Bill Gurley & Malcolm Gladwell in Conversation | SXSW Live 2015 | SXSW ON

Hello, everyone. This is Bill. I'm Malcolm. We're going to talk about a series of random things over the next hour. I told Bill backstage that this would work best if he treated me with a mixture of skepticism and hostility, and I'm hoping he lives up to that. Those of you who don't know, I write books. Bill is the tallest venture capitalist in America. He's at Benchmark. And we've known each other for a number of years now. I run into you randomly in places and we have conversations. And so we thought we would just replicate. So we're really going to talk about sports. But before we do that, we might talk on some more boring things, serious things first. Last time I saw you, we talked a lot about healthcare, and I thought it would be really fun to start with the question of whether your world, the technology world, can help us fix our healthcare problem.

Yeah. I've been involved in a number of companies that have used the resources that Silicon Valley is helping create—the smartphone infrastructure, the internet, the cloud—to try and make industries more efficient, things like OpenTable and Uber that have been successful. And you look at the technology and what's possible, and your immediate reaction is, 'But of course, there must be hundreds and hundreds of opportunities for us to solve the healthcare problem.' And so about a year and a half ago, I tweeted, you know, 'I'd love to roll up my sleeves and see if I can find one of these opportunities and help an entrepreneur in this field.' And I met with, I don't know, 100 companies? And I became more and more skeptical as I went through the process.

And the real problem is, and I don't think entrepreneurs realize this, but there's an assumption of market forces when you do a startup. Like, you expect customers to pay for value and to not pay for bad things and to want to be more efficient. And the physics are just completely mucked up in the healthcare system. It's like those drawings where you can't tell which way is up. And it's—it's everywhere you turn, it's like that. So, I'm going to—I want to tell everybody, I bet most people don't know this, even though if you do a Google search, you'll see that it's absolutely true. So, in 2009, when the government passed the Reinvestment Act to try and get America going, or whatever it was called, they put in place a program where a doctor is paid

believe this or not, paid $44,000 to implement EHR, electronic health records. So it'd basically be like telling a construction company, if you put in NetSuite, you get paid 50 grand. And it's just shocking to me. So the payments for these incentive programs, because I just looked it up to get the recent update, $29 billion. So our government's paid $29 billion for doctors to implement software that they weren't naturally implementing on their own. And of course, the question that comes to your mind is, 'Well, will they use it? Or are they just saying they're going to use it to get paid? Or is it any good?' Yes. So, two things about that. Putting it in was Meaningful Use 1, that's 44K. Because they're not sure you'll use it, Meaningful Use 2 is proving that you're using the software you put in place in Meaningful Use 1, and that's another $17,000 for the doctor.

And it's insane. It's asinine. And if you go online and do searches for Meaningful Use, you'll find spreadsheets where they list the features required to be in a qualified EHR program. So you basically have some government bureaucrats sitting around somewhere going, 'These are the features that should be in the EHR program of the future.' And then everyone has to implement exactly that or their customers won't qualify. And if you go on the websites of the companies in that space, they have huge sections of their website dedicated to how you can qualify for this stuff. And so that's the exact opposite of what entrepreneurs do. The exact opposite.

And by the way, almost any doctor you talk to hates using these kinds of digital health records and will make a very compelling argument about why it makes the practice of medicine worse, not better.

So it's tough. It's like really tough for entrepreneurs to navigate.

Another thing I ran into is—well, give me an example of an idea that would have worked if it was a real market and which you had to, which wouldn't have worked under the current system.

I think there's a number of things that could be done to make the process more efficient.

An idea that someone came to you with that would have worked if we had a real market.

Oh, sure. There's a number of entrepreneurs that have built really, I think, some of the most compelling EHR programs that the doctors like the most, fail to get traction because they don't have the salespeople out there to tell you how to get the 44K. There's one called DrChrono in Silicon Valley. Beautiful iPad-based, super quality, but the market's so distorted that you're hesitant to even bet on a good product. That's my point about how the physics are all messed up. I think price discovery is something that could be super powerful if consumers were price shoppers, but they're not, because they don't pay for healthcare, you know?

Well, the metrics, I was fascinated, there's a little subspecialty of, of social science which deals with reputation, how people assess reputation. So people have a sense of what a quote-unquote "good" hospital is. So a couple years ago, a guy did this really beautiful paper where he took 10 objective measures of hospital quality and tried to figure out whether they correlated with measures of hospital reputation. And there's zero correlation. You know, a hospital with a really good reputation is Columbia Presbyterian in New York. Columbia Presbyterian in New York, according to the last CDC data, is worse than average on four out of six measures of hospital infection rates. In other words, it's the kind of place where you go, get one of those infections, and you nearly die.

So why does it have a good reputation? And you're left with, well, maybe it's because it's got the word "Columbia" in its name. Maybe it's because it has really impressive buildings. Rich people put their names on them. I mean, it's completely irrational.

