A Conversation With One Of The World's Leading Venture Capitalists

|
Includes: AAPL, AMZN, EBAY, ETSY, FANUY, FB, GOOG, GOOGL, IRBT, KUKAY, LYFT, TWTR, UBER, YHOO
by: David Pinsen

Summary

Union Square Ventures is, by at least one ranking, the top-performing venture capital fund in the world. Last week, we interviewed one of its partners, Albert Wenger.

In a broad-ranging discussion, Dr. Wenger compared the current ferment of digital disruption to previous industrial revolutions and sketched out where he sees it heading from here.

He also discussed his approach to investing in publicly traded stocks in his own account, and what lessons individual investors can take from venture capital.

Interview with Albert Wenger of Union Square Ventures

According to at least one ranking, Union Square Ventures is the top-performing venture capital fund in the world. Last week, we sat down with one of the firm's partners, Albert Wenger (pictured below).

Photo of Albert Wenger via Union Square Ventures website.

Among the topics we discussed were the future of work in a world of increasing automation and 'Uberization', the thesis driving Union Square Ventures' approach to investing, current USV investments in digital approaches to health care and mechanical engineering, and how Albert applies lessons from venture capital to his personal investing in publicly traded companies.

The transcript of the interview below has been lightly edited to include links and images related to topics discussed.

David Pinsen: Thanks for taking the time to speak with us, Albert.

Albert Wenger: My pleasure, David.

DP: You studied economics and computer science at Harvard, went on to a PhD in Information Technology at MIT, and then you were a serial entrepreneur, and angel investor. Talk about how you went from that to getting into venture capital.

AW: Well, it actually started when I had my own startup at the end of '96, beginning of '97 in kind of first go-round of the Internet. And I did something in health; it was a company called W3Health. And I made a ton of mistakes along the way but got the company eventually funded, had a product in market, had customers -

DP: I thought you were going to say ton of money.

AW: Sorry?

DP: I thought you were going to say ton of money.

AW: No, no, no. And one of the things I realized is that I loved the startup process but that I wasn't a really great operator nor that I thought I was particularly going to want to be good at what it takes to be a great operator.

DP: Okay.

AW: And then I really thought that the investing side was absolutely fascinating. And from that insight it took me quite some time to get to be at VC. And I took some detours; I started an incubator here in New York City with two partners, was called LC39, started that at the height of the bubble in '99. We were at $25 million in a couple of weeks and then it took me sort of two and a half years to kind of unwind the whole thing.

DP: So, sort of a Y Combinator type of thing.

AW: No, it was much more of a - it was more along the lines of like what later LA Corp. did and what Bill Gross is doing with Idealab, he's been doing it very successfully for a long time; it was the idea that we would incubate businesses. And I think what the accelerators got right later on was that they had just much more formulaic approach. So, I think the accelerators, especially Y Combinator and they've got that cut down to an art and I've since learnt that there're really only two places that you can go. You can either do an accelerator model or you can truly incubate your own ideas, the way Idealab and LA Corp. have done it. We were trying kind of a mix which was kind of a no man's land between the two.

DP: Okay. So, let's talk a bit about where we are now. Union Square Ventures, which describes itself as a thesis-driven venture capital firm, investing in networks and infrastructure for the new economy. That evolution from the initial focus, just on the large networks like Twitter (NYSE:TWTR) and Tumblr [which was acquired by Yahoo (NASDAQ:YHOO)] and that sort of thing, could you just kind of elaborate how that evolution worked?

AW: Yes. So, fundamentally, the thing that we are big believers in is the idea of network effects. So, historically, if you think of the industrial economy, the industrial economy is all about economies of scale, growing and reducing your cost per unit basically. We think what's unique about the information age is that if you grow you can actually increase the value per unit to all of your customers. And, so that part, that network effect part is still central to our investing. It's just that we've shifted more towards doing that on the B2B side. So for instance, one of our portfolio companies, a company called Sift Science, they do fraud detection. And they use machine learning to do that and they do machine learning across the data from all of their customers. So, every customer gets a benefit of this fraud detected at one customer, that benefit accrues to all the customers. And that's an example of a network effect on the B2B side.

