11/7/2012 - Vinod Khosla. Founder of Khosla Ventures and Sun Microsystems

Copyright Stanford University. All rights reserved. All right. Thank you very much. I would like your attention, please. Thank you for doing this little exercise.

Yeah, don't shoot, please.

Thank you for doing this little exercise. I hope that it was helpful. I would like to introduce our speaker for tonight. I met him at a Churchill Club event where he received the award for Legendary Leader. He's one of the most prominent VC in Silicon Valley. He's the founder of Khosla Ventures, one of the venture capital firms on Sand Hill Road in Menlo Park. Prior to Khosla Ventures, he co-founded Sun Microsystems, a billion— multi-billion dollar company that was acquired by Oracle. And he also worked at the venture capital firm Kleiner Perkins. He's a member of Forbes 400, and he's also an alum from the GSB. So I would like to ask you to join me in welcoming Mr. Vinod Khosla.

So thank you, everybody.

I'd like to explain a little bit the format of this seminar. When you register, we ask people to provide questions that you would like the speaker to address. So rather than a formal presentation, this seminar is going to be an open-ended Q&A so that you have the opportunity to ask the questions that you are dying to ask to a VC like Mr. Khosla. But to start, I would like to— before I open the Q&A to the audience, I would like to ask you a few questions. And first, I would like to check How many of you in this audience are working in the area of cleantech?

Okay.

That counts. High-tech? Biotech?

Medtech? All right.

So a good proportion of the people. Now, how many of you are thinking about doing a startup or already created startups before? Wow, that's impressive. A lot of entrepreneurs in this room. Um, on behalf of the people who are working in the cleantech, high-tech, health-tech sectors, I would like to ask you, um, how do you think these sectors are going and what are the future trends that you see coming from an investor point of view?