I think the glaring counterexample, which you and I have discussed before, is Singapore. The US spends 17% of GDP on healthcare and Singapore spends 1.7. It's a 10x delta and they have relatively reputable healthcare system. The key thing they do is they make everyone a payer. If you're high income, the government only pays 20% of your bill, and if you're low income, the government pays 80% of your bill, but everyone's a shopper all the time, and—and that makes every provider a seller all the time. And I think that's the kind of market dynamic that makes other industries improve over time. But hold on, it's not necessarily the case that the less we spend on healthcare, the better off we are. There is such a thing as too much, but also such a thing as too little. One percent strikes me as—I mean, I don't know, strikes me as—

You could be right, but we've got a system now that's ballooning due to largesse because the hospital groups are in a war with the insurance carriers and they're both trying to get bigger and bigger and bigger so that they have power against one another. Where I live, Stanford is grabbing up every general practitioner around and putting them under the Stanford umbrella. And you're seeing Stanford hospitals that are 40 miles from the Stanford campus. And so the system's growing because they want the employer not to be able to count them out. And so now they have ultimate leverage against it. But I looked at an app recently for MRI where I live, and the independents were at the bottom. There were 30 different providers of MRIs.

The independents were at 400. Stanford was at the top 30th at 3,200. So you had an 8x delta in the price of an MRI. Yeah. I mean, it just highlights how mucked up the system is.

Yeah, yeah. Are you saying there's nothing, or not nothing, nothing substantial that can be done to fix the system absent some kind of...

Radical reformation.

Institutional fix, or I mean governmental fix?

A lot of people are optimistic that we move towards high-deductible. So Obamacare is basically a high-deductible plan. All the employers I know would love to move there. They almost need it to be forced because they're in a competitive dynamic to try and grab employees, and they don't want to—they don't want to look like they have a worse plan. But if—if we move to high deductibles in any—all but the acute cases, that could create shoppers out of everyone, which I think will have a huge impact. And—and a lot of smart investors I know, like some of the PE firms, are—are buying up these urgent care facilities because they believe they'll get—once people move to high-deductible, that they'll get a bigger share because they're more customer-oriented. Yeah, yeah.

I want to, before we—and I realize we don't want to spend the whole time on healthcare because we do need to spend the time for sports, but I want to make one last point, which is I'm always curious when I talk to people like you about whether you make the error of assuming that more of the world is amenable to technological fixes than is actually the case. So, let me just give this one example with digital health records. So, when I talk to friends of mine who are doctors and who use these things, they all hate them. And it's not clear that they hate them because they're badly designed. They hate them conceptually. And they hate them conceptually because they say, 'Here's the problem: you're the patient.

I'm the doctor. Before the digital health record, I'm talking to you, listening to you, emoting, holding your hand, whatever. Then I would have, you know, if it was 1970, I would talk into a Dictaphone and my secretary would transcribe it, and that would be the—right? Now, they say what you're doing is you're on the keyboard and you're not looking at the patient, and the patient gets really upset. And what you've done is you have, in the cause of efficiency, disrupted the very thing the patient really wants in the encounter, which is a personal connection with the provider.'

Now, is it possible that this is simply

an area of deeply important personal interaction that is not amenable to technological—in this specific area, amenable to technological?

I would like to believe those things aren't at odds with one another. In other words, if you can make the bureaucratic piece melt away, you could increase the kind of intimate piece of it all—you know, instead of filling out forms. I mean, it was paper before it was typing, right? And to the extent—but, so look, I think there's a number of ways you could improve efficiency. You shouldn't have to—every single medical place you go to, you shouldn't have to fill out 10 forms on a clipboard. That's ridiculous. But I think the bigger opportunity for having massive impact on healthcare is the data, right? There's no one that's aggregating all the data, doing the proper analysis. So, I actually think to the extent we could get more of the data aggregated, we could have much bigger impact on different root causes, that kind of thing.

But once again, the regulation fights against it. A lot of the incumbents use HIPAA as a blocker to prevent data exchange.

I always thought the problem with—I have the same reaction to the word "healthcare" as I do to the word "Africa", which is to say it is a term which is used to obscure understanding and not enhance. Africa is a meaningless word. It describes an extraordinarily vast and diverse area, and there are, you know, one end of Africa has less in common with another part of Africa than it does with us, right? So when we use that term, we're grouping together a group of cultures, countries, whatever, that effectively have nothing in common. Healthcare's the same thing. Why do we insist that someone undergoing the most complex kind of brain surgery ought to be grouped under the same rubric as someone who gets a cut stitched up, three stitches for a—like, it's absurd.

They have nothing to do with each other except that the person who performs the operation went to the same kind of school.

I think that's true, but I think you're seeing a separation. These acute care cases, even when I talk to employers that are running self-insured plans, 60% of the cost for 4% of the employees, and they treat those things dramatically different than the way they're managing maintenance care.

It's not, institutionally. One of the biggest mysteries, if you talk to people in the healthcare world, is why does the hospital continue to exist? It doesn't make any sense. There's no reason anymore why you should be delivering babies in the same places you are treating colon cancer, right? Why? No one else does that. And if you dig into it, they'll tell you actually it's simply a function of the way the payments are made by Medicare.

I believe it. It's messed up.

Do you have anything helpful before we—except that we should...

I mean, seriously, I think high-deductible plans could move. You need something that simplifies the whole system and brings back market forces. They're just absent. And maybe not for the brain surgery, maybe not for every element, but for a large majority of it, that's what you need to do.

David Goldhill, who wrote this really brilliant book on this very point.

The very best book I've read on the topic, if anyone's interested. It's called Catastrophic Care.

