It’s a myth that startups happen in isolation. Those legendary two people in a garage are often building on the deep and basic research — long funded by government and conducted in universities — that has come before it. But with the advent of the internet, what’s the future of peer-to-peer collaborations and startups-as-“science experiments”? Can venture capital disrupt academia… and vice versa? And finally, what’s the secret to universities like Stanford making money off the entrepreneurial ideas coming out of them? (Hint: It starts with a ‘p’. But not what you think.)

a16z’s new professor in residence Vijay Pande interviews Marc Andreessen at our 2014 Academic summit on these topics, as well as ‘regulatory arbitrage‘, how to mix humanities and science, and what Marc would have majored in if he were 18 today.

Show Notes

  • The future of research and the influence of philanthropy [0:00]
  • Merging computer science with other disciplines in education [10:07]
  • How tech varies between West Coast and East Coast, as well as abroad [16:35]
  • The future of AI and ML [19:47]


The future of research

Vijay Pande: So, I’m Vijay Pandey, I’m a professor at Stanford, but also in an interesting new role here as a sort of professor-in residence at Andreessen Horowitz. Towards that end, you know, there’s a couple interesting things that we can think about for how venture can disrupt academia, and how academia can be disrupted in an interesting way. And so, one way that we talked about is related to Bill Janeway’s hypothesis that investing, especially long term basic science investing over decades, is really what’s responsible for the success that we see in IT, and that biotech hasn’t had that quite that time yet, and that cleantech has really had nothing close to that. But now, with government probably not going to be able to put the same type of money and emphasis into things, how’re we going to have the seed corn for the future.

Marc: Yeah. So, this is — for those of you [who] haven’t read Bill, so Bill Janeway is a venture capitalist, and actually a Ph.D in economics who studied from a student of Keynes. And so, he’s kind of straight in the, what is it, the Oxford sort of lineage of economics. So, he’s both a practitioner and academic. And his book is called “Doing Capitalism,” and it’s probably the best single book on the theory of venture capital that I think came out last year. It’s one of his books that came out from academic press, and so it’s got, like, a terribly ugly cover and like no marketing. And so, nobody’s heard of it. But it’s fantastic. It’s an absolutely outstanding book. What he basically observes, he says, look, venture capital has tried to engage in all these categories. IT has been a huge success. It turns out, four decades of federal R&D money preceded venture capital success in IT. Biotech has been a moderate success, two decades of federal R&D money. And then everything else has just been a train wreck, with cleantech being the most recent train wreck. And he says, look, there was no federal research funding, right? The federal government went straight to the industry subsidies without passing through R&D. 

And so, consequently, there was nothing to draw on. There was not enough science to draw on. And so, he kind of makes the, you know, sort of interesting, profound, potentially disturbing statement of — if you want to think about the future of entrepreneurial capitalism and venture capital and startups, look where the federal R&D money is, and basically invest, you know, behind that between 20 and 40 years. Which also, by the way, goes to something I’ve observed, I completely agree. Somebody earlier said such a William Gibson quote, that the future is here, it’s just not widely distributed yet. I have just [been] continuously struck by the number of times the hot, new innovation that we see — or the hot, new innovation that becomes, you know, this huge thing, is something that was running, virtually, invariably was running in a research lab 20 years earlier. By the way, often 30 or 40 years earlier, right? I mean, in a lot of ways, like, the, you know — Facebook, you know, very successful company, right now. In a lot of ways, Facebook is <inaudible>, right from University of Illinois, right? 50 or 60 years later, right there. You know, these ideas play out over very long periods of time. And a lot of that has to do with the early research. So, the bad news is, right, to the extent that the research funding is not what it should be. And I think it’s imperative on all of us in industry, you know, to try to push for as much basic research funding as possible, because that clearly is the key.

Vijay: What else can you do?