Well, it's, uh It's great to be here talking to everybody. I mostly like talking to people who are, one, deeply technical and who want to start companies. And I saw a lot of entrepreneurial hands coming up, so that's really exciting to see. Let me answer your question a little bit indirectly and give you a broad introduction of my views. There are a lot of challenges society faces. I'm one of those people who believes that most important component of solution to those challenges is technology. You asked about areas. What surprised me is every single area— and I'll give you some fun examples— I've looked at I have a tendency to talk too much, give long answers, so shut me up anytime you want. But every single area I've looked at, no matter how unlikely it seems amenable to technology solutions, has a technology solution. And that's why I love people who are applying science or technology in one way or another to these large problems. What I find gets in the way of young people— like most of you are much younger than me, I consider myself relatively young— I find people like you get swayed by experts who tell you everything that can't be done. So I find the single largest problem in society is all the experts who tell you the way things are going to be and why you can't do what you think you want to do. I'll give you an example. Professor Tetlock at UC Berkeley did a statistical, rigorous statistical study of this topic. I've studied this topic of expert opinion. Mostly in my gut, I always wanted to ignore them. So I wanted to understand what I was saying. But Professor Tetlock did a more rigorous study over 20 years and across 30,000 real forecasts and found that the average statistical accuracy of the best experts in their area of expertise was about the same as dart-throwing monkeys. If you don't believe me, read a book called Expert Political Judgment. And it wasn't just about political judgment. If you're a stat nerd, you'll love the book. If you're not a stat nerd, you should read a consumer version of it that Dan Gardner wrote called Future Babble. Almost all experts are babble. I was explaining day before yesterday as I sat with my daughter watching election coverage before the election, how it seems like the chief election analysts and experts that both CBS and ABC had on didn't understand statistics. They were of course all completely wrong. Even midday yesterday, they were talking about how close this election is. And Monday morning, based on rigorous analysis that Nate Silver, who writes the FiveThirtyEight blog, had done, I said to my partners, I'll take bets with anybody that Obama will get more than 300 electoral votes, no questions. Of course, that's what happened. And all the experts and pundits were wrong. Why do I tell you this story? I tell you this story because no matter what your area is— now I'll come to the key areas. I said I wouldn't do a talk, but I'll give a talk anyway. These experts are dead wrong. They know how to extrapolate the past, but they don't know what the future is about. I ask people like you to not worry about what's happened in the past, but think about what you can make happen. I like to say, instead of extrapolating the past to predict the future, you want to invent the future you really want. And you have that choice in much more powerful way than most people believe. So I want to get you off the belief that you should look at market studies or other things. I'll give you, since we're in the School of Medicine, I'll give you my current latest belief system. I looked, I started looking at it because I was looking at healthcare in India. Turns out you need 10 times as many doctors as India has today to provide reasonable healthcare to a billion-plus people. There's no way to afford that. But as I start digging in, what became clear to me is in most parts of the world, you can replace 80% of what doctors do. Computer systems are much better at diagnosis than the median doctor, maybe even the 75th percentile doctor. Now it seems improbable, but all the early studies that have done research on this topic say you don't need doctors most of the time. Not for diagnosis, not for prescription. And then some of you will say, well, but you need the human element. Well, if you do, you don't need an MD to have the human element. In fact, there was a study done at an Oregon, I think it was an Oregon urgent healthcare, urgent care clinic. You could replace doctors in 73% of the, of the, incoming patient visits in very conservative sort of notion of safe triage. In another case, they did a study, and without medic labs, high school graduates who want 1 year of medical assistance training replaced were accurate at 93% level with the help of such a system. That's scalable medicine. So that's just one example of belief systems, and experts will tell you why that's wrong, but they can never stand up to scrutiny. So my— I urge you to ignore the experts and just invent the future you want. It doesn't matter what area, and I'll get to the various areas. I've mentioned healthcare. The same is true of almost any other area. Data science and big data is opening up so many possibilities that you probably don't need teachers either. Computer systems can do education much better than teachers can. There was a little prototype study done at Notre Dame University where while the student was working on his computer, the computer was watching the student. And every time they changed their facial expression, they knew what the student's emotion was. If the student was bored, they changed the topic. If the student was puzzled, they made the problem simpler. There's no teacher who can do that. So my point is, everywhere you look, you can do this. Professor Pat Brown, who is a professor at Stanford, probably one of the 5 or 10 best-known names in genetics in the country, just left Stanford, gave up his tenure, gave it up, gave up a Howard Hughes Fellowship, which very few people do, just to reinvent ground beef. You heard me right, reinvent ground beef. He decided that since animal husbandry—cattle, chicken, things like that—take up 30% of the land area of this planet, in taking plant protein, like corn protein or soy protein, converting it to meat on your table, and each pound of steak you eat takes about 15 pounds of corn, in about 1,800 gallons of water, that that was so inefficient he was going to dedicate his career to making that process much more efficient. I won't go into the details, but my point is, as unlikely an area as ground beef seems, it's up for reinvention. As unlikely an area as replacing doctors or teachers seems, it's up for reinvention. So almost every area I can think of, and hopefully you'll have lots of questions, is up for reinvention as long as you have the skills, often technical, but not always technical. Sometimes there's other skills that can be easily as effective to reinvent the world we live in. So let me— I've always wondered why just beef and not pork and chicken, chicken fake meat? It's just beef is a big market in this country. There's $100 billion of just hamburgers, which is a relatively easy target. But absolutely, you can do— you can do chicken. You can do pork. You can do fish. Look, all they are is perceptions in your tongue and nose. There's aromas associated with it. There's feel associated with it and taste associated with it. If you replicate those experiences, you have that product.

So my second question is a very practical question. A lot of people in this audience are working on their startup projects. If there are 3 key points that they should pay attention to, what would that be? And in particular, if at the end of the session someone comes to you and has 2 minutes to pitch their idea to you, What should they highlight?

