading that will help you delve deeper into this week’s episode, and vigorous conversations in the form of our comments section below.


Ralph Casale
Oct 10 2017 at 8:24am

Enjoyed the episode. I was fortunate to be introduced to Toffler as a teen and have always enjoyed the futurist mindset. But … it did give me a bit of a chuckle to note that when I went to buy the book it was not available for a kindle reader, now my preferred platform. I went to buy the book direct from the author’s website, which links to the publisher, but … at a > 60% premium over Amazon. So …. in trying to better glean the future via this book, I feel I’m getting a lesson in what the future isn’t.

Oct 10 2017 at 4:12pm

Great episode!

Russ Roberts: Let’s make a list–and of course it’s a very partial list–of what had to be in place for Uber and Lyft and others–Airbnb being another example–for them to exist. … Which–that’s so fortuitous–just all those things had to be there for this to happen. Nobody who did those individual things planned that it would lead to Uber. Uber is a fantastic–and all these companies are fantastic examples of emergent phenomena that couldn’t be designed–that could have been, of course, at enormous cost; but they emerged through different pieces of design, but mainly pulling these different technologies and insights together.

Tim O’Reilly: Yeah; and there’s one other piece–I totally agree with that list. And it is astonishing when you look at the way these pieces gradually accumulated until somebody could put them together in a novel way.

Exactly so! That is what I’ve called Dynamic Distributed Design.

No one designs the whole top down. Yet each piece is a different piece of design. The pieces of design accumulate. At any time someone else may put them together in a novel way, which is more design.

The whole is not designed by one person or one committee. Yet all of it is very much designed — by many people over time. (No baker decides for all bakers. Yet every bakery is the result of design.)

How can it work to do Dynamic Distributed Design, i.e. for design to be distributed across people and across time?

One model is how we design and develop large complicated software systems like operating systems.

Russ Roberts: … This is something I started talking about starting around 2000, was that the Internet was becoming an operating system with services like location, like payment, like identity.

Tim O’Reilly: … So, again, this is this Internet operating system I’ve been talking about for the last 20 years, you know.

Another model is any economic market, which was mentioned in many ways (e.g. even designing a company within as a market).

I find it fascinating that both of these cases are mentioned in this episode as models to aim for in the future.

Roger Barris
Oct 11 2017 at 7:34am

Russ, another excellent episode. Thanks.

One comment about the retail business of Amazon. A thought experiment: if Amazon raised its prices by 10-15%, would you change your use of it? I know that the answer for me is “no.” I use Amazon primarily for its convenience, especially its broad selection, incredible client service, and the saving in time and effort it allows.

If Amazon raised its prices by this amount, I suspect that it would lose very little in the way of sales. It would also go from being barely profitable to being highly profitable since this extra margin would almost entirely drop down to the pre-tax profit line. In fact, the only way that I can justify its current stock price is the market’s expectation that, one day, the current or a future management of Amazon will start to “milk” the enormous retail cow it has spent a fortune to create.

Of course, this begs the question of why this hasn’t already happened since I think that Amazon is easily established enough to pursue this strategy. I think that the answer is that the current bubbly stock market has meant that management is not under pressure to produce profits — yet. Tesla is another example of this phenomenon, except that Tesla, unlike Amazon, has no path to solid profitability.

Glenn Mercer
Oct 12 2017 at 3:49pm

Good episode, as virtually always.

A couple of tangential remarks. I get the Lyft/Uber innovations: app, GPS telling me where my car is, preloaded credit card, ratings replace regulation, etc. But in the medium run (that is, before autonomous cars), taxi company main costs = driver + car + fuel, and Lyft/Uber main costs = driver + car + fuel. I am still trying to figure out how L/U can outcompete taxis, once they figure out app, GPS, cc, etc… see Flywheel and Curbed. (And don’t bring up medallion costs: in SF they are set at $250,000 by the MTA, and yet there is still a waiting list for them.)

Also, on the remark about Air BnB not getting into their own hotels, the company has actually started experimenting along these lines in a few places, notably Florida and Japan.

If there is another “theme” in tech these days, beyond “emergence” (as so well discussed in the episode), it may be “convergence,” as incumbents adopt features of disrupters, and vice versa. Thus Flywheel emerges from taxi fleets, while Uber introduces tipping and cuts way back on surge pricing. Thus corporate entities buy whole apartment buildings in NYC and convert them to Air BnB rentals. As economists often say, “Solve for the equilibrium.”

Kevin Ryan
Oct 13 2017 at 8:24am

Very good episode again

I also agree with Glenn Mercer’s comment about convergence

Indeed, from a UK perspective, it seems to me that a fair amount of talk about the newness/uniqueness/sharing economy features of the activities of companies such as Uber and AirBnB is overstated.

For example, one of my bugbears is that many people describe Uber as a ride sharing service, as if it were a bunch of commuters sharing the car of one of their members. Aren’t a very large % of journeys just like a conventional taxi ride, with a driver whose full time job is driving that vehicle; or is that not the case in the US?

Similarly what percentage of AirBnB stays are actually staying in the spare room of owners who are present, and eating with them – which I recall is what was described in the Mike Munger episode back in July 2014? While there are a lot of variants on what else is happening – whole property let out v single rooms, supply of food and other servicing or not, owners present or not, how many units are owned by the host, I suspect that an awfully high percentage really amounts to models that are very familiar to us – at least in the UK as I would not want to speak for other markets. But there is nothing new about short term holiday lets, for example; the concept of ‘buy to let’ is now very well established in the UK, and if owners are letting out multiple rooms and providing food etc then that sounds like a hotel to me?

Is there any serious analysis of the balance of AirBnB lets, and how they vary from place to place?

Todd Kreider
Oct 13 2017 at 10:01am

Tim O’Reilly said:

You know, the thing that’s interesting is the speed is extraordinary; but, there’s a really interesting economics lesson in it, also, which is that new technologies typically take 20, 30, 40 years to really affect the economy and productivity. Amazon has been around now for, you know, 22 years. And it’s really–the major impact on retail beyond bookselling are really only starting to be felt now

Russ also stated something similar to the first half of the above later in the interview but this isn’t correct. It was true in the first half of the twentieth century as the diffusion of electricity, telephones and automobiles was much slower but as the graph in the Atlantic article shows, the diffusion of home computers, cell phones and the internet have been much faster. (“The 100-Year March of Technology” April 7, 2012)

With the case of Amazon, it had a major impact on the book world within two or three years. Of course it didn’t affect the food industry in 2000 or even 2010 because that wasn’t on the radar in any substantial way. I imagine that Amazon has had an impact on retail for a while but numbers are needed instead of O’Reilly just proclaiming it has taken 22 years.


Oct 13 2017 at 10:11pm

Glenn Mercer: “But in the medium run (that is, before autonomous cars), taxi company main costs = driver + car + fuel, and Lyft/Uber main costs = driver + car + fuel. I am still trying to figure out how L/U can outcompete taxis, once they figure out app, GPS, cc, etc… see Flywheel and Curbed. (And don’t bring up medallion costs: in SF they are set at $250,000 by the MTA, and yet there is still a waiting list for them.)”

Costs such as medallion costs cannot be ignored. Just because there is a waiting list, that doesn’t negate the fact that any owner still needs to set prices high enough to bring in enough to cover all costs, not just “driver + car + fuel”. Otherwise it is a losing proposition, plain and simple.

As for “driver + car + fuel“, the new technologies allow for more efficient operation (= less fuel) by bringing drivers directly to paying customers instead of having them drive around looking for someone.

But even if taxis used similar apps, their “driver + car” is not the same. Taxi companies have a fixed pool of cabs that they pay money to own and maintain. That cost has to be made up.

