Travis Kalanick and Arianna Huffington at Uber wedding | Daily Mail Online
These shares are owned directly by the Steve and Karla Jurvetson Living Trust dated August 27, 5. Pursuant to the issuer's Outside. Five hundred years from now, says venture capitalist Steve Jurvetson, less than 10 percent of people on the planet will be doing paid work. SAN FRANCISCO — Steve Jurvetson left his venture capital firm after A second woman told Recode that she began dating Jurvetson while.
And I was like this costs a lot. I just felt guilt. So, I looked at the various rules for graduation, and it felt like a simple system of equations. Like this plus that has to equal this. So, where can I explore the corner cases? And so, I came in with the maximum number of AP credits. So, it actually goes back to my high school, St.
Marks in Dallas, that incredibly well prepared me for college. So, I took the maximum number of AP units that I could. But, of course, universities limit, not surprisingly, how much of that applies. I also took a couple of summer classes that I could transfer in from another school. They have a certain number you could do. I maxed that out one summer, while I was doing summer jobs at HPS. I took night classes. But, most of all, I learned a trick from the grad students.
This is the key answer, actually. You just make it whatever you want. You have to map this out early on. You have to start freshman year to make this work. Oh, and I had to get a physics teacher to let me take a class and a double E teacher sort of against the pre-requisites saying like I really can do this. Let me try to take your class a year earlier than I should, so that I can get the pre-req chain going.
All of these simultaneous equations you have to solve. But the signing up for more classes was the key. There was like one winter quarter where I took the equivalent of eight classes where you would normally take a load of four that made that possible. So, when did, if it did, when did the budgeting stop being a concern? Because you then went on to continue your schooling.
Oh, this is at HP? Well, it was working summer at HP, but I started a master in electrical engineering. I was a research assistant for Professor Hennessey who went on to do amazing things there and was an entrepreneur himself, at one point. And I shifted gears. And so, I, basically, at that point, stretched it out and tried to stay as long as I could.
Got a balloon payment of recreation. Now, I want to stay. And I found a way to be in the freshman dorms as a grad student, which was awesome.
I was the computer coordinator there, which meant that I kept the computers running and kept the paper in the printers. But it meant I could have freshman dorm life all over again as a grad student. And it was fun. Oh, that came later. And would you do it again, why or why not? A lot of folks who I knew and know are asking that same question. Even law school and med school and what have you, you just go straight in from undergrad.
So, the main reason I went is networking really to expand the — and to learn, frankly. But the networking came first in mind. To meet a bunch of people who come from all kinds of different backgrounds that I could learn from, not only at school but throughout my life. And one of the sub points on that is, for example, pretty basic questions like what do I want to do in my life. I had bounced around from engineering to management consulting, prior to business school and thought well, from what I know today, maybe I want to go to product marketing.
And so, you have all of the consultants trying to leave consulting. And you have all of the I bankers trying to leave I banking and learn about something new. Especially when your employer hoping to leave is paying for it. In fact, that was exactly what happened to me. I was financed by my former employer. So, I had that sort of safety net. And then, I decided not to go back, I had to pay for all of my tuition in one check.
On the Edge of Automation - MIT Technology Review
Yeah, I was wondering how that worked out. That was a tough decision. But it did raise a barrier to leaving. The other thing is learning. Business was not what I went to learn per se.
I knew there would be some classes about business that I thought would be helpful. And I had taken classes back when I was an undergrad and a grad student prior in accounting and finance. So, I had already taken a lot of, again, the core courses. I placed out of every core course at business school, too, and went right into the electives and the more interesting classes.
But I also knew that it was an opportunity to take classes in other departments, to brush up on actually molecular simulation of all things. And one of the classes I took was on modeling quantum mechanical properties of materials.
Steve Jurvetson's personal Apollo collection (photos)
And that really all played out. So, did it to what I expected? In a way, it did. When I grew up in Texas, I had heard the term.
I never met a venture capitalist. So, I literally would not know how to reach one. So, the entire transition of the venture occurred because of people I met there. And I learned a lot. I should point out good location for it, too.
If you think about do you want to go to business school, I did not apply to any other school because there was no other school that would have met that goal. Also, I learned a lot. And back to the network, almost every investment I made, I think it was almost for the first year or two out of business school, and there was a lot because it was the middle of the dot com boom for me, had some business school connection. It was, literally, the entrepreneurs were from business school, or it was sent to me by a professor from that business school.
What mistakes do you see otherwise very smart venture capitalists making? Are there any patterns? Are there any particular types of mistakes? It could be investors, broadly speaking. Oh, my God, tons. All over the place. So, the first one, this notion of being fear driven.