Home page image from Sift Science Click to enlarge

Now, we've also added some other areas, so we've said, okay, everybody is building these new companies, they need certain infrastructure component. And so we invested in companies like MongoDB that provides the database, Twilio that provides connectivity to telecom. And those are core components that everybody needs. More recently, we invested in a company called Clarifai here that does machine vision. And so, that's become another part of our investment thesis.

DP: Okay.

AW: And then more recently, what we've done is we've also made a bunch of seed investment in blockchain technology, which we believe has the potential to maybe do away with some of these network effects. Most of the network effects really are the result of somebody having more data than anybody else. That's really at the heart of the network effects. And to the extent that over time you could use blockchain technology to have a consistent view into a dataset, so consistent but not controlled by one corporate entity that could help undermine some of those network effects. We think that is kind of a long-term play, and we've made a bunch of seed investments in companies in that space.

DP: So, it's kind of a hedge for you in the event that networks get broken down?

AW: I wouldn't say it's a hedge for us as much as we actually believe that - one of the core beliefs of the firm is innovation. And we think that if the world were to wind up being dominated by a few very large networks, we think that would not be very good for innovation. Over time, whenever somebody becomes very large and dominates the market, the rate of innovation in that market slows down. And so, to the extent that we believe in innovation and we want to invest in innovation, we think we should invest in things that allow for decentralized and what we call permissionless innovation where the beauty of the Web was you want to publish something, you didn't have to ask anybody for permission, you could just put it up on your website and there it was. And that gave us a huge amount of content innovation. And so, we're thinking that the same ought to be true for data.

DP: Just to drill down a little bit because I think the word blockchain, most people are thinking of Bitcoin, but it seems like that's sort of the tip of the iceberg.

AW: Yes. So, when we use the word blockchain, we're talking about all the systems that are currently being created to essentially allow for decentralized but yet consistent data stores. So, you can think of it this way. If you have an Excel Spreadsheet and I send a copy of the Excel Spreadsheet to you, well, if you changed some cells, your Spreadsheet is now different from mine. And if those are rows with let's say entries as to how many items have been sold in a marketplace, well, you can now have different entries of how many items have been sold from what I have; who now has the true list of items sold, well neither one of us does, right?

So, one way to solve this problem is to create something like Google Docs where there is one version that is constantly being synchronized. But now you put a central corporate entity in charge of that synchronization. And the beauty of blockchain technology is that it gives you both, it gives you synchronization but without an entity in charge of that synchronization. And so, the data is - so if for instance, to give an example of a company that we've invested in, we've invested in a company called OB1. They are publishing protocol…

DP: Like the Star Wars character?

AW: It's called OB1, yes, but it's O B and then the number 1. But it's a…

DP: A play on that.

AW: Play on that, absolutely. And the protocol that they are working on is the protocol called OpenBazaar; that's where the OB comes from. And OpenBazaar is a protocol for buying and selling things, but without a centralized marketplace operator. And yet, if I say I've sold you this unit, it is now registered that I've sold you this unit and I can't say, no, no, I actually didn't sell you this unit, right? That's partially what eBay (NASDAQ:EBAY) does, like if you and I are on eBay and I am selling something, eBay takes care of the thing where I say I sold this thing to you, right? And so OB1, the OpenBazaar protocol does the same thing, except that nobody - actually not OB1, no single entity controls that ledger where the 'I sold you the thing' is being recorded, that ledger is in the so called blockchain; in this case, it's in the Bitcoin blockchain.

DP: Okay, interesting. Now, in a previous article - and this ties in with some of the things you've been working on - we speculated about the rise of populism, exemplified by Donald Trump and Bernie Sanders currently in the campaigns. How that might be problematic for gig economy startups such as Uber (Private:UBER) where the employees don't have the benefits and the pay and protections of traditional employees. A question we asked a previous interview subject, angel investor Tyler Willis (Interview with an Angel Investor): is the solution going to come from the technology sector or is it going to come from government? And from your writing and speaking on it, it sounds like you envision a combination of both.