Uber and Lyft do not own a fixed pool of cars. Their pool can grow and shrink over the course of the day, giving them a variable number of drivers and cars and a greater ability to satisfy peak demand times with rapid responses over a wide geographical area by using a larger fleet during those times. That makes them attractive compared to waiting a long time for a cab. But they can also scale down during hours of low demand. They aren’t paying for cabs that sit empty without fares during low demand times.

In the long run, any kind of taxi company with a fixed fleet size will have trouble competing with an adjustably sized Uber or Lyft.

See the chunk of the conversation starting at 20:38 for the discussion of this.

Kevin Ryan
Oct 14 2017 at 3:27am


Fair enough, but driver-owned cabs are common enough in many markets and would seem to have the same feature of the co-ordinating business not incurring fixed costs.

Similarly the old technology of ordering cabs by phone obviates the need for such vehicles to drive around looking for fares.

Indeed, when I dropped out of university 45 years ago I was thinking of earning a living working as an owner driver for such companies in London – really a lightly regulated form of taxis known as minicabs.

Admittedly I do not know how such systems manage variable demand in practice – infinitely variable pricing was certainly not a feature; and I can understand that if you are able to gain market dominance then you will have more cabs who can get to each customer more quickly than a would a small system

Oct 14 2017 at 12:30pm

Keven Ryan,
If your main point is that some of these ideas are not quite as new as some people imply and that there are predecessors (driver-owned cabs, telephone requests instead of apps), then that is good balance and perhaps another example of the idea that there is nothing entirely new under the sun.

I would just observe that wouldn’t be an argument against the effectiveness of this model. It would merely be making a good case that others have previously used some aspects of the Uber/Lyft model. Yet Uber and Lyft are now combining ideas in a significant way.

Combining variable supply with variable pricing to meet variable demand can be quite powerful. When many people need a ride and want it soon, more drivers and cars can be incentivized to meet that need. (Some find a way to take offense at this arrangement. But would it really be better to have fixed supply and fixed pricing with a shortage of the drivers and cabs when they are most needed?)

BTW, Amazon Web Services (AWS), which is discussed in the chunk at 42:46, also includes some variable pricing aspects that allows people to get a break on price by doing computations at low demand times when there is unused capacity. In that case, supply is not variable but variable pricing can redistribute demand.

Kevin Ryan
Oct 14 2017 at 2:59pm


Thanks for your response

Yes, you are right about my main point. And I do agree with you that Uber etc have been combining ideas in a significant way. But I also am sympathetic to Glenn Mercer’s thought that the end result will be both traditional ‘taxi’ companies and the new ones carrying on business in a similar converged way.

And on a specific point I do not think I am opposed to combining variable supply with variable pricing to meet variable demand.

That said, there are clearly some things about the Uber model that irk me. I’ll try to avoid a long rant and boil these down to the following:-

1) As a former regulator I have a bias, which I am sure is not shared by many Econtalk listeners, against new entrants being able to overthrow well-established regulation. This is not to suggest that, say, taxi regulation, should not change or even be swept away. But that it should result from the reasoned decision by a community of the need for that change, as opposed to locally based authorities and incumbent businesses being outmatched by the lobbying of global businesses that are not exactly frank about what they are really doing.

2) As a (car owning) resident of London, what I observe is a city whose streets are clogged with traffic. And with a mayor who wants to reduce car use. Of course the congestion has not been solely caused by taxis and ‘private hire vehicles’ as our mayor calls them. But the growth of the latter has contributed. And it looks like the only practical solution lies in more use of/ more efficient public transport. So our ‘taxi’ problem looks to be one of excess supply – not a need to find ways of accommodating larger numbers of cheaper journeys.

Kevin Ryan
Oct 15 2017 at 6:50am


Thanks for your response

Yes, you are right about my main point. And I do agree with you that Uber etc have been combining ideas in a significant way. But I also am sympathetic to Glenn Mercer’s thought that the end result will be both traditional ‘taxi’ companies and the new ones carrying on business in a similar converged way.

And on a specific point I do not think I am opposed to combining variable supply with variable pricing to meet variable demand.

That said, there are clearly some things about the Uber model that irk me. I’ll try to avoid a long rant and boil these down to the following:-

1) As a former regulator I have a bias, which I am sure is not shared by many Econtalk listeners, against new entrants being able to overthrow well-established regulation. This is not to suggest that, say, taxi regulation, should not change or even be swept away. But that it should result from the reasoned decision by a community of the need for that change, as opposed to locally based authorities and incumbent businesses being outmatched by the lobbying of global businesses that are not exactly frank about what they are really doing.

2) As a (car owning) resident of London, what I observe is a city whose streets are clogged with traffic. And with a mayor who wants to reduce car use. Of course the congestion has not been solely caused by taxis and ‘private hire vehicles’ as our mayor calls them. But the growth of the latter has contributed. And it looks like the only practical solution lies in more use of/ more efficient public transport. So our ‘taxi’ problem looks to be one of excess supply – not a need to find ways of accommodating larger numbers of cheaper journeys.

Oct 15 2017 at 1:42pm

Kevin Ryan,

I’m most surprised by your second point. Is too much car traffic really something “about the Uber model” that irks you? In other words, does the Uber model make the problem of having too many cars worse or better?

Certainly any shift toward mass transit would “reduce car use”. For those who choose to use cars, however many or few that is, they could a) use their own car, b) take a traditional taxi, or c) go with Uber/Lyft rides.

If they go with (a), that means they will need parking for the whole time they are at the destination. Shifting from (a) to (c) would reduce the number of parked personal cars.

Shifting to app directed rides and variable fleet size instead of hailing traditional fixed quantity taxis would mean either fewer taxis parked on the street waiting for people to show up or fewer circling taxis hunting for a fare to flag them down. Each of those taxis without customers will be somewhere trying to get business.

Finally, to the extent that a fleet scales up in quantity during times of need and down at other times, that would mean that car traffic is kept more to a minimum as needed. Plus, if variable pricing is used, the Uber model can encourage use during cheaper times of lower demand. Those who can shift the time of their travel can be incentivized to travel when demand and traffic volumes are lower.

So, while it is good to promote public transit, any excess of car traffic seems to me to strike more against the models of traditional taxis and personal ownership than it does against the Uber model. Because the Uber model is more efficient, that means it would help the traffic and car use problem as compared to other kinds of car solutions.

Oct 16 2017 at 9:32am

O’Reilly sounded intelligent until his comments that we now have the “divine right of capital” and that all the “productive gains [are] going to Wall Street” revealed his comic-book grasp of economics. The behavior of Google, Facebook, Amazon etc. is pretty consistent with behavior of big, incumbent corporations of the past. If they seem less innovative, less nice, and less dynamic than they were as startups–well, that’s one reason we need startups, because disruption almost never comes from dominant incumbents, especially when they’re highly profitable. And weren’t both Russ and O’Reilly talking about the fact that Amazon was foregoing profits?

To suggest, just after an election in which a declared socialist barely lost one party’s nomination and in which a protectionist won the general election, that the Zeitgeist venerates capital is silly. Marginal University has some good videos on industrial organization. If O’Reilly wants to understand the economy and what drives its participants, he ought to watch a few.

Steve Wood
Oct 17 2017 at 4:24pm

Agree: Great/Timely discussion. I do have a couple comments, one positive, the other not so much:
1. Re Eric’s comment, invoking Mr. Robert’s list of factors that contributed to the emergence of ride-sharing; this links to the recent Auerswald EconTalk podcast (another stimulating visit that seemed to range far afield, but in a welcome way) on the rise of populism … in terms of “combinatorial” factors, which links to “collision density,” that came up in an earlier EconTalk podcast, possibly Geoffrey West on Scale – Cells & Cities. Gotta love overlaps, linkages, synchronicities.