Yeah. He has a really interesting idea, many interesting ideas, one of which is that the normal trajectory with insurance is you keep adding things that are covered by the insurance. He wants to do the opposite, that you have an insurance plan and every year you stop insuring something until you're left with just the—just things that are actually—insurance is supposed to be for unforeseen catastrophic events. You don't insure the painting of the front door of your house because you know you'll have to repaint it every 10 years. You insure a fire. And only in medicine do we insure, you know.

Well, and the doctors try and create more and more and more chronic instances and get them on the schedule, yeah, yeah, so that they can... We could reverse, yeah.

Have we dispensed with healthcare? I think we should move on. We should move on. Uber, your company is a—

I'm an investor and a board member at Uber.

And you're on the board? Yes. And I was intrigued by this notion, which we touched on when we were chatting yesterday, of whether we are properly accounting for the transformational nature of something like Uber. In other words, are we underestimating how significant a transformation of our society could come about from things like this?

Based on what I've seen in San Francisco, where we've been the longest, it is transformational in many, many ways. I'll give you a few examples. Probably the largest job creator in the town today.

So how many jobs have been created by Uber?

There's about 300,000 Uber drivers right now, and they're adding right now at a pace of about 50,000 a month. That's on a global basis. And I don't know of another entity that's adding jobs on the global basis at that level. For a lot of these people, it's a flexible lifestyle. So it'll be a student putting themselves through school. I met a guy in a rock band that likes to go on the road for six weeks and then come off. And one of the most amazing things is just how self-leveling the whole thing is. I think it's going to have a dramatic impact on DUI. Studies are starting to come out showing a drop. There's a great article in the L.A. Times where they interviewed a bunch of young kids who said, "Yeah, I have a car, but I don't drive it on Friday or Saturday night."

And I want to stop there, because this was the reason I asked the question. That alone is a sufficient argument. So I have this, for now, as one of my little hobby horses. We have somehow overlooked the fact that highway traffic fatalities continue to be a catastrophic form of death in this country. We have 35,000 people die every year on the roads, of which 10,000 of those are the result of drunk driving. So that dwarfs every... More people die of drunk driving than have died in every war since Vietnam. And we treat this like it's some casual thing that we should accept as part of... If something like Uber could even cut that by a third, we're talking about 3,000 lives a year. This is...

Yet in a city like Austin where people are known to enjoy their libations, the mayor here and the city council were slower than almost every other city in the country—there are a few that are worse—to get on board with this thing. You just send one person to L.A. to understand the impact this is having on youth and drinking and not driving, basically.

Yeah. So we'll keep going on what the San Francisco— Yeah, so some really interesting things.

We've had some condominium developers in San Francisco approach us and try and get us to help lobby the city to remove the parking requirement for a new building. So apparently there's a rule you have to have so many parking spots in a new building. And the developers obviously would prefer to have more apartments than have parking spots. But it's basically changing to where you don't need the parking spots, you don't need the cars.

And the cost, by the way, of the parking requirements—the parking mandates add a significant amount, as much as 20% of the cost of housing.

Larry Page has been running around saying parking is like 17% of a city's footprint for his autonomous cars idea. And I don't know if that's true or not. I have heard another thing that like 30% of traffic is people driving cars around looking for a parking spot. Maybe those two things would be different if one were bigger, the other small. A recent study came out that was using the traffic-light analysis in San Francisco and said congestion was down year over year. And this is the city that's booming more than any other city in the U.S. So you ask yourself, how could traffic drop in the city that's got the highest employment growth rate in the country? And I think one potential example is that this new service is out there.

The most fascinating thing for me is that for basically 80 years, we grossly underestimated the demand for transportation services. And we limited it at a city government level with a taxi medallion.

Yeah. So this raises the question of where is the Uber demand coming from? To what extent is it displacing other forms of transportation? And to what extent is it suppressed demand, new demand?

Well, so another stat for you. In San Francisco, I think the numbers are something like this. I could be off by a little bit, but today we are 5x bigger in San Francisco than the entire limo and black car market was when we started. And we don't have 100% of the market. And we're still growing 300% year over year. So that suggests that the demand is 10x or more bigger than it was before. And so that's new demand. So I think at first it was just an alternative that was better than taxis because it's safer. And now it's cheaper. So Travis has worked hard to use efficiency to drive price points down. So we're 40% to 60% cheaper than taxis, so now it touches more of the economic cycle. And then it started competing with other things, rental cars.

I mean, if you travel to LA, I tell you, it's 10 times better to get a car instead of getting on that bus and riding it out to the parking lot and then waiting in line and having someone type for a little bit and then you get your keys and then you have to find you have to have your maps and you have to know where you're going and you have to know where you're gonna park and then you finally pull up to your hotel and they charge you forty dollars for parking your car and it all goes away you just don't have to do it There's an interesting thing that Morgan Stanley put out where they calculate that 97% of cars are idle. In other words, we've got an asset that's 15% of the global CapEx deployment, and it goes 3% utilized.

Which may not make sense. And the technology may be what's allowing the efficiency so that we can all share this asset and not be required to own it. Two reactions:

That surely the biggest area of suppressed demand is among the elderly. And I would love to know to what extent people of retirement age are using Uber now, because it strikes me they're the ones for whom this service has the most potential.