AW: Yes, absolutely. And I think historically, we've always made real progress with technology when we figured out that we needed both, the technology and the regulation. I always use the car as a good example, right? When the very first people started building cars, the very initial reaction by regulators was, oh! These are not allowed to be faster than horse drawn carriage that's because they were like how we're going to protect the horse drawn carriage industry. And then they were also like - and these things are dangerous, so, they have to have somebody walking in front of them, waving a red flag, right [see image below, via Amazon (NASDAQ:AMZN)]? And so, all the initial regulatory reaction was sort of bad, the kind of anti-innovation. But it's also a wrong narrative to say that we got the car innovation without regulation. What eventually we had is we had rules of the road, right, and we had government building road networks, and that gave us a car innovation. And so, we needed both the technology as to push the car forward, make the car safer, better, and faster, cheaper and we needed the regulation that was like okay, here's the rule of the road, here's the highway network, and it was those two things coming together.

And I think the same is going to be true for all these digital technologies, right? So, the idea that I have a phone, the driver has a phone, I'm mobile, the car's mobile, there ought to be a way to put the two together like that makes all the sense in the world. And so, if we wind up with regulation that makes that not possible, that would be a really bad outcome. But conversely, we can't just say, oh! Market is just going to take care of it because you know instead of these you have such things as congestion. That's the real problem, right? You have this question of okay, are these people really like making enough money and how do we make sure that they make enough money. There are some cities where Uber now pays so little that it's not clear that you can even maintain your car. So, we can't just say the market's going to take care of that, when there are structural problems in the market that may prevent the market from taking care of that.

February 2016 Driver Protest against Uber in NYC, via Gothamist

So, I think we want to wind up with regulation that makes it possible for lots of people to freely act in these systems. So, I'm a big fan of the idea, for instance, which we may get to on the universal basic income, if people had some base level of income then it really becomes a free choice to drive, right? It's not, I have to drive because otherwise I don't know where to pay my rent or how to feed my kid. So, we want people to be free and make this as a free choice. We want to blow up these old and artificial distinctions between employee and contract worker. We created those distinctions at a time when it was very hard to gather data. It was very costly to gather data, so we put employers who had employees for long periods of time, we put them in charge of collecting their own data and contractors who bunched around we put them in charge of collecting their own data. Now, we have all these systems collecting data all the time. So, the distinction makes no sense anymore. But we've created a system where we tied all these other things to the distinction, while if you're an employee, you get your benefits from the employer but if you're contracting it, you have to figure out how to get your own benefits. So for instance, I'm a big fan of this idea of portable benefit. You should just be able to say, here's my benefits organization and anybody I work for has to funnel some money to my benefits organization. It doesn't matter whether I work for you for an hour or for a day or possibly I'm your employee. So, we - there are ways of blowing up these artificial distinctions. But they require the government to regulate.

Sometimes it will require - this is one of the hardest things to do in democracy, sometimes, it will require the government to undo existing regulation. We're not very good at that. Often what we do is we just pile more regulation on top of it and that's not generally a good approach. So, we should take some of the existing regulations we have around employees and contractors and kind of nuke it and replace it with better regulations. But I don't believe we should just say no regulation required.

DP: Okay. A couple of things, when you mentioned the portable benefits, one thing that came to mind was - I'm not sure exactly how it works - but I think there's some elements of that, for example in Hollywood unions, because these are people that are used to not working for the same employer for a long period of time.

AW: Absolutely right. So historically, like Actors Guild and so forth, they were supposed to take care of things like income smoothing and benefits and how you get insured, because they already have that model. And Alan Krueger was a very well-known economist; just published a paper. And the paper shows that all the net job growth in the U.S. since 2005 has been in these type of jobs. So, when you look at employment growth, the net growth has come from these kind of jobs that really aren't connected to benefit, that really are highly contingent jobs, part time jobs, et cetera, because even part time jobs, you know a lot of employers are intentionally creating part time jobs that they push below a certain number of hours.

DP: So they don't have to offer the benefits.

AW: Exactly.

DP: Okay. So, it sounds like the benefit portion of it is the part that you could see private sector companies offering.

AW: Absolutely, and I think it just requires a regulatory framework where it says, hey, you are allowed to designate a benefit provider for you and then everybody you work for has to contribute to that benefit provider. I think decoupling that from a specific employer, I think would go a long way to also creating more of a market, right? Because right now, if I'm a full time employee, it's that employer - it's a take it or leave it of whatever benefit that employer is offering. I would much rather have the employer pay some amount of money, and if I want to use a different benefits provider, I can make that my choice.

DP: Right. Now, the other end of it, I guess there's two other parts broadly speaking, there's the benefit part, there's the government. I plan on linking to your TEDx talk on the universal basic income and also being represented by bot, algorithmic organizing [embedded below]. So, I guess this was kind of the three legs of the stool but why don't you just elaborate on your view on the basic income and also on the bot idea?