Also, I was pleased to hear Mr. O’Reilly later invoke the idea of “surplus capacity.” That is the one crucial factor I would have added to Mr. Robert’s checklist.

2. RE Tech/Amazon creating jobs. The conversation was rather glib about this point. Those lost retail jobs are being replaced by warehouse jobs which are part-time, low-wage, low-interaction, low-reward endeavors that will likely be replaced by robots in the near future. So the top line numbers may not honor much deeper issues.

Steve Wood
Oct 17 2017 at 5:07pm

Apologies re earlier comment: Reference to Geoffrey West, Cells & Cities, was to a Waking Up With Sam Harris podcast, 07/14/17. https://www.samharris.org/podcast/item/from-cells-to-cities

Kevin Ryan
Oct 18 2017 at 10:30am


I will try to be clearer on my second point. Unlike the first point, this is not meant to be a criticism of the Uber/Lyft model, but represents opposition to an increased use of ‘taxis’ of all kinds (in London).
So I would like to think that normally I am in favour of innovations that deliver a better service at a (generally) lower cost – which is what I think this model is promising/delivering – and I am trying to explain why I am not in favour in this particular case.
There appears to be some evidence that the number of ‘taxi’ (broad sense) rides has been increasing in London. So this is increasing clogging. And I am suggesting that the success of Uber etc is contributing to this. And this then is making my preferred form of travel, driving my car, less attractive. I accept this is at least partially selfish.

I can, of course, agree with you that a switch to a more ‘clogging efficient’ form of car traffic is a good thing, ceteris paribus, but I am sceptical it outweighs the volume effect.

If I can comment on your points on the various forms of car use:-
1) In London we have a fuller range of alternatives:-
a) Traditional taxis hunting for fares;
b) Traditional taxis on ranks;
c) Traditional minicabs at base;
d) Pre-booking of traditional taxis and minicabs – by telephone or, more latterly, online
e) Uber/Lfyt rides
f) Drive own car

If we define clogging efficiency as distance of customer’s journey/total distance travelled by vehicle, I can accept that intuitively Uber/Lyft are likely to be more efficient than other forms of ‘taxi’ ((a) to (d)), although the extent to which this is true in practice will vary.

2) I do not really understand why any form of taxi would be considered more efficient (in the clogging sense) than a customer driving his own car. The former will always have to travel some distance without passengers in practice. While if I can’t park at or very close to where I am going, then I will not take my car. I don’t really see being prepared to spend time driving around looking for a parking space.

Further, there is no real gain to clogging if there are fewer parked cars – as London does not have the type of streets where additional driving lanes would be freed up if the parked cars disappeared.

3) I am also sceptical than variable pricing would of itself have any material benefit to clogging in practice. I cannot really think of any circumstance in which a lower ‘taxi’ cost would cause me to change the time of a journey

Butler T. Reynolds
Oct 19 2017 at 9:26am

We think that we are living in a unique time of unprecedented change.

Spend a while browsing the archive of old photos at Shorpy.com and today’s changes don’t look quite as dramatic.

For that matter, visit Disney’s Carousel of Progress! 🙂

Comments are closed.


EconTalk Extra, conversation starters for this podcast episode:

This week's guest:

This week's focus:

Additional ideas and people mentioned in this podcast episode:

A few more readings and background resources:

A few more EconTalk podcast episodes:



Podcast Episode Highlights

Intro. [Recording date: September 6, 2017.]

Russ Roberts: Our subject for today is Tim O'Reilly's new book, WTF?: What's the Future and Why It's Up to Us.... Now, your book is quite amazing. It's part memoire, part history of technology. There's also some preaching about how we might mold the future. And to start with, I want to talk about how much your book reminds us and makes us pay attention to how much and how fast the world has changed in recent years. Here's a quote from the book where you talk about going back, way back to the past--to 1998. And 1998, it's hard to remember--it's less than 20 years ago. So, here's the quote--of course, some of my listeners--this is not the quote yet--some of my listeners won't appreciate this quote because they are 19. But, for those of you who lived through 1998, it will remind you of just how much things have changed. Here's the quote now:

Software was distributed in shrinkwrapped boxes with new releases coming at best annually, often every two or three years. Only 42% of U.S. households had a personal computer, versus the 80% who own a smartphone today. Only 20% of the U.S. population had a mobile phone of any kind. The Internet was exciting investors but it was still tiny, with only 147 million users worldwide, versus 3.4 billion today. More than half of all U.S. internet users had access through AOL [America Online]. Amazon and e-Bay had been launched three years earlier but Google was only just founded in September of that year.

End of quote. Less than 20 years into this revolution, the speed of it is really rather extraordinary.

Tim O'Reilly: Yeah. You know, the thing that's interesting is the speed is extraordinary; but, there's a really interesting economics lesson in it, also, which is that new technologies typically take 20, 30, 40 years to really affect the economy and productivity. Amazon has been around now for, you know, 22 years. And it's really--the major impact on retail beyond bookselling are really only starting to be felt now. And we're only starting to see the articles that say, 'Oh. Maybe it isn't going to be the death of retail. It's just different.' Because, there's starting to be articles about Amazon actually hiring more workers, or just e-commerce in general leading to different jobs rather than, 'Well, yes, people standing in stores are being put out of work.' But it's actually kind of amazing: numbers of jobs being created in warehouses, in delivery. And this amazing capability that we didn't have where you could say, 'Oh, I need something,' and have it show up. Sometimes even the same day. And that, to me, is this parable that we also saw in the Industrial Revolution, and that we forget in the early stages of a new economic revolution. And that is, that technology lets you do more. And, you know, so, going back to--I always like to think about the Luddite Rebellion. And, imagine that, well, you know, we started using these machine looms and they say, 'Great. Now we can make everybody their one suit of homespun very, very quickly.

Russ Roberts: Yeah.

Tim O'Reilly: Instead, no: it was all of a sudden, you know, people could have more clothes. Fashion became something that wasn't just for the very, very rich. Ordinary people were able to change and update their clothes. And now, of course, clothes are a commodity and a disposable. And, so, part of the question that I ask in my book, which really is meditation mostly about Silicon Valley technologies but more broadly about technology as a whole, is: What do we have to do differently to accelerate the progress towards making the winds that we get from technology come more quickly? And, get out of the mindset that says, 'Oh, this technology is going to destroy jobs'--because, yes, it may well do that. And instead start asking, 'What new jobs will it create? What things were previously impossible that it will let us do?' You know, back in the early, Industrial Revolution, they had no idea that we would fly through the air; that we would be able to move goods all around the world incredibly cheaply. That we would build a tunnel under the English Channel. That we would build skyscrapers half a mile high. None of that was in the imagination. And now, here we are, with this enormous failure of imagination about what might the cognitive technologies of AI (artificial intelligence) and what's happening on the Internet make possible in the way of a future economy? And so, I don't have answers for that. But, I do know that there are certain design patterns that you see in the great companies of how they actually drive forward. And one of those things is they don't simply try to cut costs. They don't see technology as simply labor-saving. They'll see technology as a way to do more. And part of my call to action in the book is: Let's get out of this financial, market-driven, this is all about improving, you know, cost structures and making companies more profitable so their stock prices go up; and instead say, 'What problems can we solve now?'