So let me mention 3 things. First is the quality of the people you work with. You should never compromise that. That's sort of always number 1. The second thing I'll say is a little unusual. I tell people, don't be afraid to fail. People want to make plans that are going to be successful. Everybody wants to be successful. But the way to be successful— there's two ways to be successful. I like to say, I don't mind failing as long as if I succeed, The consequences of success are consequential. So think about it. You can always reduce the problem to a high probability of success, and the consequences of success being inconsequential. Mostly I'm saying those people do incremental things, and then you have a high probability of success, but who cares after you're successful? On the flip side of this is take on projects that might be bold in their vision, may be difficult, may be hard, have a lower probability of success. But if they're successful, they'll be disruptive and cause a major change. Those are, to me at least, the more interesting projects. So the bottom line, don't be afraid of failure. Make sure that when you fail, you fail in small ways. And for you, any small failure doesn't ruin your big idea. And fail quickly, iterate quickly, move up the path of learning. That's how you do successful startups. You don't get there by trying to avoid failure. You just manage failure. I always love giving examples of perfect in the human body. In a computer system, you cut a wire. Most likely the system grinds to a halt. You can cut off my finger, I won't grind to a halt. My body keeps going. It's a resilient system. That's how you should think about failure and the way you think about what you're trying to do. Try lots of things, fail quickly, admit, failure and move on and learn from it. And then do the next thing. So the last thing I would say is the most common wrong failure mode— there's right failure modes where you're experimenting, trying things, and learning. The wrong failure mode is to not be open to your ideas. Any psychologists in here? People with psychology? OK, nobody. It's easy to believe your own bullshit. It's easy to take all the incoming data and reject the stuff that doesn't confirm to your thesis and accept the stuff that confirms to your thesis. Psychologists would call it confirmation bias. In fact, since many of you are in the medical school, I presume MDs, there's a very high degree— this misdiagnosis happens all the time, maybe 10% of the time. But the much larger error is not misdiagnosis, but missed diagnosis. And that happens for the following reason. Patient walks in, says, something for 30 seconds, you've made up your mind. This guy has XYZ. All your follow-on questions are designed to confirm that bias, that early conclusion. Dan Kahneman's written a great book called Thinking Fast and Slow, highly recommended reading. We jump to conclusions and then look for confirmation. Startup entrepreneurs do the same thing. I like to say, don't be full of your own bullshit. Look for as much criticism and critique as you can get. And if you can get all that and consider it seriously, you'll be in a much better place as an entrepreneur. So those 3 things I'll bring up.

How about the 2-minute pitch to you if someone runs into you and want to— share their idea of startup? So what are the points?

The 2-minute pitch has 2 simple ideas. Why is this an attractive market? And 2, why is the technology breakthrough relevant to that market? Why would that be disruptive? But there's a meta question I would add. Almost the first thing I judge when I look at a business plan, a PowerPoint, somebody talking to me is, the clarity of their thinking. You'd be amazed how many people are not clear and succinct about what they're trying to do. And then they're very amenable to drinking their own Kool-Aid or their own bullshit because they haven't dealt with the reality of what they're trying to do at a ground level.

So my third question is about social ventures. There are people in this room who are interested in creating social ventures. Could you tell us why this is also important to you and what can be done to promote the creation of this kind of businesses that are not driven only by profit but also by the, to make a positive impact on societies?

It's a great question. Let me answer that by first See, I'm pointing out another problem I see in entrepreneurs. Very often, they're confused about why they're starting a company. There's lots of reasons to start a company. All of them can be good reasons. One reason is, I'm just passionate about this technology. I've worked on it for 5 years. I want to see it out there in the world. Real good reason. I want to make lots of money. Real good reason. I just want to have my own business in my playpen. And real good reason. But then you do things differently than if you're trying to build the largest business. I just want to work with friends and I have a nice work environment. Real good reason. But be clear. If you care about impact, to your question, then social businesses make a lot of sense. And it's a pretty common reason, especially why technical people want to start companies. If you want to change some areas, I was mentioning Pat Brown who's doing beef. But he really is passionate about eliminating all animal hus— husbundry on the planet. That's a pretty big goal. That's not to say he won't make money, but he's clear why he's doing it. I like doing it because I sort of get bored of just things that are businesses and have no consequence socially. We have a separate impact fund where we take the approach that we only invest in companies that produce products for the bottom 3 billion people on the planet. So it's a separate little fund, different than Khosla Ventures. It's called the Khosla Impact Fund. And we do that too. So if you have a great diagnostic for the developing world, or great drip irrigation system for the developing world, there's a Stanford design student we funded who developed this great laser technology to do drip irrigation. They don't have pressure in your pipes or have a pump or electricity, and it's precision and it doesn't go bad and all those kinds of things. Those make sense. But I can tell you, in the end, if you look at life, it's very rewarding to make a difference. Way more rewarding to make a difference than make more money. And, and I find those people who pursued those as the happiest people with their life.