AW: Sure, so, just to explain what a basic income is, the idea of a universal basic income is to give everybody in society some amount of money, every month. I've been saying in the U.S. something like $800 to a $1,000 per adult and less for teens and even less for children. And the idea behind this concept, and there is sort of variance on this like the negative income tax and so forth. But the idea of this concept is basically to say look, we want people to be completely free actors to freely allocate their time. And there are many things in the world that need taking care of that are unpaid that many people would like to do. I mean that people would like to take care of the elderly, there are people who'd like to take care of animals. But they're like, nobody's going to pay me for this so I have to go do this other thing that I don't really like to do just so I can feed myself. And I think we've reached a point of automation where we sort of earn the collective luxury to decouple the just enough money to feed yourself and house yourself and cloth yourself and have a phone and have access to the Internet.

So, I think of this - some people think of this defensively like we have to do this because people are going to be unemployed. I think of it offensively like we want this, we want people to have this opportunity because historically one of the key drivers of innovation has been when labor wasn't cheap. If you think about the difference between what happened in places like India and what happened in Europe, is that in Europe we got the unions, so the unions made labor expensive and that meant capital had to be invented; people had to invent machines, and they had to invest in building those machines; they had to make those machines better, right, whereas if you have like hundreds of people you could just line up and let them do everything with their hands, you don't have any incentive to want to invent a machine.

DP: Yes, someone speculated that ancient Greece, for example, with those brilliant minds, maybe the reason they didn't have industrial revolution is because they had slaves?

AW: I think that's a totally fair thesis because they were like we can just tell these people to do whatever we want them to do. And so that leads you not to want to invent things that like we would end up inventing as part of the enlightenment or scientific revolution. So, I think that when people say, well if you pay people universal basic income, they're going to stop working. I think, well, they're not going to stop working; there many places where people want to work. But in some areas, you are going to have to pay them a lot more to work like you're going to have to pay a lot more to flip burgers and that means we're actually going to get burger flipping automation faster which net-net I think is a good thing, it's not a bad thing.

So, I mean I've always - when I talk about this, like I point out that if we hadn't radically mechanized agriculture, you and I wouldn't be sitting here right now…

DP: We'd be working on a farm.

AW: Talking about this, we'd be working on the farm.

DP: Right, right. It used to be, I think 80% of the population or something worked in agriculture.

AW: Yes.

DP: And another example I guess, today would be in Japan where because they don't have much immigration, they don't bring in low-wage workers, some of the things that would be done by those workers here are done by machines. Like I've seen they have machines to bathe the elderly for example.

AW: Yes, they've definitely made much more progress on automation. And I think we want to create a system like that. We want everybody to have the benefit that can be had from automation because ultimately we want to get to the point where if you want to take time to bathe your child or your parent or your sick friend, you can go do that because you don't feel, oh my God! If I don't show up for this job tomorrow, I am going to get fired and then I am going to not be able to pay my rent and I am going to lose my apartment and be homeless. I mean the pressure we're putting people under at the sort of bottom of the income pyramid is very real. And I think that's what you're seeing in this election. What you're seeing is that I think a lot of politicians are disconnected from that pressure that a lot of people are feeling. And so, they are suddenly surprised when somebody like Trump gets a lot of votes and they're surprised when Sanders gets a lot of votes, because they're like no, all seems fine, sort of looking around among their peers and like this - but people who are confronted with these kind of existential threats they're like no-no, this is not some like little change to the economy and everything will be fine again. This is not like let's cut interest rates a little bit and voila, everything will be there. And they're like we're going to need a much bigger change. And so, I think people are ahead of that relative to politicians and this election is bringing that out quite clearly.

DP: Yes, there was a - I don't know if Seeking Alpha would let me put this in the article but there is a little flow chart someone Tweeted at - David Pilling from FT retweeted it, it was just - and on one side basically it was crap, to use a better word, crap is broken and you go that to the left side and that has Trump and Sanders on it and then crap is fine and you have Hillary…

AW: I did see that flow chart…

DP: Yes, yes, okay…

AW: It's a simplification but as a first approximation, not exactly wrong.