Russ Roberts: One of the things your book reminds us of is the--how much of our world that we enjoy today, that's information-based or digital--is, was created by passionate people. Passionate about a solution, passionate about a technique, passionate about creating a tool. I want to go back to something--I think you mentioned this twice in the book; and it grabbed my attention, because I've mentioned it before on the program--that when Amazon first started, I was very uneasy entering my credit card into an open space on my computer, thinking, 'This could be bad. This could end up really badly.' And I was talking to the CEO [Chief Executive Officer] at the time, and I remember him saying, when I said--he said he ordered a lot of books on Amazon; and I said, 'You do? You give them your credit card?' And he said, 'Sure.' And I thought, 'Oh. Okay. Maybe it's all right.' And I went and did it; and I found out, I could get all these extraordinary books. But I want to raise the possibility, which you allude to in the book, that creating that confidence--that you could give your credit card to a stranger--not literally to a stranger--to a website, entering it into a website's database was okay. That changed the world in ways that you can't begin to think about. Just an obvious thing that that does is it allows Uber to exist the way it does, which is extremely pleasant, not to have to take your wallet out, worry whether you have enough money, go through that transaction, and just--it's done. To have your credit card quote "on file" is just a wonderful thing, that now we just take for granted. But it was not obvious that it was going to happen.

Tim O'Reilly: That's right. And I actually talk about that in the book, this idea that often these technological leaps forward require thinking the unthinkable. And we don't even know until--in retrospect it just seems so obvious: Why would you not do this? But, the example I give regarding Uber and Lyft is: We actually had connected taxi cabs; and what did we do? We put a TV screen in the back and a credit card reader in the back. It was just like the old days of movies where you filmed the stage play and then bit by bit they figured out, no, actually you could move the camera around, you could do different takes and then stitched them together with editing. And we invented movies. And, same way: it took a while. And people can even have the idea, but it doesn't all come together. So, there's a guy named Sunil Paul who in 2000 filed a patent for a whole set of ideas about how a smartphone could be used to coordinate cars, could support fractional car rental--all kinds of things that were really part of today's world. The thing that that really illustrates is you can have an idea and the world is just not ready for it. There weren't enough smartphones. And Sunil kind of took a run at it and couldn't make it work--couldn't get funding. And then later he came back to it, but by then it was too late: Uber had started. And it's also really interesting if you look at how that whole market evolved: you see how markets kind of invent things together. People get an individual piece and then someone else gets a different piece, and then they get assembled in magical new ways. So, Travis Kalanick and Garrett Camp came up with that marvelous idea that a smartphone in the hands of a driver and a smartphone in the hands of a passenger could be used to coordinate pickup in real time. That was that magical user experience of Uber. But they thought about it for black cars--it was sort of an app for rich people. And it was Sunil who came in with a company called Sidecar, which was rapidly copied by Lyft, that kind of had this idea--ordinary people coming and doing this in their own cars. Uber didn't actually adopt that till a year later. And so, then this market sort of exploded. But, it was really these two streams of ideas that were both needed, because, you know, John Zimmer and Logan Green, the founders of Lyft, had previously been working on a startup called ZimRide which was inspired by the jitneys in Zimbabwe. It wasn't actually named after John Zimmer but after Zimbabwe. And it was a web app for coordinating long-range travel. And then when they saw Uber, they went, 'Wow. This would be a great way to get to our vision of a future transportation network that's a 21st century version of those jitneys that supply public transportation in Zimbabwe.' And, so, you see this sort of magic of somebody has an idea; and then it makes somebody else have a different idea. And then there's this game of leapfrog. And I've watched this throughout my career. And kind of back to credit cards on the Internet--think how long it's been. It was, what? Twenty years almost between Amazon One-Click and Uber's realization that you could do one click in the real world. Apple was pushing Apple Pay, which is like, 'Wow. We'll just use your smartphone as a token instead of a credit card.' It was [?] say, 'No, we'll just store your credit card and [?] the act of consuming the service will trigger payment automatically.


Russ Roberts: So, I made a list, actually, responding to what you wrote about Uber and Lyft, and you had a lot of interesting observations on them and we may come back to them. But the point I was thinking about, to echo what you said a minute ago, was: Let's make a list--and of course it's a very partial list--of what had to be in place for Uber and Lyft and others--Airbnb being another example--for them to exist. So, lots of people use smartphones, for sure. You mention that. People being able to pay for things with their phones--so, a credit card system that generated enough trust through experience and software that you could be comfortable that you wouldn't be constantly having your identity or money stolen. GPS [Global Positioning System]--location, on your phone--which is crazy. I mean, that's amazing. We don't even think about that, either, but that's an amazing thing. A little bit scary, too, but also amazing. The underlying algorithm to align people with cars without having to have like a dispatcher figuring out, 'Who is closest?' That's a mind-blowing achievement, I assume. I assume some really thoughtful people did it. An interface on your phone to make that whole thing pretty straightforward, really shockingly straightforward. And then, finally--and this may be as important as all the rest--a way to ensure trust without heavy, top-down regulation or monitoring of drivers using the rating system of both customer and driver. Which, as you point out--I think you say this--it came from e-Bay. Which--it just--that's so fortuitous--all those things had to be there for this to happen. Nobody who did those individual things planned that it would lead to Uber. Uber is a fantastic--and all these companies are fantastic examples of emergent phenomena that couldn't be designed--that could have been, of course, at enormous cost; but they emerged through different pieces of design, but mainly pulling these different technologies and insights together.

Tim O'Reilly: Yeah; and there's one other piece--I totally agree with that list. And it is astonishing when you look at the way these pieces gradually were accumulated until somebody could put them together in a novel way. And again, some of those actually--you had to think about them--they really are being delivered as services. This is something I started talking about starting around 2000, was that the Internet was becoming an operating system with services like location, like payment, like identity. And the applications like Lyft and Uber are examples of that in the real world. We now have this sensor-infused world where things are connected to the Internet, where there is this data services that are provided by third parties. So, Uber doesn't build its own payment service. Lyft doesn't build its own payment service. This is just something that you can get from someone else. So, that kind of comes back to this really interesting idea in economics of why it takes 30, 40 years for a new technology to really show up in the productivity status, statistics. There's a wonderful book by an economist named James Bessen--

Russ Roberts: --Former EconTalk guest. We talked about that book. It's a great book.

Tim O'Reilly: Yeah, it's a great book. And just this idea that you have to have--a lot of pieces come together. It wasn't just those weaving machines. He talks about the weavers in the 1840s, Lowell. And you have to have a critical mass of people who have the skills to use the technology. You have to have people who can fix the technology, who can extend the technology. But I wanted to come back to this one thing that was missing from your list, which is that critical mass of buyers and sellers.

Russ Roberts: Yep.

Tim O'Reilly: And, it's interesting, because the reason why Uber and Lyft are still losing money is that they are competing very hard to get and maintain that critical mass. And, I'm not sure--it's sort of interesting because it's hard to say what would happen if they weren't spending all that money. If you look at the earlier waves of the Internet, things kind of evolved at something of a natural pace. I mean, sure, Amazon raised a lot of money. But, Google grew pretty organically. Facebook grew pretty organically. And, so, the market just--it wasn't force fed. And one of the things that I see that's really interesting in this ride-sharing market is, these companies are basically force-feeding the development of the labor ecosystem. And the thing that's sort of interesting about that is, I wonder to what extent that large amounts of capital that are raised have actually led them to distort the natural economics of that market. Because, they basically have been trying to lower costs for consumers, subsidizing that with venture capital. But then they end up, the driver is going to end up making enough money so they end up having to pay all kinds of incentives--

Russ Roberts: bonuses, yeah--

Tim O'Reilly: which--bonuses. And that's very costly. And I do wonder if they shouldn't be paying effectively efficiency wages: Pay more for the drivers so that they stick around; they don't have to spend as much money on recruitment; and don't spend quite so much money on, 'Well, we're just going to try to kill the taxicab companies by undercutting them on price.' And find a natural, you know, balance in that market. It's hard to say, though, because it's also true that the immense amounts of capital that they've spent are really spent on educating people about this new possibility.