Every company will succeed. The earlier you go, the higher your death rate is going to be as a startup. So, the majority will fail. Just give them time. And that sad reality took a while to sink in for some. So, most investors I see just let all of the biggest home runs pass them by. Nobody else would invest in the company. What were the reasons for not investing?
So, and, in often cases, conventional wisdom would support their point of view. Many in batteries, many in solar, just scores. Maybe hundreds of companies have failed. Pick whatever it is du jour. Consumer internet or social this or block chain that.
Yeah, but there are also other people who have the exact same idea you just did. Why are you different? So, this lack of strategic thinking and also lack of thinking of a long term strategy.
What worked for you, in the early days, when you were just getting started? First, it was a stroke of great luck, which is the dot com boom, as we now call it, was just starting to happen.
And I was ready for that and open minded enough. That would be the consumer software segment. That was just a horrible place, games, history of business. It still is a horrible segment. It is to do non internet consumer price point stuff. So, they thought the internet was that. And if you put it into a prior bucket, it would be those are bad sectors. So, unbelievably, one of the biggest investment opportunities probably the venture industry has ever seen was largely avoided by most venture capital firms.
And when I started inwe did about a third of all internet investing industry wide. The entire venture industry. And when Netscape had its IPO, boom, everything changed because, now, there was an example of a winner.
This pattern will come again in space investing and everything. As soon as you have a single winner, then, the sheep phenomenon and the herd mentality kicks in. To think how much bigger might this opportunity be than even the people pitching it are imagining. Normally, an entrepreneur is trying to sell, to the maximum extent possible, to predict the largest possible addressable market to think well, Phase 1 of my business, I do this.
And Phase 2, I take over the world in some way. So, I credit Tim Draper. I credit the internet. It just seemed like the natural thing to do around early stage investing. So, when I thought about this, I was like, okay, most of the things fail. They should change the world. They should be disruptive. Those are words people use. Well, how would I know if something really is big and disruptive?
Well, every entrepreneur claims it. It could be doing a payment processing thing for back in whatever payroll system. And there are all kinds of [inaudible] that come out of that. There were no other quantum computer startups, zero, on the planet earth. There was not a single company claiming to build quantum computers, big or small. Actually, for five years afterwards as well.
So, that one is the most extreme. Maybe for good reasons, no one was doing it back then. The way that it did peer to peer voice over IP.
And, similarly, with Tesla and electric vehicles. And I can get into why that one maybe is the most corner case of wait, there are a lot of electric car startups. Why is that one special? We can talk about it. But, certainly, Space X would qualify as well. And but the benefit from my career was it has me perpetually, searching for the next great technology wave. But, on the other hand, I was not finding anything unique.
I saw four business plans for selling pantyhose on the web. Like three others like really? Four business plans to sell pantyhose or just balls.
What does just balls do? Just Balls, ventured backed, just sells balls. The name was awesome. What do you sell? You want a bat? You want a glove? Balls of all kinds, small balls, big balls, golf balls, tennis balls. Were you a founder? Okay, just making sure. But this shows you how absurd it got to say take the economy and the internet changes everything. Now, to my — so, I avoided a lot of pain. If you add up every internet — remember, we did so many of them?
If you add up every internet company we invested in that lost money, add them all up, it was less than single deals lost from other firms. And every one of them was unique. So, you get diversification for free.
That might have a good 10 year run. Why did you choose nano technology? That was a bit of a mistake. But it was a mistake in retrospect. I did some chip design at Hewlett Packard. And the angel investors are flooding in. And every banker or consultant is flooding in. I had been exploring at the Forsythe Institute and a number of other sort of think tank like conferences that the radical potential this had to revolutionize a bunch of things.
It more was a set of science and technology capabilities that would stretch across many industries. These questions are process questions. Like how do you do innovation? How do you do entrepreneurship? But that platform technology, which let me just tell you what I mean by this. Nano tech was like a platform applied in [inaudible] new ways of doing material science, if you will. And then, more recently, deep learning, neural networks, what we might describe as narrow AI.
That whole field is widely applicable. I think every company will use it eventually. So, again, we can come to that later. All they do is stuff related to machine learning.
But much more interesting is all of the industries that will be changed by the use of machine learning, the vertical industries. Thermo dynamics as a concept, there was a time when, actually, companies named themselves Thermo Lectron and Thermetics.
Are there seed changes in, basically, how we do science and engineering? And I think, today, deep learning is the best example of that. Say that one more time, the last part. So, I think deep learning or let me just call it machine learning, broadly, because there are a bunch of sub branches of this. So, [inaudible] is just imagine you started with a random dot or two.