DP: So, we talked about the basic income, we talked about the portable benefits; let's talk about the other part which is your idea for algorithmic organizing.

Albert Wenger: Yes. So, I think one of the things that's happened in the world is that if you look at the companies that are very valuable like Google (Alphabet) (NASDAQ:GOOG) (NASDAQ:GOOGL) and Apple (NASDAQ:AAPL) and Amazon and Facebook (NASDAQ:FB) and so forth, they operate vast numbers of computers. And we also have computers, we end-users. And for a while, we were all using desktop and laptop computers, and on those computers, we're using the web browser, mostly. And the web browsers are really interesting piece of technology in the HTTP protocol which underlies the web. It is referred to as the user agent, the piece of code that executes on behalf of the user and is controlled by the user. And so what that allows you to do, for instance, is it allows you to keep a record for your own of all the web pages you've accessed. It lets you do form filling, lets you keep your own copy of all the forms you filled out. It lets you block ads. Ad blocking is an assertion of how powerful the browser is. It can say I don't like the way you are presenting this content to me. I actually want to see this content differently. But, as most of the interaction is shifting to the phone, we're losing that agent that's representing us because a lot of what's happening on the phone happens through apps. And then the app, all the code is once again controlled in Facebook's app by Facebook, in Google's app, Google.

When we use the browser on the phone, we still have less control because these browsers are provided - there are only a couple of them. And like Google's browser, while it's extensible on the desktop, it's not extensible on the phone. So, the transition of computing from the desktop-laptop open web era to the mobile era is closing down a lot of that end-user power. And I believe we need to figure out a way to re-empower end users in this era, which will be the era of computing for many, many decades or more to come.

And so, if you think again about something like Uber, we've talked about the gig economy. If I'm driving for Uber, I'm interacting as a driver with Uber through Uber's app. Behind the scenes what's going on is that Uber has an API, an application programming interface that Uber's app talks to. But I as the end user can only use the app. If Uber were somehow required to open up their API to drivers, then drivers could have apps that were written by third-parties, a driver union or just somebody who wants to help drivers who writes an open source Uber client, you could then write a client that maybe plugs into Uber and Lyft (Private:LYFT) and Hailo or some other service, and that would reduce the power that the central networks have. And it would shift some of the power back to the end user. And I think that's a very important thing.

If we think about what's happening in the economy, Larry Summers just wrote a blog about this yesterday. We're seeing very high corporate profits and one thesis why we're seeing very high corporate profits is that we've been seeing a lot of increased concentration in market power in a lot of markets and we're seeing it especially in the digital markets where Facebook totally dominates social; YouTube massively big in video; Google massively big in search; Amazon massively big in e-commerce; Uber massively big in the sort of on-demand driving. And one way to curb that power that these very large players have is to not have regulators come in like what the European Union wants to do which is use antitrust law or behavioral regulation and sort of say, you - thou shalt not do this, thou shalt not do that. Government is very bad at trying to make these decisions as to what a company should or shouldn't be able to do. But what we should do is figure out how to get to the point where I as an end user have more control over my interactions and where that control can be done programmatically for me where I can run software just like what the web browser used to be. Some layer of software that I control, compute power that I control as opposed to that the provider controls. And that's what I call the right to be represented by a bot; I mean here not a robot in the sort of physical mechanical sense but a software bot.

DP: Right. Something you mentioned there, you talked about how power has become so concentrated, Facebook, Google and companies like that and then earlier you were talking about cars and how that evolved. Just thinking about the contrast there, because 100 years ago I mean there were probably hundreds of car companies, there were lot of small car companies. Is part of the - would you say part of the issue today is that things have gotten concentrated so fast relative to previous eras?

AW: Yes. And the reason this has gotten concentrated so fast is because of network effects, right? We talked earlier about the difference between economies of scale. So, when you compete on the economies of scale, it often takes a long time for consolidation to happen because you achieve scale economies to a certain level and then suddenly it turns out that somebody can at lower volume achieve the same low cost because of increases, improvements in automaton, right, better machinery. And so, those benefits don't often last very long. And so in those physical industries, it's taking a lot longer. And a lot of that consolidation is driven by the corporate finance stuff. But nonetheless, we've seen concentration rise even in those industries. But we've seen it go very rapidly in the information industry because of that network effect. And if the network effect is really working for you, it's very easy for the leader to be dominant, to be 10X, a 100X larger than the next player in the field.