Marc: Well, so I think it also — I forget who said it earlier, I think maybe Balaji said it earlier — which is I think the future — the hope, the optimistic view — in lieu of lots of federal research money, the optimistic view would be a more, I would say, open network, collaborative peer to peer approach to research development. You pull on the following threads. You would say, you’ve got the internet as, like, a new — basically coordinating mechanism for lots of smart people all over the world to be able to collaborate and share information and sort of build on each other. You’ve got open source. Literally open source, like open source software, but also the open source mindset — open source data, and you know, open source, you know, all these other things. Designs that could be open sourced. So you’ve got that as a thread to pull on. 

You’ve got globalization as a thread to pull on. And that there’s just more people worldwide now engaged in research and development than ever before. And collaboration between countries is going to be very powerful. And then you’ve got this really interesting kind of intermingling between research and development. Again, going back to historical, you know — arrows going back to original natural philosophy or going back to the original engineering, which was like, let’s go try to make something — and then let’s derive, you know, principles out of what we’ve tried to make. And so — and then industry — and, you know, for all of the criticisms that, we all love, you know, <inaudible> against either venture capital backed startups or big technology companies — you know, generally speaking, there’s a worldwide boom in industrial companies and technology companies getting built on the basis of R&D. 

And so, a very — I would say, imperfect — like, you know, since the 1950s style top down, you know, there shalt be the NIH and NSF. And then 30 years later, you know, Microsoft appears. In a sense, like, that’s very idealized, and that’s very predictable, and that’s very wonderful. It may be that we’re just gonna be living in a world where it’s going to be much more bottoms up, much more collaborative, much more diffuse, much less well organized, much sloppier. Maybe, by the way, more dynamic, maybe more creative, maybe more intermingling between disciplines. 

Vijay: What do you think of this crazy idea. So, let’s say, you know, a VC firm wanted to put in $50 million, in terms of fifty $1 million seed funds, right? And maybe with a relatively low valuation, because it’s very early stage. So, it’s something that you normally would stay away from in science projects, but at that sort of small stage, seed stage, you could actually see. And actually, if you had a big enough valuation, it would be worth [it] for you later on. Do you see yourself doing that? Or what’s the issue with that? Maybe there’s just all the bandwidth to facilitate that. 

Marc: So, we do some of that. And actually, I would say in the Valley, Khosla Ventures is probably the most advanced on this. They actually call them science experiments. By the way, the terms are like, you know — literally, like, they’ll invest a million dollars at a $1 million pre-money valuation.

Vijay: One on one?

Marc: One on one, yeah. So, you know, you have to like…

Vijay: That makes one on two pretty good.

Marc: One on two looks great. Yeah, exactly. We might even offer one on three. So, there’s a little bit of that. I guess the counter argument on that would be that there is a difference between science and technology, there is a difference between research and development. It’s not clear that research as research benefits from having any, you know, short term commercial. In fact, research is probably compromised by having short term financial incentives. The other lever that I think you can pull — you know, if the question, ultimately, is how to get the $50 million — if the question is how to get $50 million, you know, sort of from the venture capital <inaudible> into research — the lever that you probably pull instead I think is philanthropy. And I think this is the other part of the system that’s working better now than before. And I think there is going to be — I think this will be maybe the big upside surprise in the next 20 or 30 years — the bow wave of philanthropy coming out of the high tech community and coming out of high tech founders and CEOs and people who have been successful in these companies going straight back into universities. 

I mean, right now, I mean, it’s staggering. Like, you just walk through the Stanford campus, and you just, you know, the Yang center, and you see all the buildings, and it’s just mind boggling. Actually, this week or next week is the unveiling of Ram Shriram, who was an angel investor in Google — just funded a new biomedical research institute. So, there already is, like — you know, Tom Siebel, at University of Illinois has played this huge role. My father in law, actually, at Stanford has played this huge role. And so, there is a lot — I think you could see, you know, 10 times, 100 times the amount of philanthropy from industry and from successes in industry flowing back into universities. And I suspect that might be the real lever. By the way, that goes to something somebody mentioned this morning, which is — the really, in my view, the enlightened universities that really think strategically about things like spin offs, and students and professors taking leave and so forth, are the ones who realized that the long game here is probably philanthropy — as contrasted to, you know, university venture capital, or as contrasted to, you know, patent licensing or whatever it is.