Russ Roberts: Yeah. No, that's a good point.

Tim O'Reilly: And a lot of people who say that these companies can never be profitable really don't understand that, there will come a point--this is just the way you do it. And, this is the way transportation happens; and it becomes much more ubiquitous. Which is what they've been really pushing for. And at that point, it will be very interesting to understand to what extent the companies really find the right balance between the benefit to consumers and the benefit to the workers that they depend on. So, there's a real interesting--one of the books that influenced me a lot when I was writing my book was a book called Who Gets What--and Why by Alvin Roth.

Russ Roberts: Another EconTalk guest, by the way. [?]

Tim O'Reilly: Yeah. And I was introduced to that book by one of the economists at Uber, Jonathan Hall. And he said it was totally shaping--he was like, 'You've got to read this book. It's totally shaping my [?].' But it's really about this idea of, you know, you need to build a critical mass of buyers and sellers.

Russ Roberts: Yeah. It's about the challenge of matching. And a thicker market is easier than in a so-called thinner market.

Tim O'Reilly: Right. And so, it's really, it's so interesting, because, as I wrote this book I realized how much technology and the economy are becoming deeply, deeply, deeply intertwined.


Russ Roberts: With our culture. With our habits. With our--I mean, I recently had Benedict Evans on; we were talking about autonomous cars, electric cars. And I said something like, 'So when the autonomous car takes you to the grocery,' and he said, 'Do you think you'll go to the grocery?' and I thought, 'Well, actually, probably not.' So many cultural experiences, habits, as you point out--years, it might take decades to change the way we think about how to use these tools we have. It's going to change so many things about how we live, how we shop, how we interact with each other. I think, for example, the autonomous car is going to change a lot more than what we think of as taxis, obviously, or our own driving. It's going to change everything--just the way the car changed everything relative to the horse. You make a great point, by the way, in the book: when you talk about the labor model of Uber and Lyft, you say,

There are those who argue that Uber and Lyft are simply trying to avoid paying benefits by keeping their workers as independent contractors rather than as employees. It isn't that simple. Yes, it does save them money, but independent contractor status is also important to the scalability and flexibility of the model.

And then you go on to make the following point, which I think is fantastic:

Unlike taxis, which must be on the road full time to earn enough to cover the driver's daily rental fee, the Uber and Lyft model allows many more drivers to work part time and take passenger requests simultaneously from both services, leading to an ebb and flow of supply that more naturally matches demand. More drivers means better availability for customers, shorter wait times, and far better geographic coverage. These companies are able to provide 5-minute response time over a far larger geographical area than traditional taxi and limousine companies.

That's a great point. I've never seen it made before. And I just think, again, the creativity and the getting-used-to and the what's-going-on underneath this is just so hard to know. And you are worried about whether they pay enough, whether they should have lower turnover. My first thought is, 'Well, they'll figure that out. They have a pretty good incentive to.' They might not figure it out for a while. It may take 'em, their culture and CEO [Chief Executive Officer] and leadership and management may struggle with those issues. But I think they'll get the hang of it. And I think--as you say, I think it's the future, whether we like it or not. I like it. I think it's unbelievable. It's changed my life when I travel, in so many pleasant ways.

Tim O'Reilly: Yeah. So, one of the things that is super-important to remember or to understand is this great phrase from, that I learned, from these consultants back in 2000, called the beams[?]. And it was really--they were talking about business models as, the way that all of the parts of the business work together to create marketplace advantage and customer value. And, it's so often you see people who don't really understand how all of the parts of a business have to work together. And so, for example, taxi companies, they say, 'Well, we just have to have an app.' Thinking that the app that lets you summon cars is the magic. And that's one small part of the magic. Because, you do have to have that critical mass of drivers. If you are a taxi company and you basically say, 'Well, we have enough taxis, with our taxi licenses and this licensing regime, to meet average demand--because we can't have enough to meet peak demand, because they will be idle most of the time.' Right? So they can't do it, with their model.

Russ Roberts: They can't do it.

Tim O'Reilly: Because, that's the beauty of the model, is that you have more drivers when you need them and fewer when you don't. And so, the challenge, then, is, if you realize that that's actually fundamental to the ability to deliver, you realize a couple of things. One is the huge challenge that the transition to autonomous vehicles provides for companies like Uber and Lyft. A lot of shallow analysis says, 'Oh, well, those just get rid of all the cost of the drivers. Won't that be great?' And I say, 'Well, no. They're going to have to have enough cars to meet peak demand.' Right? So, logically, they are going to have--

Russ Roberts: Totally different. Yeah.

Tim O'Reilly: [?] A lot of the time, a lot less of the time, unless they find other uses for them. And, so you say, actually they are going to have to feather in humans and autonomous, and they are going to have to figure out how to balance which--do they take the best routes, or the worst routes? The worst times? the best times? There's all kinds of interesting challenges there. And, of course, probably the very best solution would be for Uber and Lyft to transition to a world like that of Airbnb, where other people provide the autonomous vehicles in the same way they do today. Because, that swarming marketplace model isn't backed[?]--the very center of their business model.

Russ Roberts: It's a great point. Again, I don't think Airbnb is going to move into their own hotel chains as a way to expand their business. They've got a good thing going. And that's the fact that we all have, many of us have unused capacity.

Tim O'Reilly: Right. And so, that's why I think what's going to happen--and, again, the logic of this says to me that in autonomous vehicles, these companies shouldn't be saying, 'Well, we're going to develop our own,' unless they feel like they can develop their own and then give it away--for example, Google said, 'Well, we're going to do Android and then we're going to give it to all the phone manufacturers so there's a Google-friendly phone operating system,' that they could say, 'We need to have an Uber- or Lyft-friendly autonomous vehicle system.

Russ Roberts: Yeah. Well, they could go that way. It's going to be--I think we'll both see this in--we'll find out, this, when it ends. It's going to be--probably, I hope--I'm 62--but I think I will.


Russ Roberts: I want to shift gears. I want to move to Amazon. You have a lot of interesting things to say about Amazon. You mentioned it earlier, about whether all this VC [Venture Capital] money, all this venture capital [?] poured into ride-sharing and before poured into Amazon--I always say, 'Well, this is great. I get cheap books, and I'm going to keep buying them. All these venture capitalists are paying.' And that really improved my life a lot. But I think--Amazon's interesting because--and you have some interesting observations on it. There's a temptation for most of us on the outside to see Amazon as just a big retailer that once sold books but now they sell lots of other things. But, you write the following about Jeff Bezos. You say,

How Jeff took the idea of Amazon as a platform out of the realm of software and into the realm of organizational design ought to be taught in every business school.

Explain what you mean by that.

Tim O'Reilly: Well, I think--there's a couple of things. First, the actual background. At some point, Jeff realized that Amazon needed to become a platform. I had actually given a talk about how Microsoft had beaten a whole series of challenges, from Lotus all the way up to Netscape, you know, in the PC market, because they were a platform provider and they owned the platform.

Russ Roberts: Explain what that means. Let's use Amazon as an example. What does it mean for Amazon to be a platform?