We hear anecdotally a lot from the child of that parent who's thrilled to not worry about that person driving and giving them an alternative. I think you will see more on this area. I think you'll see more products and services delivered that have certified drivers that are targeted for a very specific example like that. Children's another one. A lot of parents are putting kids in Ubers. A lot of parents of teenagers are putting the Uber app on the phone and saying, 'If you get somewhere where you're unsure about who's driving, just press this button.' And I think you're gonna see a lot more of that. Interestingly, we just got lucky with some of the demographic changes. The Millennials don't give a shit about cars.

When I was growing up, all you cared about was your car. I had like a cigar box where I'd cut out pictures of the car I thought I wanted to buy. I'm sure it was like an obnoxious Trans Am or something like that.

Oh, I'm sure it wasn't, Bill. I'm sure it was something tasty.

And I would put like quarters in it, you know. And today they don't care. There are kids that'll turn 16 and won't go get their driver's license.

Yeah.

You know, I hear parents frustrated their kids won't get a driver's license. So the Millennials view cars as a utility, not as a social statement, which is a huge shift for North America.

Now, which goes back to when you said that you were talking about how many jobs Uber had created. The issue, of course, is not how many gross jobs, but how many net jobs. If the growth of Uber has the effect of idling 50% of the capacity of the automobile industry, we're not ahead of the game. How much as a society have we been coasting on precisely these inefficiencies?

I don't know what percent of U.S. GDP is car manufacturing. Is that what you're getting at? I think it's enormous. I mean, we did the math. So you want to keep that going even though it's inefficient?

No, I don't. I'm just saying this is, if you're, let's not lose sight of the fact that when we, to use the terrible, clichéd word 'disruptive,' this is disruptive in a very real sense. Absolutely. The economy of the Midwest may be dependent on something that is about to be devastated.

I think that's true in every disruptive change. I think Amazon causes bookstores problems. I think that's been true every step along the way. But I would also argue, and this is a super-high-level thought, but I think if you look back across the history of time, it's when industries—and you have these large movements of economic gain—that you get the most standard-of-living improvement for the whole populace, which is what's happening in China right now.

I would only say that I feel better about disruptions when I feel they've happened naturally than when I feel they've had artificial assistance. So you mentioned Amazon. I sort of feel they got some help along the way. You have some particular issues with Amazon. I had a few issues. Did you ever see the video I made about Amazon? I did. For those of you who don't know, you should look it up on YouTube. I made a video with Dick Cavett and the comedian Dave Hill, which we pretended, you know, Amazon went through that thing last summer where they were stopping shipping all the books of my publisher. So we made a video where we pretended that Dave Hill was the new director of fulfillment for Amazon because they were transitioning from

delivering things to not delivering things. So they would need to bring in someone who was adept at not doing his job. We thought this was wildly hilarious. It may not have been, but we did.

Thank you. Speaking of jobs, so what happens when driverless cars come? Then do all the jobs that Uber created go away?

So I'm more of a skeptic on driverless cars than most people may be. And I've got a couple reasons for that. One, I think the number of nines, using the parlance of 'four nines' or error rate, that are going to be required, especially in this litigious...

Wait, wait, wait. Do a better job of explaining 'four nines.'

There's a science where people talk about what percentage of time you need a product to work versus not work. And for uptime on a high-end website, people will strive for 'four nines,' which would be like it's down for an hour a year, something absurdly low. Oh, I see. OK. And I would argue that for a machine to be out there that weighs three tons that's moving around at that kind of speed, it would need to have at least 'four nines' because the errors would be catastrophic. And this is more your world than mine, but I think humans will be much less tolerant of a machine error causing a death than a human error causing a death. So I think it needs to be way better than a human driver as opposed to just as good.

And I think we're a long way off on that. So I also think that, for whatever reason, all the large tech companies in the US have decided, in a way to distract the press, that they're going to launch these special space-y initiatives and put them off to the side. And when the journalists come, they let them in that room. And they don't let them in the other room. I think Brad Stone did a good job of talking about this in The Everything Store, the Amazon book. But the whole drone thing they did at the end of 60 Minutes, it's just a stunt. And now there's been like 18 companies do a drone PR stunt.

Yeah. Wait, but I want to backtrack for a moment. So wait a second. 35,000 Americans die on the road in car crashes, of which probably 34,000 are the result of human error. Someone comes along with a driverless car, and you're saying that it has to be perfect virtually to get going.

That's my assumption. I would tell you that I think where you may see it happen first is dedicated lanes. Those might be on a highway, they might be in a downtown area, but I think you separate—I think anyone who's studied complex systems, if you remove the random variables and make the constraints way more systematic and tight, your ability to get to, up that error rate, is high or low.

Let me just play devil's advocate for a moment. That's bringing in, that's solving the one piece of the problem that isn't a problem. That is to say, it's not highways that are in need of fixing when it comes to safety; it's almost everything else. Where you want the Google car is not on the freeway; you want it on—the driverless car, rather, shouldn't just say Google—you want it in the places where accidents happen: secondary roads, late at night, available to teenagers. I mean, I sort of understand what you're saying with this, we're not going to find this acceptable until it's nearly perfect, but when are we going to wake up to the fact that the system that we have right now is so catastrophically imperfect? 35,000 lives—

I'm sorry to come back to this. That's a lot.