Vijay: Well, I think Stanford certainly sees it that way. And from what I’ve seen, especially since, you know, this ability to go back and forth is something that is not looked down upon, but it’s something that is actually a real flexibility. One of the questions I have is when thinking about academia, there’s this large spectrum of people we could talk about. We could talk about the undergrads, the grad students, junior faculty, and senior faculty. And while senior faculty are probably the most interesting to talk about , you know, actually, the undergrads are very interesting to talk about. I mean, could you imagine, let’s say — put yourself as a first year undergrad, like, what would you want to be doing?

Marc: Oh, so I think about this a fair amount. We try to think about this to kind of think about what we would be investing in. So, I would definitely be computer science. For the last 10 years, I thought it would be some sort of combination of computer science and biomedicine. I don’t know enough about biology to really know exactly where I would go, but I would look for wherever the heat is at the intersection of biology.

Vijay: Why biology?

Marc: I just think there’s so much — I think, software and big data is going to be such an enormous lever to be used on affecting human health in the future. And we’re making a whole series of investments against that. But I just think that we’re at the very, very beginning of the intersection of those two fields, and the outcome is, you know, the outcomes in people’s lives, whether it’s personalized medicine, or new kinds of medical devices, or, you know, sort of different forms of human augmentation are just going to be really breathtaking. That’d be one. The other one more recently, though, is I think if I were 18 — I think I would be very tempted to go headlong into cryptocurrency and into distributed systems. I think that there’s the potential. I think if we were to create the internet today, I think we would do it completely different. I think it would have it be completely decentralized. And I think we would have cryptocurrency built in at the core, and I think it would be far more robust, and would have all kinds of interesting properties that it actually doesn’t have today.

Vijay: So, no more 404s.

Marc: Yeah. But like, you know, you build in the concept of monetization. So then you can do — you know, all the issues with resource allocation on the internet, starting with spam and going all the way through to things like quality of service. Basically, computer science meets economics would be a huge opportunity.

Vijay: So, why not just start that now?

Marc: Well, you know, the internet does kind of have the snowball rolling down the hill kind of thing going for it, for all of its issues. I mean, and the answer is, by the way, we are, to the best of our ability, we are — and there are a bunch of startups trying to recreate DNS. There are startups trying to recreate, you know, file storage. And I mean, even like, you know, there’s applications even for things like, sort of, countering censorship regimes. If you could do peer to peer routing, then all of a sudden the Great Firewall of China doesn’t matter so much. And you see more and more of these ideas popping up in startups now.

Computer science vs. other disciplines

Vijay: Yeah, that sounds fantastic. You know, it is exciting to see, with undergrads being so excited about CS and [it] becoming a dominant field. I think one of the things that’s appealing, I think, is I see what entrepreneurial power you can have with a CS degree. But I was always very curious to see, too, how that could be broadened. So, you talked about biology, and I can imagine science has been entrepreneurial. If you had to advise people for how to have entrepreneurial liberal arts — let me tell you my motivation here, which is that, you know, I think all of us have colleagues that look at computer science and see all the great things it’s doing. And it’s both exciting, and maybe a little scary, and computer science is dominating in terms of undergrads, and so on. And I don’t think any of us want to see that other part of the university disappear. That’s something we’d love to help. 

And I think the entrepreneurial aspect of computer science has been very powerful. So, to give examples, I mean, so, in social sciences — you can imagine computational social sciences becoming big, especially with big data. I’ve been trying to think. So, this is something where maybe we’re not going to solve the problem over 30 seconds, but I try to think about how we can sort of incorporate other aspects. And there’s creative aspects, in terms of art and dance and things like that I can imagine being part of it. But if we can find a way to, sort of, take that entrepreneurial spirit and apply it to other parts of the university, I think that would be disruptive in itself.