DP: Okay. You've invested in a few startups in Germany. Is SoundCloud the largest of them would you say?

AW: It's also the investment we've made longest ago, yes. So, SoundCloud is definitely the largest of the European startups. The second largest of the European startup or maybe as large, I don't know in terms of valuation, I haven't really thought about that part, is we invested in a company out of London, called Funding Circle that's in the P2P finance for small business loans.

DP: Okay.

AW: And then we've made some investments that are more recent and still…

DP: Such as SimScale?

AW: SimScale.

DP: Let's talk about that because that sounds pretty interesting.

AW: Yes, it is interesting.

DP: Yeah, I'm just thinking because if you think of Germany, you think of mechanical engineering a lot and here's a web company that draws on that…

AW: And that's exactly what happened. It was started by five students who were all studying mechanical engineering at the Technische Universität München.

DP: Which is Munich…

AW: Yes, the TU Munich, Technical University of Munich. And what they saw was that first of all, CAD, Computer-Aided Design, has made great progress; a lot more people are using CAD. But once you have a CAD model you can actually take that model and simulate it before you put it out in the real world. So, if you have a valve, you can simulate fluid flowing through that valve, and you can see whether the valve works the way you intended to and where it has pressure points, whether it has the right valve thickness and all those things.

Now, the simulation software that exists today is very, very expensive, you are talking $50,000 or more per seat per engineer, and it's as a result largely used in aerospace and high end automotive manufacturing. But there is a lot more mechanical engineering in the world that is now creating CAD models but that doesn't yet have access to simulation. And so, SimScale is using - there is a lot of open source packages; because a lot of the simulation stuff comes out of research, it's often made available as open source. SimScale has taken these open source packages and making them available in the cloud. And so, all you have to have is a CAD model, you upload your CAD model to SimScale and then SimScale helps you to create a simulation out of that. And these are interesting mechanical things like applying force and seeing where a part would break or under how much force it would break. You can simulate that for different materials that the part will be made of. You can have air resistance. So people are simulating drone parts, for instance, on this. And it's just a way to make - to democratize the access to this kind of technology to a much larger potential group of end-users.

Sample SimScale model -- flow through a Naca duct. Click to enlarge

DP: You mentioned, in one of your posts, you're a sailor and applying this to sailboats?

AW: Yes, I mean there are already a couple of sailboat models on there, but sailing is an area where at the high end of sailing, the Americas Cup, for instance, people are using aerospace technology, and they've been using these kind of simulation technologies. But if you're a smaller boat…

DP: Like Larry Ellison, he can afford a $50,000…

AW: People like Larry Ellison, exactly. But if you're a smaller boat builder, you will not be able to do fluid simulations of the hull that you're planning to build. And all the way down because of this model, all the way down to amateurs. So, it's business like - there are boat builders all around the world, people who build individual boats, one off boats and you will now be able to take the design of your boat and run it through them and see how well it works.

DP: Alright great. What other investment of yours do you think deserves more attention than it's gotten so far?

AW: There is a company we invested in out in San Francisco called Human Dx and they're doing something very, very interesting…

DP: Dx is in diagnosis correct?

AW: Yes, exactly. And what they do is they have an app on which doctors can solve cases. And they send out a new case everyday at the moment to a growing group of doctors. And what's interesting about this is that they can use this as a teaching tool for a medical student to see whether they get the case right or wrong; they can also, however, gather data on what are the various diagnoses that doctors think this problem is. And so there's also a way to come up with a collective diagnosis. And they are already finding that in many cases, the sort of collective wisdom of a few different doctors looking at this is far better than any one doctor.

DP: The wisdom of crowds.

AW: Exactly, it's an example of that. And what's also interesting is if you think about in the existing medical system, it's very costly to get a second opinion. Yet, if you go and you have a doctor, you should always get a second opinion, anything that's even slightly serious, you should always get a second opinion because sort of quality goes, I mean immediately now imagine that second opinion coming very rapidly from a group of doctors who could just review your case very quickly. It will allow us to at scale produce second opinions and dramatically improve the quality of healthcare. And so, this is the kind of thing that has me very excited because I think when I talk about basic income, people are always like but how are we going to afford healthcare? And they look at these risks and costs we have for healthcare, and like you don't understand that all the stuff coming out that's going to make healthcare so much better and so much cheaper and it's just around the corner and we're just going to get there.