I hear you. Look, I think, and I know it's going to look like I'm just talking my book, but I think Uber is a great first step, and I'll tell you why. The driver is a professional driver that's focused on the task at hand.

Look, the biggest problem you got now is shifting from drinking and driving to texting and driving. And if you get behind a car that's driving erratically or slow and you go around it, almost every time somebody's doing this. And so let's just have less people drive.

Although the last Uber car I got in New York, the driver had no fewer than four screens. I don't even know how you—there were four separate systems he was monitoring while he was driving. And I wanted to kind of stage an intervention. He was probably playing Minecraft while he was driving. I was in the back in a fetal position. Oh, the other thing I wanted to talk about was this thing about catastrophic error. So this is another one of my little hobby horses. This is the central switch that we have made in the kind of modern digital era. We have exchanged what I would call casual—we have reduced the number of casual risks in exchange for an increased number of rare catastrophic risks. The analogy of, if you switch from—

Steel fails every now and again, but when it fails, it's not that bad. Carbon fiber almost never fails. When it fails, it shatters. Plane crashes now. You automate the flight deck, you eliminate a whole bunch of casual risks, but you get Air France a couple of years ago, where the whole thing crashes when nothing went wrong because of some kind of... So it's like, I feel like we have, or even you could make the same argument with hijacking. We used to have 1,000 hijacking cases a year, all of which were of incredibly low level of severity. And we got rid of 999 of them. And now we're left with one every five years, which is really, really, really serious. We've sort of done this. You blame that on technology?

No, technology's a part of that, though. This is the typical... What the effect of the modern response to problems has been is that we've picked off the low-hanging fruit and we've pushed the problem into a corner. And we've done this without having a discussion about it. Do you want a world with a small number of catastrophic risks as opposed to a large number of low-level risks?

I personally want one.

Yeah, I would rather have more low-level risks and fewer rare catastrophic events.

Which may not be rational.

I think I'm the rational one.

I mean, call me crazy, although the evidence would suggest, just based on the size of our bank account, that you are the rational one in this conversation. No, I don't know. Would you... I don't know. I found that... I don't know if you read that piece by William Langewiesche, I think, Atlantic, Vanity Fair, on the Air France crash. It's really, really, really disturbing on some level. And you realize that the effect of attacking ordinary errors is to produce this kind of rare catastrophic event.

And at the very least, we should talk about, like... Here's a really... I think that's a lot more likely with autonomous cars than with man-driven Uber cars, just for the record. That example you're talking about, like the whole system going haywire and there being some kind of catastrophic...

Yes, no, you're right. No, I actually... I might... I just wish... The only thing that makes me like autonomous cars more than Uber is autonomous cars have the potential, at least it strikes me, as having far greater penetration. So if you view the act of driving as being... So, it is the most hopelessly, ridiculously, catastrophically flawed thing we do, we should basically be trying to get rid of it as soon as possible. And I say this as a car nut. Then the thing that can most quickly and radically change our behavior is the driverless car, probably. Anyway, I've heard... Do you know that the... Well, I was going to ask, someone was telling me that there are... We err in only talking about Google's approach.

There's a number of other approaches, including one out of Israel that many, that some who know about these things suggest is actually far better. Do you know about this? I do not. Oh, you don't. Okay. Let's talk about arms races. It is a feature of certain kinds of problems of modernity that they are arms races. There is never any clear solution in sight because for every innovation that the defense comes up with to stop the offense, the offense just does one better. Hacking being a great example of an arms race. And so far, the offense always wins. Is there any way to...

And it may tie into your catastrophic thing, because I would suggest that we're seeing the blow-ups appear to be more catastrophic, at least for the organizations that are exposed to them. So in the past 12 months, we've had Target's CEO lose his job, supposedly related to how they handled a breach or failed to prevent a breach. The Sony situation is one we've never seen before, which is of a different form, where there's a combination of penetration and espionage, where, 'We're just going to show every single email that's been in your organization to the public, and you're going to have to pay.'

Not only that, we're going to enlist the help of American journalists in publicizing the contents of those emails.

They were very helpful on that front. That's an interesting question, whether they should feel like they shouldn't write something that they know was stolen.

Well, funny, because in the Sony hack, some of the emails that were uncovered were emails sent by me. And so I got a call from a guy who said, "Oh, I found this email." He's like, "Wait a minute, how did you find it?" He's like, "Well, you know, I don't know if you know this, the hackers, the North Koreans, would send journalists..."

You sure it was a North Korean? I think it was.

Whoever it was, they would send them all the emails. And it was a searchable database. So basically, they gave you all kinds of tips. There was probably some kind of PDF file you could download, which you could go through the directions to find the most salacious emails. So anyway, the guy, following the directions of America's enemies, found an email which he felt was embarrassing for me. And he's like, "Well..." He's asking me questions about it. It's like, "Wait, if someone broke into my house, stole documents, and gave them to you, would you write a story on that?" He's like, "Well, it's not the same thing." I was like, "Well, explain to me how it's not the same thing."

I agree with you. But my partner Mitch's emails were in there with your emails on the same subject. I think they related to Snapchat.

But it's an odd thing when you... I agree. I mean, anyway.