Automata like automaton but with — Steve Jurvetson: And it sort of unveils. One directly above me, left, and to the right. And so, you would think that such a simple rule that just says look at the three dots, and there are literally only sixteen possible conditions that I could be seeing up there. But in a few case, like Rule 31, it unpacks into this incredible complexity that is not repeating.
You have to run the experiment. And the reason I find this so philosophically interesting is this is the same thing in evolution.
You actually have to just run this little algorithm over and over and over again. It can learn to do what I train it to do, but with some edge cases. So, I can come back to it if you want to talk about AI and deep learning. But I just — you asked about [inaudible], and the process learning — the question you started with was engineering, right?
So, specifically, engineering since the scientific method itself. So, way back when, engineering may have been random like I think astrology is true or I think this is true.
And then, you have the scientific method. You release the method of compounding knowledge over time. This is a different way of, in a sense, growing solutions to problems instead of purposely designing them. You can build things that exceed human understanding.
We have sort of reached the limit, in many cases, of what a human, any human or even group of humans, can put their heads together and design. And now, we can push way past that and build things beyond our own understanding, which is pretty powerful. Could you define — just revisit, you might have already done it indirectly, but the deep learning and neural networks? I guess I have sort of an approximate equal sign here.
On the Edge of Automation
There is a reason the word deep came in. Well, first, you want a new name. But there is a reason. Deep in the sense of more layers. So, what do they have in common? And, in fact, neural networks, I think, could be a more generic term for all of this.
So, now the whole brain. Well, you have cells that connect to a bunch of others. And our brain is hugely a fan out of a thousand or so to ten thousand connections from each one.
So, a huge fan out. Each cell connects to a bunch of others. And when one fires, it sort of goes down the dendrite, and the little synapse fires, and it triggers another one. Well, how can you make a computer like brain? And the essence is something activates. So, the simple idea is, to recap, neurons fire. You sum up all of your inputs.
And if it crosses some threshold, then, you fire. And so, the deep learning is having more and more of these layers. Instead of one or two or three layers, having scores of these layers. And a variety of other algorithmic advances. We take it for granted. Of course, you want a million of those, do you want a billion? How many pictures of cats do you want?
You could spend your whole career trying to get a good data set to train digital networks with. So, data is the first. The other is GPUs from Nvidia. They started about 10 years ago as a side project in scientific computing realizing — Tim Ferriss: So, any gamer would be like I want the latest to play my video games to do all of these polygon renderings offloaded from your main computing chip, the Intel chip, usually, off on the side what used to be a peripheral processor or co-processor, an afterthought, if you will, that only gamers cared about.
The only thing you cared about was your CPU, the thing that computes your Excel spreadsheet or whatever, your database. But the GPU people were pushing a whole different kind of computational math forward, which was a lot of math. A lot of polygons. The more polygons, the more sort of resolution you have in your game.
- Steve Jurvetson
- The Tim Ferriss Show Transcripts: Steve Jurvetson
- Travis Kalanick and Arianna Huffington at Uber wedding
Multiply a whole bunch of numbers together and add them up. Multiply, add them up. Can we try to see if we could use it? And so, it was like the skunk works project. Management had no resources against this. I knew, actually, one of the first four customers working in this RND area of all things to model how neurons fire scientifically.
To model how a neuron actually — the actual biology of a neuron down to as much detail as possible. A company called Evolved Machines. And they were using GPUs. So, this incredible computer resource was pushed to the limits from gaming.
We get an avalanche of information on our retina and our cortex. And we down scale that to image recognition building over there your face, the symmetry or your eyes. If a startup or new business venture has created a job that involves human labor, it probably has done so in a way that is pretty marginal. There will be massive dislocation.
Which jobs will survive? In the long run, years from now, everyone is going to be involved in some kind of information or entertainment. Nobody on the planet in years will do a physically repetitive thing for a living.
There will be no farmers, there will be no people working in manufacturing. To me it is an impossibility that people would do that. People might do it for fun. You might have an organic garden in your backyard because you love it. It pretty much will be what life was like for most of human history—just without the gruesome servitude. If you go back a few hundred years, everyone was either a slave or a serf, or living off slave or serf labor to pursue science or philosophy or art. Is there some way, some government policies or strategies, to minimize the pain of such a dramatic shift?
No politician has a year horizon. I see zero chance that long-term thinking will govern policy. I do lament how many investors focus on all the short-term sugar buzz of some marginal improvement in something—nothing history books are ever going to be written about.