Tim O'Reilly: Well, literally, Amazon decided, at some point around 2003, 2004 they were going to take this big e-commerce application that they had built and they were going to turn it into a set of services that would be used to compose the application. Like, earlier, remember when I was talking about Uber and Lyft, I said, you know, they don't provide the payment. You know, you can get payment services from Stripe, or again--companies like that. You don't have to do it yourself. Well, Amazon basically said, 'We're going to decompose our own application and do a set of services.' So that when somebody wants to do computation, there will be a computation service. When they want to do storage, there will be a storage service. And then they basically turned the company inside out and said, 'We're going to offer these services to the outside world.' And, as a result, they built this Amazon Web Services business, which, you know, is now, a $10, $15 billion dollar business--you know, it's really established the whole new cloud computing market. And, really, it was because Jeff realized that if somebody else became the platform, that Amazon would be vulnerable in the same way that, you know, Lotus and WordPerfect and Netscape had been vulnerable to Microsoft, who owned not just applications like Microsoft Word and Excel and Internet Explorer, but also owned the underlying operating system. Then they could basically use that to make, to privilege their own applications. And so, you know, Amazon has not used [?] to privilege their own applications. But they did do this amazing thing, where they built themselves into a sort of a platform that they could then deliver their application with, which improved their own services. But they opened it up to the world. So, that's Part One. But, a second part that's behind that quote is that they realized, ultimately, that being an internet-scale platform requires a different kind of internal company organization. Because, you decompose your massive application into a bunch of small services. And those services are managed and created by small teams. Amazon calls them 'two-pizza teams'--you know, that is a team that is small enough to be fed by two pizzas. Right? And each team has some function that they are trying to optimize. And, it's--so effectively they've made an internal marketplace--almost managed by the invisible hand. I mean, obviously, there's a guiding spirit where we're trying to do this big thing. But, how people do the pieces, you know, is not really tightly specified.

Russ Roberts: I think a very good book could be written about how large corporations have tried to introduce market-like forces within their command-and-control regimes. It's a fascinating example, this one.

Tim O'Reilly: Yeah. Absolutely. Amazon has really done a great job of it. The thing that we have to really understand is that today's companies are infused with the digital. You know, when you look at a company like Amazon or Google or Facebook, it is the application. One of the earliest insights I had about Internet applications was that they still had people inside them. You know, this notion that, you know, unlike, a program like Microsoft Word, which is, as you quoted from me earlier, would ship every 2 or 3 years on a CD, an application like Google is constantly updated. Literally, it's a set of business processes where it just ain't--the activity of billions of people in real time, and then turning it into a unique work product, i.e., search, ads, whatever--Amazon in adjusting this huge workplace of products, continually updated, continually changing, you know, having people comment on the products that are featured are partly a result of what sells, what people like. You know, so lots and lots of people are giving a product 5 stars is going to appear at the top of the listings. And it's this dynamic, wonderful business process. And, it's not--you know, when people say, well using marketplaces inside of companies, this isn't like, you know, Accenture has done some really interesting work where they have built inside the company a jobs marketplace. You know, a little bit like Upwork for occasional--that's the small end of this. The big end is the entire company is infused with digital. It is an application. Amazon, all the people in it, you know, are part of Amazon not just the app that you see. And so in some sense, humans and machines are working together to deliver that service. It's all the way from, not just the programmers and the editors who are editing the content of the site, but also the people in the warehouses, the people and the delivery drivers. A [?] into this vast cybernetic, cyber organism.


Russ Roberts: Yeah. Well, you have a great quote. It's one of my favorite quotes in the book, and I think it's incredibly deep. You say,

If you think with the 20th century factory mindset, you might believe that the tens of thousands of software engineers in companies like Google, Amazon, and Facebook spend their days grinding out products just like their industrial forebears, only today they are producing software rather than physical goods. If instead you step back and view these companies with a 21st century mindset, you realize that a large part of what they do--delivering search results, news and information, social network status updates, relevant products for purchase, drivers on demand--is done by software programs and algorithms. These programs are workers; and the programmers who create them are their managers. Each day, these managers take in feedback about their workers' performance, as measured in real-time data from the marketplace. And if necessary, they give feedback to the workers in the form of minor tweaks and updates to the program or the algorithm.

End of quote. Standing that on its head it's just--that's a mind-blowing, beautiful way to think about it. And it emphasized--I forget, long ago, an EconTalk made this point that we're all just a giant focus group, constantly giving information to these products, and they are being tweaked and improved and changed. Of course, maybe not always improved for our benefit. I think mostly; but we'll maybe at the end talk about whether we should be worried about this. But, the process is a brand new phenomenon. It's an incredible phenomenon. And, as you point out a number of times in the book, and as you just said: It's hard to talk about where the human and where the technology start and end. They are just intertwined. They are augmenting each other.

Tim O'Reilly: Yeah. And you pick a key word here, which is 'augmenting.' When I think about the key--I started wrestling with this question of income inequality and how are we going to make a better word using technology, and it really struck me that the key design pattern of technology is augmentation. And I don't mean augmented reality, although that's going to be a pretty interesting set of new developments. But, just as the technology is the 18th, 19th, and 20th century were about augmenting our muscles, from the 20th into the 21st century we were really about augmenting our minds. And, you augment in a word to increase our capabilities. And that is the thing that--again, I kind of feel like I have this call to action which is at the heart of the book, which is: How do we use this technology? How do we get out of this narrow idea that the only function of technology is to productivity in the narrowest sense simply by doing the same thing more cheaply? I talked with one investor who said, 'I invested in a startup, an AI [artificial intelligence] startup that will get rid of 30% of call center jobs.' And I said, 'And I talked with Abby Johnson of Fidelity and she wants to make her call centers work better.' She doesn't want to get rid of the workers. She wants to avoid the situation I was in where I was trying to transfer one of my wife's accounts and I spent two hours on the phone. And still didn't get the right information.

Russ Roberts: Yeah.

Tim O'Reilly: You know, it's like make it better.

Russ Roberts: Well, I think the best entrepreneurs do that, obviously, and the best investors understand that if you eliminate 30% and give worse customer service you are not going to be a very effective innovation.

Tim O'Reilly: Right.


Russ Roberts: I want to come, in the last part of our conversation, I want to talk about these issues of sharing these gains more widely and what's going to happen to the jobs of the future. But, I want to add, before we leave Amazon, first I want to apologize to listeners but these are so interesting--I want to read one more little story about Amazon and then I want to ask you a question. Quote:

In an Amazon management off site, Jeff Bezos once famously responded to a suggestion the company needed to improve communication between teams. "No. Communication is terrible," he said. The reason can be explained by the old joke. One person sits and drinks. Two person clink and drink. The more people you add, the higher the ratio of clinking to drinking. What you want, said Bezos, is a situation where people clink only with the people doing shared work, not with everyone they touch. This is simple math. Communication gets worse as team size grows.

That last part may have been you, not Bezos. But that's the point. And it just--

Tim O'Reilly: Yeah. Just to be clear: The only part of that that's a quote from Jeff is, "No. Communication is terrible." The joke and the rest is me explaining the point.

Russ Roberts: That's just so great though. This is crazy, but it reminds me of the insight we've talked about here before on the program of FedEx realizing that if you go to a hub, you need a lot fewer planes than if you have direct flights between every place you are going to deliver and pick up packages, because you are going to need a lot more planes. It's the same clinking and drinking problem. Ironically or not.

Tim O'Reilly: Yeah. Although, there's a couple of interesting counterpoints to that, and one is in airlines, the hub and spokes system versus Southwest, which did do point-to-point, continues to do point-to-point, and actually for a long time was the only profitable airline.

Russ Roberts: Yep. Good point.

Tim O'Reilly: And the second thing is that I think is--because again, they realized that serving an unserved market was a market opportunity rather than just serving the major markets more cheaply.

Russ Roberts: Yup. True.