Back to the other issue. But yeah, so hacking appears to be that kind of situation. And certainly, as we all connect our organizations more and more to the Internet, the number of openings that are exposed is going up exponentially. ERP systems in the cloud, all their financials are in the cloud. They're using email systems in the cloud, and so a lot of it's even out of their control. Yeah, because other third parties are hosting these things, and so I do think you need a different approach. We've recently funded a company called HackerOne that helps companies run bug bounty programs. So bug bounty programs are where you pay white-hat hackers rewards for finding things, hopefully proactively, before the black-hat guys get in there and exploit them.

These programs have worked really well for Microsoft and Google and Facebook. It's not very efficient for every company under the sun to do this. And so this company was started by some of the people that ran those programs at those other companies. And in nine months, we've added 150 companies, including Twitter, Yahoo, Square—long, long list.

So I'm a hacker in Bulgaria. I want to make some cash. I look at the list of companies, try and hack my way in. And if I succeed, I inform you, and you pay me a reward. We pay you off.

Yeah, absolutely. And it's probably the most elegantly scalable approach, relative to the other side, is to just keep putting gum in the holes in the dam over and over and over and over and over and over again.

How do you decide how much to pay for one of those things?

Because there are a number of companies doing it now, there are different vulnerability types, and there's like a price schedule that's kind of emerging for these types of things.

This actually might be one of the most genius ideas. So basically, what you've done is you have regulated a criminal market.

I think you've outsourced your security investigation to some very smart people on a global basis and giving them new job opportunities they wouldn't have otherwise.

No, but I have two things to say. One is, to the general point, actually I'll say the first thing first. I was having a conversation with someone from your world who was talking about this company he really loved that he was backing, where you would simply take a photo of an object, send the photo to the company, and they would arrange to pick it up and ship it wherever you wanted to ship it. So he was telling me, full of excitement, jumping up and down, best thing ever, this is going to revolutionize everything. And I said to him, 'Well, how do you know whether the object that the person took a photo of is owned by that person?'

He was like, 'You know, it never occurred to me.'

Anyway, but back to the other point. Does this get us off, does this idea circumvent the arms race problem? Do you think this actually puts defense in a position?

If you talk to—I mean, Sheryl Sandberg was being interviewed recently and she talked about this at Facebook, and they believe that it's had a material impact on their exposure. I don't think you can argue any security approach is holistic, like it's a hundred percent gonna, and—and this is not something you do and don't do the other smart things. Yeah, it's something you do on top of it to try and be proactive, yeah.

I mean, it always goes to the question of, what is the primary motivation for a hacker? Is it financial, or is it a kind of internal, within the hacker world, pride at being able to pull off a stunt?

I think some of it's the latter. And we will obviously have leaderboards for every different type of— So you are going, you want to try and give them a kind of street credibility for— Absolutely, absolutely. That, actually, there are things that will evolve on top of that because you'll find there'll be a group of hackers that will develop a reputation around a very specific type of institution, like a financial institution. And so then if a new bank wants to launch a new program, we'll probably line up a very special restricted area where those specific people get to come in.

Will you have an awards dinner every year? And only—only if you'll come. Just think about it. You can—you'd host—you hold it in Belgrade, you'd call the prize a Snowden, you get like a brass statue. These are good ideas. Like, hold it up, you need to write them down. No, it could be very—uh, this is a lot of—uh, potential, this idea. Um, but I— That is because normally, I've always thought that when you think more generally about arms races, it's so—we're finally at sports, we have 17 minutes left, we're allowed to go there. You know, the problem when you think about PEDs, that's an arms race where we're completely out of ideas about how to give defense an edge over offense. There seems to be no—every new bit of intelligence we get suggests that the problem is bigger than ever.

And there's no, I mean, there's no kind of corollary HackerOne for, or maybe. The only interesting thing I would say along these lines was that when they caught Justin Gatlin last year, the sprinter, instead of—they gave him a very, very light sentence, and what they really wanted him to do was to rat on his suppliers. So if you stop punishing the dopers in exchange for them giving up their suppliers, maybe that's a way off the arms race.

I think in the major sports you have just this huge problem where the advantage is real. I don't think anyone doubts that the advantage is real. And the payout is high. Yeah. And so, you know, the incentive program is remarkably high. And I think a lot of people, I don't know, but I think a lot of these people get dragged in around an injury, right? And recovery times are dramatically faster. Yeah. And you just slowly get drawn into this thing. But if you are better, and I think everything suggests that you're materially better, the temptation is astronomical.

Yeah, yeah.

And for whatever reason, the players' unions get in a fight with the owners, but no one's created a systematic testing program that's really intended to catch people. They seem like they're intended to say we're doing it, from what I've heard and read. Do you agree with that?

Well, certainly certain leagues like the NFL, to name one, doesn't have any real interest at all. The analogy would be if you believe that 85% of your employees are smoking pot, would you put in place a drug testing program? No. I mean, your company would disappear. So if you're the NFL, you're in a somewhat analogous position. You realistically couldn't have a team on the field on Sunday if you were at all aggressive in this area. So they're powerfully incentivized to look the other way. Sounds like the healthcare problem. Yeah, it's a version of that. But I wonder whether there's a point at which we need to... The problem is that we have a definition of sports that is too tied into absolute levels of achievement.