Marc: Well, I think for sure we can bring computer science and science more broadly, and math, into more liberal arts fields. But the arts alone, you know — music and movies, sort of, having a huge impact, and there’s people doing very interesting work. I will tell you, I’m an unabashed bigot — and I probably have a lot of people who won’t argue with me in this room, although this is a really good way to clear out a dinner party, if you haven’t tried it, in about 10 minutes. You know, there’s two basic, I think, aspirational modes for universities. One is, sort of, the classic life of the mind, you know, kind of Allan Bloom, you know, philosophy, science, or philosophy and liberal arts. The other, I think, you know, engineering, learn how to make things. Maybe a Midwestern farm boy — maybe we were just raised this way — but I think making things is what you do as your job, and the life of the mind is what you do for fun.

There’s this book that is coming out of this, is it a Yale professor? Dershowitz is writing this book. And basically this book, it’s gotten some heat on it. He’s like the ultra left wing alternative to Peter Thiel, will be the way to think about it. Basically, it’s a comprehensive condemnation of basically what he views as the usurping of the academy for anything other than the life of the mind. And it’s this, like, searing indictment of today’s undergraduates for, like, being too focused on, like, professional success and being too focused on gaining practical skills and being too focused on learning how to make things. And I just think he’s out of his mind. Like, I just think he’s, like, completely crazy. It feels like this is really going to intensify as an issue unless we figure out how to bring technology and science into more of these fields.

Vijay: I think that’s exactly right. And I think this is a tension that we see right now in the academy, and I think the solution is not to have one side dominate, but to sort of figure out how we can raise everything. One other issue that comes up that’s very important is the tension between technology and regulation. So, with me spending a lot of time with drug design, you know, if we had to do clinical trials for Google the way we do for drugs, they’ll never be able to make any sort of changes or improvements to their search engine. It’d just take too long. You can’t do AB testing the same way. And now we’re seeing this and other things in terms of Uber and other areas, they’re trying to push the envelope. I mean, how do you see this tension between regulation and innovation, sort of, panning out?

Marc: So, I think you — have you talked about Eroom’s law?

Vijay: Yeah. Oh, no, I didn’t bring it up. I think Balaji did.

Marc: Okay. Eroom’s law says every four years, the cost to develop a drug doubles and the timeframe elongates.

Vijay: Absolutely. Eroom is Moore’s law backwards.

Marc: Yeah, it’s the side of it you don’t want to be on. So, this goes actually straight back to the previous topic, which is, I think societally, we decide collectively, we decide how much innovation we want, we decide how much risk we want, and then based on that, we decide how much innovation we get. And I think that there is a general principle that as societies sort of, you know, advance/get older/mature, you know, we decide we want less and less risk. As a consequence, we get less and less innovation. And then, at some point, it’s sort of like the cycle of civilizations. At some point, the barbarians show up, and the barbarians, like, have less to lose, right? And so they’re like, oh, well, we’ll just go for it. 

So, it may be that the right thing to do — I mean, there’s kind of two ways, I think, to think about it. One is how to fight inside a system like in the United States, and try to figure out how to navigate through this. And you know, different companies have different tactics on this. The other is to let, you know, basically, go to new regulatory regimes. You go after basically what we call regulatory arbitrage. So, if you can’t do it in the US, find a country where you can do it. And by the way, it could be us finding a country that can do it, or it could just be an industry forming in a country where you can do it that has nothing to do with us, because it just happens to be legal and supported there. And I think stem cell research in Korea has been an interesting example. In drones, we’re seeing this now. You know, it’s still illegal to fly drones in the US, and so the initial drone deployments are all outside the US. So, we’ve been talking publicly about a couple of ideas on this. One is, you know, if you think about a lot of countries, and in fact, a lot of cities in the US a lot of countries around the world want to have their own quote, Silicon Valley, the answer probably is not to have their own Silicon Valley — it’s probably to have a different kind of Valley — you know, biomedical valley or a stem cell valley or, you know, take your pick.