DP: Along those lines, and this ties in with I guess with Human Dx, years ago, Holman Jenkins from the Wall Street Journal had a column where he talked about the high error rate that physicians make. And he made an analogy between physicians and pilots, where pilots use autopilot for a lot of things and that probably reduces a lot of turbulence, and maybe physicians should have the same relationship to medicine that a jet pilot does to an airplane where there is a computer or a computer system…

AW: Assisting, yes…

DP: And they're kind of - they're monitoring it, but it sounds like Human Dx is sort of moving in that direction.

AW: Yes, absolutely. And part of the goal is eventually to be able to diagnose a lot of cases directly without requiring a physician or just having a physician look over it. And part of the problem is there, there are a lot of people in the world who will have access to smartphones but won't have access to a doctor anywhere nearby. And so, I think it's just one of the many waves in which digital technologies are going to contribute to better education, better healthcare at much lower cost. And I believe fundamentally that we're living in a world of technological deflation. There is a great chart [screen captured below] that I've shown in my blog where you can see cost curves for different things. And you see almost everything is trending down, except for education and healthcare which have been trending up.

And so, I am a firm believer that we will be using digital technologies to make education cheap, almost free, and then we will be able to even make healthcare cheap and almost free. And that will be a huge gain for society, for everybody in society.

DP: Well, you could argue that, education pretty much is almost free now. If you're not interested in the credential part of it, if you just want the information.

AW: Absolutely, I think you're totally spot on. I think from that perspective, we're getting there very quickly and there are amazing free resources available. And it's just going to take a while for the whole system to shift because everybody is still like, well, if I don't have this credential, I can't get this job, so that means then they work backwards. And so, this system has a lot of inertia, it's going to carry on like that for quite some time. But I think we'll - if I look at some of the early signs whether that's in our portfolio, companies like Duolingo where you can learn languages for free Codecademy or Skillshare which is not free but very cheap and has a bunch of free content. I think…

DP: You've taught some classes….

AW: I did, yes, and I don't have any of the - they switched to an online model, I taught some offline classes; I've got to get back and teach some online classes, actually a very good reminder there, I should do that.

DP: Great. Earlier you mentioned Sift Science as sort of an example of artificial intelligence that you're invested in. You also on your website, you talked about creative artificial intelligence. And the first thing that comes to mind when I think of artificial intelligence today, so, I guess two things, one is Google's AlphaGo which beat the champion at Go and then the next is Microsoft's (NASDAQ:MSFT) Tay chatbot, which you're probably aware, went on Twitter and within 24 hours it offended millions of people with all kinds of…

AW: Yes, it's interesting study in contrast. Obviously, Go is a extremely well defined and narrow domain. You literally have two types of pieces and you have a defined board. So, we should expect to make more rapid progress in such a narrowly confined domain. What's interesting, though, about the AlphaGo victory over Lee Sedol is that the way AlphaGo plays is that it needs to come up with ideas for moves, and that's very different from chess. In chess, there're far fewer pieces and there are very strict rules about how the pieces can move. And so, you don't actually need a chess program to be creative about how to come up with moves. It can generate moves, it can evaluate all the moves and then it can decide which of those to look further out. But in Go, that initial step has to already involve some kind of notion of creativity of selecting candidate moves from the very many moves that would be available even at the very first level. And so, that's why I think this is interesting because the program is being creative at some level when it's playing Go. It is coming up with new moves. It's not just saying oh, this is like this other old game that got played and now I need to make this move. No, it is coming up with new moves in this particular game on the fly. And so, I think what we will see is another narrow domain; we will see machines that can be creative.

So, if you think of a narrow domain, for example, a relatively narrow domain, music, especially music in a specific genre. I am quite convinced that we will have machines that will make very good music in a particular genre that will be entirely new and novel sounding and you won't be like oh, clearly that was composed by a machine; you'll be like, well, that's a pretty good song or that's a pretty good jazz piece. And so, the reason I'm saying all this is because there is sort of this story line out there is like oh, you know it's great that this automation is happening, we don't really need to think about the future of the job market, we don't really need to think about how to think about society because we can just do the creative things. And that's just way too simplistic an answer because that's sort of suggesting that machines can't be creative and I think that's wrong.