That is to say, as long as you're, I'm a runner, so long as runners are obsessed with the precise time with which someone runs a race, there's always going to be this massive incentive for people to cheat, right? But when you get into a sport which relies entirely on relative levels of performance, then I think it becomes easier, easier not, not easy, easier to impose some kind of, so if you look at what happened, cycling has cleaned up its act a lot now. And the average times in major cycling races, the Tour de France is biked at a much slower rate now than it was in its heyday when certain prominent local cyclists were involved. So they've had to accept the fact that the product you're seeing is in some sense inferior.

And you have to go through that. Imagine football that was played by people who were on average 75 pounds lighter, right? Imagine if I was suddenly a running back. I think people would be okay with that.

I think people would be okay with that.

I don't know how you get there. We have some questions here on Twitter. And by the way, I encourage people to shoot in Twitter questions. What is the biggest problem that you see that you wish someone was solving?

I'd probably go back to it. I don't want to be redundant, but I'd probably go back to the healthcare thing that we talked about. And I understand the point you were making about how maybe efficiency isn't what affects very specific one-on-one care. But when you look at the system, the amount of waste, like the fact that one MRI can cost $400 and another $3,200, is just ridiculous. I met an entrepreneur recently who was in the medical field that thinks he has a very, very exciting cure for cancer, and he was doing a fundraiser for the clinical trial. And I started asking questions. I went deeper and deeper and found out the clinical trial was at Stanford, where these costs are dramatically high.

And so he needed to raise money. 80% of the costs were going to go into just being a patient at Stanford, where you get the $3,200 MRI. And so it's just something's got to change. Like, it's just really broken.

Yeah, yeah. Is it as hard to get into Stanford Hospital as it is to get into Stanford University?

I do not know the answer to that question.

I'm sure there's all kinds of tests. You've got to be the 99th percentile, have an interview. Sorry. No day is complete without a shot at an elite university in my world. What's next? What industry is ripe for disruption?

Probably the one that's happening the fastest right now is in the video world, right? And we've just backed Jason Kilar, who did Hulu, his next thing called Vessel. But for a while, I thought the forces wouldn't be there to upset the kind of apple cart around the cable bundling, but it appears like it's really coming apart fast. And I happen to think that there's still quite a bit to do. Most of the really large online things, like, like YouTube, aren't very curated. It's very hard. Like, if I just wanted to go find all the Malcolm Gladwell interviews on the web, it's not easy to find that. No one's put that stuff in there by design.

I have them hidden to protect my reputation. But I do.

I think we're going to see some really cool stuff over the next 10 years in that area.

Yeah.

I would have said healthcare again, but that'd be redundant. Going back.

What company are you not invested in that you wish you were?

There's a couple that probably stand out today. Obviously, Airbnb's doing extremely well. We're not an investor in that. And more recently, Stewart Butterfield's thing, Slack, has got just enormous—if you believe what everyone's saying, just has so much enormous viral growth in the enterprise that it's something that, kind of a once-in-a-decade thing that happens in Silicon Valley.

Yeah.

Ironically, a pivot. A lot of these companies—we're in one that I'm super excited about called Nextdoor that's a social network for your neighborhood. It was also a pivot. So the team headed down one path, just stopped, and then started down another path and was very successful. How often does that happen? Slack was a game company. It wasn't even an enterprise company.

How often does that happen?

It seems like it's happening more and more than it used to. And why do you think that's happening more? I guess that the people that are running those types of companies—so both those two companies were run by repeat entrepreneurs who I think have the confidence to admit defeat and the confidence to try something new without giving up along the way. I think for a first-time entrepreneur, those kind of things are horrifying, right? If what you set out to do looks like it's going to fail, and they usually use up every last penny trying to prove that it'll work and just hit the wall.

What percentage of the people that you invest in are repeat entrepreneurs?

As much, as high as we would like. I would tell you that it actually depends. In the enterprise, it's 10 times more effective than on the consumer side. So, in the enterprise, a lot of problems are already rinse, repeat. Like, you get three customers, you start listening to them, they drive your product management, and it just goes. And so, repeat entrepreneurs in the enterprise, you know, typically just stand up and be successful again. In the consumer space, we've seen a lot of high-profile repeat entrepreneurs try and launch something and it just be a dud. It's just more of a hits business. And so the number of repeat entrepreneurs on the consumer side that is—Rich Barton, I know, has done it twice, and there's just not a lot of people that have.

Yeah. A Snapchat question. Okay. Is it a media company? How does it handle relationships with publishers who are putting content directly on Snapchat?

So Snapchat was a messaging company until recently when Evan launched Discover. And that product, as I understand from publishers, is taking off like crazy and is doing extremely well, high engagement. And so now they potentially are a media company, even though maybe they weren't 12 months ago. You know, I've had long conversations with Ryan Spoon at ESPN, and he's particularly thrilled with the ESPN on Discover. And so there are a number of—and I've even heard some of the other platform companies now wish they had built a media type that lives in their site as opposed to just having pointers away to the traditional media type because they view it as a higher ground, you know, on which to stand. But, you know, that's just how these things evolve, and I think if the platform's strong enough, then the publishers will, will get in line and do with what the product wants to do. Uh, do you think we're in a tech bubble?