Vijay: With a friendly regulatory environment.

Marc: Exactly. Specifically, and basically make a specific decision in a specific area of research to legalize a certain kind of risk, in order to get that kind of innovation in that place. The other — Balaji has been talking a lot about this — would be the idea of taking the old Economic Zone idea, that was used so successfully in places like Hong Kong, and apply that into this domain. And Larry Page has been talking about this as well, which is — maybe we don’t want, like, self-driving cars everywhere all of a sudden, but maybe, you know, a city that wants to have an economic boom around it would decide to legalize it.

Vijay: Well, and that will help people with the fear that we talked about, that you can sort of ease into things without having something be national law. And the nice thing about the way the US is set up is that there is this possibility for local versus more federal government. And as long as that was possible, I think that could be a way that things could innovate.

Marc: The problem is, a very large number of people are very invested in the current power structures, and so, the minute you start talking about this, it immediately gets painted as like you’re advocating secession. And it’s like, no. We just want to have, like, a place where we can do something new. It doesn’t need to, like, you know, violate all the rules. It doesn’t need to secede from the US. It’s just like, how about we can just, like, have a drone fly. But back to the C.P. Snow thing, like, the reaction is — it gets very visceral very fast, and that is something that requires a fair amount of energy to fight.

Cultural differences in tech

Vijay: There’s also, sort of, an East Coast, West Coast, rap war of sorts. In terms of East Coast, you know — paper belts, newspaper, Wall Street — versus West Coast, you know, computers, innovation. How do you see that changing? Or do you think that those styles are pretty ingrained?

Marc: Well, to start with, it’s like in hip-hop, it’s descriptive rather than prescriptive, which is, there are plenty of people on the East Coast who are very innovative, and there are plenty of people as opposed to who are very hidebound. I mean, just drive through San Francisco and look at the zoning, and you’ll discover plenty of conservatives who don’t even realize they’re conservative. But with the constraint that it is descriptive rather than prescriptive, there is definitely to that. And I would argue this actually goes deep in the American culture. It has for a very long time. Which is, this is the whole go west, you know, kind of phenomenon, which is, you know, where’s the frontier. It’s not a surprise that we happen to be right here, like — we all, literally, are the spiritual descendants of people who came as far west as they could, literally until they would drown if they took another step. We are the extreme case. 

And so, what I find so interesting about that, from a cultural standpoint, is — that’s a 150 year old or 200 year old — it’s the urge to go to the frontier. And I really feel like the Valley, you know, really benefits from that to this day. The number of people who pick up stakes from, you know — I grew up in the Midwest, or growing up on the East Coast, or in other countries, and coming here is a wonderful thing. At the same time, I do very much believe it’s a state of mind, not a geographic place. And I think one of the really exciting things about the last, especially the last 20 years — like, way more so today than I think when I came out here — the ideas and the attitude of innovation and new ideas seem to be spreading worldwide at an accelerating pace.

Vijay: You think so? I mean, what’s an example of that?

Marc: Oh, just the rise of — I’ll just give you an example. A good friend of mine, Chris Schroeder, wrote another great book I’d recommend, called “Startup Rising.” So, he’s a former internet CEO and State Department official who travelled around the Middle East for a year and a half and has friends in all these countries — you know, Jordan and Egypt and all throughout Syria, and all throughout the — it turns out there’s an explosion of internet entrepreneurship happening all throughout the Middle East. It’s basically unknown. And his book goes through the whole thing. He just got one of the 20 business visas a year to go to Iran. He just went to Tehran and spent three weeks, and it turns out, there’s this amazing, effectively underground internet startup ecosystem in Tehran. 