Tim O'Reilly: But, there's another piece, which I've been chewing on, and it really has to do with: How does this new economy end up creating more jobs? And there was a great example I came across recently in Alexis Madrigal's wonderful "Containers" podcast. He has an 8-part series about container shipping. And, Episode 4, he talks about the changing economics of coffee: because container shipping led to this world in which you could take massive, bulk products, load them into a container, ship it halfway across the world at very low cost. You got rid of all of the loading and unloading--the longshoreman jobs. Huge job losses that happened in the 1970s and 1980s. And now, the jobs are coming back. Guess what? Because we've gone beyond commodity coffee back into this differentiated marketplace where people say, 'This is single-origin coffee from this particular region--Guatemala, or Honduras, or Sri Lanka. And it's roasted by this particular roaster in Emeryville, California.' And so, what happens is: container comes into the port; but then all the containers have to go to this new distribution center, where they get unloaded, because, you know, Blue Mountain Coffee is getting a certain amount of coffee from 15 or 20 different individual growers somewhere else around the world. So, you've ended up re-putting labor back into the process, because we have also added into coffee the idea and the experience--the aesthetics--of uniqueness. And that's one of the themes that runs through the book--is this wonderful idea that Clayton Christensen put out in 2004 as the 'law of conservation of attractive profits'. And I saw it in the software industry--and I won't go into where I saw that there. But I see that as the key to the next economy. Because, as one thing becomes a commodity, something else becomes valuable. So, coffee was commoditized all the way down to those 5-pound cans; and then the second wave of coffee was, 'Well, we're going to differentiate, and we're going to have better coffee and unique roasts.' And we saw the rise in people like Starbucks. And now you see all these specialty coffees, that people pay more for. And people more--sure, food is this massive coffee and people pay for the admixture of human aesthetics. And even in a world of robotics and AI, I think that can be the key to a future economy of prosperity. Because it's ultimately about us sharing stuff with each other, and persuading each other that the stuff we share is worth having.

Russ Roberts: Yeah. That's really nice.


Russ Roberts: Before we leave Amazon, I want you to teach me something, which I got a glimpse of in the book, which I think most of us don't have any idea about. So, you talked about how their cloud service, AWS, Amazon Web Services, or Servers--what's?

Tim O'Reilly: Services.

Russ Roberts: Services--how that initiated the cloud revolution. Well, for me, naively, until I read your book, I always thought: 'Well, I'm into the cloud. I use Dropbox to save my photographs, and I keep documents there; and it's nice to have a backup, and that's really great.' And I have no idea where those servers are that are storing all that stuff. I assume they are out in someplace remote with a lot of air conditioning keeping them cool. And, you point out that that's not what--I mean, Amazon makes Netflix possible. I didn't realize that. So, talk about that. And then, finally, if you can remember, talk about whether Amazon is going to be profitable in its current form, because the recent numbers I saw--they are making a lot of money, but it's all from AWS. They are just kind of limping along, barely breaking even on this unbelievably large retail presence.

Tim O'Reilly: Yeah. Well, a couple of things. Like, Amazon has built, has continued to invest in the future. Yes, they did become the AWS company in some ways, that has become the most profitable part of their business. Retail in general is not that profitable. Right? So, I don't think that saying that Amazon--

Russ Roberts: They can't make it up on volume, if you are losing money. So, they are trying to do that, it looks like.

Tim O'Reilly: No, I don't think so. I mean, I think, you know, they are basically--they are holding it together pretty well. But, let me come back to your question about AWS. Yes, actually Dropbox was originally built on top of Amazon Web Services. Netflix uses Amazon Web Services to deliver its service. So, again, this is this Internet operating system I've been talking about for the last 20 years, you know. Just like the PC [Personal Computer, non-Mac-type] had this platform of Windows, AWS really became the platform which tens of thousands of startups built their services. And, of course, there's now competing clouds--Google Cloud, Microsoft Azure. And you can, if you are a big enough company you can do it yourself. But you don't have to. I think, for example, Dropbox has now internalized their own storage platform, because they are big enough to do that. But, the enabling technologies, the platform technologies, are key to the advances of civilization. You know, when you think about an economy in general--roads were a platform technology. Electricity is a platform technology. Running water is a platform technology, and garbage collection is a platform technology. We even use the term 'garbage collection' in computer programming: it's part of what operating systems do. They get rid of all the--you know, stuff that gets left around by badly-performing programs. And so, anyway, Amazon has built this amazing business layer there, created this new market. But I think that they continue--their growth in retail, I think is still actually incredibly powerful part of who they are as a business. I wouldn't think of them as, 'Well, they are really just the AWS company.'


Russ Roberts: Let's talk about the future, and the worries that a lot of people have. You give some examples in the book, more than one, of how--you just gave another--of how employment often expands unexpectedly. That is the history of humanity--literally. As you point out, I think, very early in the book, the work doesn't disappear. The jobs do, but the people don't die. They go off and they can do the new work that's still out there. There's new things to do, when technology and tools advance. That's been the history of humanity. But some people worry--not unreasonably--that maybe it's different. Maybe this is--it's one thing to augment muscles with machinery, but augmenting brains with artificial intelligence may limit what many people can do productively to serve each other. Which is what a good economy does. So, one of the things, obviously, is education to help us make that work better. So, I'd like to hear what your thoughts are on education; and just in general, what do you think needs to be done that won't take care of itself? I tend to believe in the power of things taking care of themselves. But, this may be different. We'll find out. But, what are your thoughts on what we can do explicitly and directly to try to help people cope with these changes?

Tim O'Reilly: Well, first off, education is, has always been central to progress. And, a wonderful quote from Robert Putnam, author of Bowling Alone and many other books--but, I was in this Markle, sort of multi-year process called Rework America, and Bob Putnam said, "All of the advances in our civilization have come when we invested in other people's children." And he kind of went through--you know, universal grade school education, universal high school education, you know, the GI Bill [Servicemen's Readjustment Act of 1944; GI = Government Issue--Econlib Ed.]--and each of them led to actually a more productivity economy. So, education is a huge lever for making the world a better place. And the problem we have is that we have a sclerotic education system, where we are still, despite all the changes, thinking first of all that you get done with it, whereas in fact you have to keep--continuous education is the world of the future. The kinds of things that you have to teach are very, very different, and they change very, very slowly; and because of course we now have all these cognitive enhancements. And we are still effectively teaching people--you know, there's this analogy--we're doing the equivalent of teaching people how to drive a plow, push a plow, in a lot of the things we teach. Because the skills that are really needed today are very different when computers can remember things for us very easily, when they can do math, arithmetic, for us very easily, can approach subjects very differently. And we need to approach them through a task orientation. So, for example, you know, a very useful course would be: How do you survey the media landscape and figure out what's true and what's not? That would be far more useful than many of the things that are taught in grade school. I just read a wonderful account of a teacher who is doing that. So, there's that kind of thing. But also just this idea of continuous education, on-demand education. We're going to get to a point that's not that dissimilar, that fanciful scene in The Matrix where Trinity says, 'I need to know how to fly a helicopter,' and it downloads into her brain. I watch my teenage stepdaughter figure out how to do things. She goes and looks it up on YouTube, and--done. She's figured out how to make something or do something that she didn't know before.

Russ Roberts: Right. Our kids do that effortlessly.

Tim O'Reilly: They know that on-demand learning is the way that you learn. And so, the question is: How do we teach kids and give them access to the right kinds of skills and the right kind of knowledge of what's available out there.

Russ Roberts: A lot of that comes naturally to them, like the example you just gave. I was going to say, our kids do that effortlessly. I do it when I'm up against a stone wall. I couldn't get my toilet seat off: I'm changing out a toilet seat, and I thought, 'Oh, I'll just look it up on YouTube.' And, of course, there were 10 good ideas I hadn't thought of for how to get my toilet seat off. And my dad, he just--he's 87--he just calls me and says, 'Do you know such-and-such?' And I'll say, 'Well, I don't, but you could look it up on Google.' And sometimes I wonder--I like to think he just likes to call me. But, it could be he didn't think of Google. Because he does say--it seems to be actual surprise--oh, yeah, he didn't think of that. There is a real challenge for the idea of what we're going to do with formal education, what do we need and what should be taking place in it; and our current system is remarkably unflexible in trying to help us figure out what that might end up looking like. It's, I think, a real problem.