Yeah, I've talked about this a lot. So I do think that we're taking on—what I've said is, I don't know that we're in a valuation bubble, but we're in a risk bubble. And we're taking on, in these startups, especially in these so-called unicorns, a level of risk that we've never taken before in the history of Silicon Valley or startups. In '99, there were certain risks taken, and there was a lot of people that were backing businesses that weren't real businesses. But you didn't have a situation where people were putting 100, 200, 300, 400, 900, a billion dollars into a private company who might only be four years old. And these companies just haven't had the time to mature. From my perspective, as someone who's been fortunate enough to help a lot of businesses succeed into profitability and success, having constraints of profitability helps you make decisions better.

Constraints drive creativity around decisions. Being able to just do everything leads to poor business execution. So now we have numbers of companies, I think, making poor decisions. Burn rates are higher than they've ever been. This is the amount of money, say, per month, that a company loses. And I think, in Silicon Valley, you have more people employed by money-losing companies probably than you ever have before, which is tenuous, because if the capital slows, then those aren't real anymore. And, you know, interestingly, this is all driven by low global interest rates. And there are literally hundreds and hundreds of investors who, if your company happens to break through to a certain level of success, knock on your door.

You're not fundraising. They knock on your door and say, 'Will you take $100 million, please?' Could you direct some of those people? Give them my email address.

I'm happy to lose money for somebody.

Warren Buffett has a great saying. He says, 'Be fearful when others are greedy, and greedy when others are fearful.' And while I'm not here to accuse people of being greedy, there is no fear in Silicon Valley right now—a complete absence of fear. And that eventually, typically, leads to things. Now, trying to predict what's going to cause all this to come down is very hard to do. I don't know. I do think you'll see some dead unicorns this year.

Wait, do you think—I mean, how gloomy are you? Are we gonna go through what we went through in—when was it, '99, 2000?

Well, so the one thing about '99, 2000 that made it so pervasive is that the venture dollars became revenue for a broader set of companies: telecom, financial. And so, like, the economy became dependent on the growth of the bubble. And that's started to happen recently in this case. Up until, I'd say, a year ago, it hadn't happened. But there are now a number of revenue streams that I think are being driven by venture dollars. Certainly, San Francisco real estate is being propped up by venture dollars. Facebook and a little bit of Twitter's revenue are now coming heavily from mobile downloads. So these are like Game of War, right? Like Kate Upton—like, those ads are now an increasing percentage of their revenue.

And they're being spent by these excessive venture dollars. And so, as you get more of these dependencies, it increases the likelihood that anything that slows will have a more—but it's early in that. It's pretty isolated right now, but it could get bigger. Is there any—what does one do in that? I mean, is there any way to arrest that process? It's, it's, it's just the interesting thing about it, because people say, 'Well, what should you do?' You can't choose not to play. Like, it's the game on the field. So, we're in a—we're in a company that went public last year called Hortonworks. Their competitor, Cloudera, raised $900 million. They sell pretty much the same product. Obviously, they both believe theirs are better.

But when your competitor raises $900 million, if you say, 'Ah, I'm just going to be profitable,' and they're going to hire 10 times as many salespeople as you do, that's a dangerous, risky game to play in both directions. And so, you're forced to play the game on the field. And I think that causes the whole system to be sloppier. Because even the good companies have to take more risk if they're gonna have competitors acting irrationally with excessive amounts of capital.

Yeah, yeah. You really have no—so, do you ever have a thought that maybe the rational thing to do right now is to close up shop?

There's a famous saying that I didn't make up in the venture world. They say the best way to protect yourself against the downside is to enjoy every last bit of the upside. And certainly, there were people in '96 that felt things were off the rails, and they missed the three biggest years in the history of venture capital return to that point—specifically to that point. And so, that's also a dangerous game to play. What I try and do with the entrepreneurs we work with is have several different financial models and to have escape hatches. So if things do change, what are we going to do very quickly to arrest the situation and put ourselves on a different vector? And the best entrepreneurs, I think, all think that way.

The younger ones don't.

We have a number of entrepreneurs today who were in, like, ninth grade in 1999. They don't have any institutional memory about it whatsoever.

Which is the interesting thing about risk, right? The further you get away from something, just most people forget about it.

But they have institutional memory of ninth grade, which is, in some ways, equally traumatic.

Maybe.

So you could say, we're down to, we have 45 seconds left, Bill. Do you have anything, any words of wisdom for us?

Yeah, there's one thing that, well, first of all, Hook 'em Horns. I'm a graduate of the University of Texas. And if anybody knows anyone in the selection committee, I think we need a lot of help this afternoon to get in. But with that aside, yeah, one thing that you and I have talked a lot about is analytics in sports. And the one thing I'd leave you with is that the thing I'm seeing in our best companies is business analytics is becoming a huger and huger part of the organization. So the best companies have a BI or data scientist that reports directly to the CEO. The amount of data is immense. And the best companies are diving through that data to learn how to operate better. And I think that's going to be a huge trend over the next 10 years.

Good. Well, we are now officially over. Thank you very much, Bill. Thank you. It's been delightful. Thank you all.

A lot of people are very comfortable saying that machines can think. They can perform cognitive tasks. They can play chess. They can plan these elaborate plane routes and so forth. We're comfortable kind of using this word 'thinking' with machines. But this emotive side, it's new. It's provocative. It's compelling. And it can also be controversial. So I thought I would spend some time.