And he said the founders are exactly the same as the kind of people who are sitting at the Coupa Cafe in downtown Palo Alto. Interestingly, he said at least 40% and maybe 50% women. He said the gender balance is actually much more gender balanced. And he said they are the most fired up young people you’ve ever met, and they have an unbelievable sense of future possibility. Coupled, of course, with enormous frustration of being, you know, in a country with all these, you know, with all the trade barriers and all the political issues. But the amount of energy, I don’t know. If you’d gone to Tehran 20 years ago, 30 years ago, I don’t know if you would have found a lot of PC startups. It feels like something new is happening. At least he will tell you — and his book is about this — he will tell you this is a global movement. There’s nothing about this anymore that’s confined to Silicon Valley.

AI and machine learning

Vijay: Yeah, no, that’s fantastic. Especially [because] the cost of the computer is pretty low, and the availability of the internet means they can hook into anything. We talked a lot about machine learning earlier, and that’s something that I’ve gotten very excited about. I think many people in the room have gotten very excited about. I’m curious to see what you think that’s going. I think often, what people think about machine learning — and this came up in the previous talk — that machine learning is, you know, the dream that we’ll have Hal from “2001,” maybe a little less nasty. But, you know, for me, actually — it came up earlier that people said, “Oh, it’s just doing classification.” I mean, classification is actually fantastic. If I can say, this is the drug I want to take, or this is the stock I want to invest [in], you know, any of those things are really exciting. And so, I see, sort of, machine learning being just classification being exciting. But I’m curious to see how far you see it going. Do we have to get to Hal to be interesting? What’s in between classification and Hal?

Marc: So, I would say — so, I was at computer science — at University of Illinois ’89 to ’93, after AI had been thoroughly discredited. So, I managed to get through all four years without an AI class. In fact, I think there was still one, but it was definitely an elective, and I don’t think you were encouraged to take it.

Vijay: So you’ve taken more liberal arts classes than AI classes?

Marc: I suspect that would not be the case if I were to go through the program today, which that was definitely the case then. So, I’m the wrong person to speculate on where the core science goes. I will say, just kind of to your point, just given the techniques that we have today, and then given the consequences of Moore’s law, and all the other changes happening at the component level, the opportunities to us seem very large to take machine learning — sort of, machine learning meets economics equals Bitcoin equals, you know, financial innovation. Machine learning meets quantified self equals biomedical innovation. Machine learning meets Uber equals — and Lyft and these things — and sensors in cars, means revolutionizing the transportation system, means — which means, you know, potentially a significant answer to the emission problem. 

And so, we see a whole series of fields in which you can basically now bring machine learning to bear. And in particular, anything where you’ve got the deployment of sensors, so machine learning meets the fact now that you have, you know, 3 billion, you know, going to 6 billion smartphones on the planet, all with cameras. What can you do with that? There’s this breakthrough new iPhone app now, where it literally — you just point the camera at a printed page and it, like, reads you the page. It was just like, you know, if you don’t have full sight, you know, it’s a revolutionary, profound thing that was just not possible.

Vijay: And maybe not that far away, it doesn’t read it to you, it summarizes it for you.

Marc: Yeah, exactly. Right, exactly. And so, you take the just massive deployment of sensors, you take the massive rise in big data, you take the ability we have to build these technologies now in a way where they intersect into the real world into people’s lives — financial services, healthcare, education, real estate, transportation, government, law — and then you bring in machine learning on top of that, the sort of cumulative effect of bringing those factors together — those are new things. Those combinations are new. And we have the opportunity to build both products and companies that have never been even imagined before. Even without more significant fundamental advances in machine learning.

Vijay: Yeah, absolutely. Well, I think we’re out of time. So, let’s thank Marc one more time.

Marc: Good, great.

image description Looking for more episodes?
Find them wherever you listen to podcasts.