Tim O'Reilly: Yeah. It's sort of interesting, because, if you actually think about how education would best happen, I think we could get away from the regimented schooling entirely. It's sort of interesting, because I have a large Catholic family, many of whom, the next generation, were home-schooled. And, there's this community, you know, of--my brothers and sisters live close together; they have lots of other friends with large families. And, they home school. Because, my sister says, 'I'll teach biology,' and somebody says, 'I'll teach math,' and somebody else says, 'I'll teach English.' They have enough kids in the neighborhood that they put on a Shakespeare play every summer. Actually, two, sometimes, in a summer, with making the sets, learning the lines, doing--making the costumes. And they actually spent--my observation was that they spent less time educating their kids than I spent with my kids on homework. And the kids came out fine. They ended up going to college, getting good jobs. And so, there's a whole question of: We assume that the way we've been doing it is the only way to do it.

Russ Roberts: Yeah. It's not a good idea.

Tim O'Reilly: And, here's this thing that's so interesting: is that, there's so many areas where, if we gave people a different mental model of how to do things and we figured out market mechanisms so people could get paid, or it might be that the market fails, and we need a different way of funding it--because, of course, we fund education publicly. We don't make everybody do it themselves--

Russ Roberts: Slightly expensive to both, K-16.

Tim O'Reilly: Yeah. What would happen if we just blew that up and said we are going to do it very, very differently? Or figured out: Where are the experiments that we could support?

Russ Roberts: I have four kids; three of them are either out of college or in college. The fourth is a senior in high school. And a part of me wants them not to go to college. But, it's really hard to--it feels risky. It is risky, obviously.

Tim O'Reilly: Well, when I think about my own life--I went to Harvard and I got a degree in classics; and it's certainly been a wonderful gift to me to have done that. But, everything in my professional life, I learned after college--by doing it. And that's part of what my own business is really about--creating learning resources for people who teach themselves the things they need to know. We publish books. We run conferences. We have an online learning platform called Safari [Safaribooksonline.com], which is, tens of thousands of e-books, training videos, live online training. Because, basically we realized that in our, my universe of computer, people have to learn new skills all the time.

Russ Roberts: All the time. Yeah.


Russ Roberts: Let me ask you a different question about the future, which is on people's minds. The economics version of this--people are worried about profit margins. They seem large. I'm not sure that's the case. But, they seem large. And one way you get larger profit margins is monopoly, or less-competitive force from your neighbors, your competitors. We are in a world right now where--at least right now, there seems to be one really good search engine. There's a couple of places you can get a driver--call a car on demand--Uber and Lyft; there are few others, also, Via, I know of. But that might end up being one. And we're talking about the fact that they are investing large amounts of money to get that thick market going. And so there might be one survivor of that. There's one really big social media place called Facebook; there's a lot of smaller ones, Instagram and Twitter, too. But Facebook is increasingly becoming where people--or had become; now, obviously, it's in question. But there is some worry that these very large companies are going to have an inordinate amount of power over us. Recently it was in the news, a think tank that's funded by Google did some things that maybe made Google happy that don't look so good. And then we have the Google memo, James Damore[?], whether that was a good thing or a bad thing. Part of me says: it's a private company; they can do whatever they want; none of my business, even. Part of me says: they are really getting kind of hard to say no to. My usual answer to this is: If you don't like 'em, don't use 'em. Use DuckDuckGo. Use a different email service. There are many--there are still alternatives. But, are these companies too powerful? Is it something you are worried about?

Tim O'Reilly: So, I have a somewhat different view. First off, I believe that many of these digital platforms do have a tendency towards natural monopoly--this, network effects that make it just simply better for everybody to be using the same one. Now, the question ultimately may be: Should they be regulated monopolies or not? But, more urgently, my advice to them, is: Make sure that it works for your marketplace. And, you know, if you think about these things as marketplaces, not just--right now, too often I think these companies think, 'Well, we have customers.' But they don't also think, 'Well, we also have suppliers.' You know, because all of the people who are creating content on the Internet, for example are Google suppliers, as well as their customers. And if you realize that it's a marketplace that you have to make work for both sides, then you kind of invest in that ecosystem. And instead what I see is this pattern that, you know, Microsoft sort of--IBM [International Business Machines] showed first in the computer industry and then Microsoft--which is, as they became more powerful they tried to take more and more of the market for themselves. Not realizing that's a short-sighted behavior. Because, if people can't make a living in your marketplace, they go somewhere else. And so, you know, there's a self-interest. So, Walt Mossberg once said to me--and I can't believe I didn't put this quote in the book. He said--he was recounting a conversation he had had with Steve Ballmer when he was CEO of Microsoft. And he said, 'Steve, if you guys would just dial back the greed only 5%.' You know, 'people would like you 100% more.' And I think that Google is right at that point--where the market is going to turn on them. And I think that the myth that it's good enough--if you are helping customers, that's all you have to worry about, is part of it. But an even more pernicious myth, I think, in our economy, is that if it's good for shareholders then you don't have to worry about all the rest of it. Because, in fact, you really do have to look at an economy as an ecosystem. And, when you have a monoculture where, you know, one company becomes too powerful, and they forget that when they are that ubiquitous they actually have to be an enabler for everybody else, as opposed to simply--you are no longer just a player in the market trying to get as much as you can. You are the market. And so, therefore you have to actually keep a level playing field. You have to make opportunity for others. And you can't take it all. You know: that's critical. And, for me, the biggest offender in that list is not Google or Facebook or Amazon. It's our financial markets. You know. We are literally--I remember there was this--I forget who made the statement, it was somebody at JP Morgan, you know, after the American Airlines was going to give a raise to their employees, and it was like, 'They can't do that.' You know, it was like, 'They are robbing the shareholders.' And I was like, 'You've got to be kidding.' You know, the amount of the productivity gains that have gone to, you know, Wall Street, the corporate shareholders, and not to, you know, the people who were working in all these companies as a result of this sort of idea that somehow--you know, I refer, in the book, I said, 'You know, we laugh at the idea that the idea that the Kings had this divine right to own things." And now we have this divine right of capital. And we are going to look back on that and go, 'No.' An economy should be for everyone. And when you have one particular group that says, 'I'm going to take it all for myself,' and that it eventually falls down. You have a revolution. You have social unrest. And it's sort of short-sighted behavior. And so, we actually have to actually understand that great companies, you know, that become platforms for the rest of society, have a responsibility for those that depend on them.


Russ Roberts: Yeah. The problem is--and you are opening remark on this topic illustrates it. It's evidently hard to do. You mention IBM, then Microsoft. I think Google's next. Where, it just--doesn't come: Here's the problem: When you don't have competitors, it's very hard to keep focused on doing a very good job. And you get a little more focused on getting bigger.

Tim O'Reilly: Yeah. Well, yes and no. But you do have competitors. You just don't see them.

Russ Roberts: You just don't see 'em. Correct.

Tim O'Reilly: Part of the reason why it's so important to take the long view. Microsoft did have a set of competitors. They were all of the people who they had disenfranchised. And, believe me: If Google ends up disenfranchising too many people, they are going to breed the next generation of competitors.

Russ Roberts: My guest today has been Tim O'Reilly. Tim, thanks for being part of EconTalk.

Tim O'Reilly: Love it. You are such a great interviewer. Thank you very much.