DP: That's interesting. So, when we have the Universal Basic Income, we may not even be doing art, we may have machines do it for us.

AW: Well, we may still be doing art but we may start to value things that are made by humans, partially for the fact that they were made by humans, right? I mean when I led our investment in Etsy (NASDAQ:ETSY), one of the things that I thought was very appealing about handmade goods was the very fact that they were handmade. So you might in some instances buy something that is potentially of inferior quality, if you see what I'm saying, but you love the idea that it was made by another human and conceived by another human.

DP: Wasn't there some controversy for at least one seller that it might not have been made by a human.

AW: No, I mean I think this is an issue with you know any large scale marketplace; it's a very large scale marketplace, you will get people try to you know sell all sorts of things and claim that they're handmade when they're not. And Etsy has moved past handmade to include small batch production as well. But my point is more, I might not want to, even if a piece of art that's generated by a computer looks really great, I might still care for a piece of art that you've made because I sort of have this human affinity between.

DP: Like paintings versus photographs, I guess?

AW: We see that slightly in photographs. Historically, there's never been a photograph that's highly valued as many pieces of art because it's sort of a sense that there was a machine somewhere in the middle, right? So, I still think that the fundamental takeaway from all of this should be that if you think about the history of humankind at a very far zoomed out level, we started out as forgers, and then we invented agriculture and that led to a massive shift in society. Then we had the enlightenment in science. We invented industry and then had a massive shift in society. And I think the thing that all of our investing and all of our concerns about innovation and my writing is telling me is that we are doing one more such big shift and it's all due to digital technology.

And so, that people would sort of go, oh, digital technology is just the continuation of having had machines and tractors and what not. That's a fundamental misunderstanding of how profound digital technology is. And it's profound because it has zero marginal costs into universal technology. And none of the technologies that preceded it had either one of those aspects; let alone both of them combined. Car, if you wanted to drive a car or make another car, there is a very clear cost in making another car. But if I have a song here in digital format or a YouTube video, there is basically zero marginal cost for you watching that video.

And universality, a car is a great thing if you want to go from point A to point B, but that car is not also going to forecast the weather tomorrow. But computers can today work on this thing and tomorrow work on this problem. They are universal machines in much of the same way you and I are universal machines, anything that we can think about, eventually they'll be able to think about as well.

DP: Okay. Now for investors in publicly traded securities, just kind of thinking about the future is definitely useful in the general sense because you want to sort of skate to where the puck is going to be. I'm just thinking you personally, you invest in publicly traded securities in your account.

AW: I do.

DP: How much of your investment process from venture capital transfers to that?

AW: It transfers in the sense that my public market investing for myself is all very long-term based. So, I don't do any trading. And so, my long-term investing is just like our long-term investing here is based on thesis. And so, for instance, I became convinced that robotics had advanced far enough that the commercial application of robots was going to grow very rapidly, and I became convinced of that a couple years ago and so I bought things like Kuka [Aktiengesellschaft] (OTCPK:KUKAY), and Fanuc (OTCPK:FANUY), and Harmonic Drive [Trades in Japan under securities code 6324] and a bunch of robotics stocks that have all done phenomenally well.

DP: But not, say, iRobot (NASDAQ:IRBT)?

AW: I didn't like the - iRobot had sort of the consumer division and the military division. I am not a big fan of that particular sort of - it's not to say that I am against having a military or anything like that, we need a military; it's just that I didn't like the dependency on large scale or government contracts that was implicit in that. I believe, I haven't been following the company closely, but I believe they've split the two recently. So, I might actually go back and look at the consumer division of iRobot.

DP: And, do you think that for individual investors there are other lessons they can apply from observing venture capital to their own investment?

AW: So, the other thing that I do is that, in my public portfolio too, is I take concentrated bets. So, I think you can do one of two things. You can either buy indexed funds at extremely low cost of administration and then just be in the market overall, or if you go into make - pick individuals stocks, have a thesis and then have a few positions, not like, once you have 5,000, 6,000 stocks…

DP: Closet indexing.

AW: Yes, you are, and probably at high cost and probably poorly relative to the actual index.

DP: Worst of both worlds. Well, listen, thanks for taking the time to speak. I appreciate it.

AW: No, it was a great conversation, enjoyed it. Thank you.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Editor's Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.