So yeah, I can believe that. Now I would like to open the Q&A to the audience. So if you have questions, raise your hand and the microphone is going to come to you.

So we have a question here. Yeah, you know, this small enough group, we probably don't need the microphone. I'll just repeat the question.

Let's try it. Okay, go ahead. So I have two questions.

The first one The question is, how do I blueprint my decision-making? There's no one formula. Every time it's different. But a few things sustain. If I don't think the person has high-quality thinking, the quality of their thinking That's a clear decision maker. Two, if I think they're obstinate, not open to input, that's a pretty clear thing. And then there's the usual. How large is the market? What's the competitive situation? How easy is it to adjust? How breakthrough is the technology? Sometimes just, do I fall in love with the idea? We just invested in a company that's taking on Mars candy bars. I love the idea of taking on Mars with candy bars. They have an M&M's and a Reese's Peanut Butter Cups and a Milky Way and Snickers bar. And it's just all organic, no junk, no artificial dyes, and 50% less sugar. Just sounds like a fun idea to try. So it can be highly variable. What I'd say in general is we seldom look for everything about a startup to be right. What we look for is a team that's open to making the pieces that— that has some pieces that are really strong and some pieces that they're open to making right. And that's sort of an aggregate judgment. Other questions? Yes.

Hi, Vinod Khosla. My name is Irfan Ali Khan. I'm currently starting a skin cancer diagnostic startup for the Canadian market. Thanks so much for coming and speaking. My question to you is, in your experience in the whole startup world, there is a difference between high tech and biotech. What's been your experience between the differences and how have you seen yourself and other entrepreneurs approach and deal with those differences?

So we don't do biotech startups. Why? Only one, because we don't have expertise in especially the pharmaceutical side of biotech, but skin cancer, is a great example. To me, that's more information technology than biotech. There's a startup in the UK, you probably know about it, that looks at your moles and turns out that the fractal pattern of the blood vessels in the mole tells you what kind of cancer it is, or if it's a cancer or not a cancer. I don't know what you're doing, but I suspect it's something like that. That's a pattern recognition information technology problem. And there is a startup in the UK doing that, and I love the idea. We have a little startup called CellScope that's doing controlled imaging of like moles or ear infections. And to me, those are IT startups in medical applications. So I forgot your question. What was the specific question?

So what do you see the differences between biotech— Oh, OK.

—and the reasons why you don't like biotech? Pharmaceuticals, which is traditional biotech, Genentech-type stuff, is so subject to FDA regulation. Most cases the cost of a trial is so high that you lose all the entrepreneurial juices. Going through a big drug company who helps you run a global trial, it's just boring for me to deal with. I don't deal— like dealing with large companies. I just— they never do anything interesting. They always find ways why something shouldn't be done. And so I just find it boring whenever you have to interface with big companies. I'd rather just deal with technology, technologists, entrepreneurs, entrepreneurial energy.

How about small life sciences startups?

So let me see if I have my iPhone with me. Yeah, I do. So this is an example. Of something I love. So this is— this is an iPhone. This is an iPhone case. I'll tell you, it costs probably $10 to produce. If you open it, you can say that. But just by firing up the right app here— let me see if I can fire it up. And see if I can hold it here. So suddenly a medical grade EKG shows up. And my heart rate's 73, which is not too bad. To me, I don't know whether this is a medical startup. Yeah, it's an EKG, is a medical device. But if I have this with me, if I happen to be a heart patient— touch wood, I'm not. No wood around here. I'd want it with me all the time. I'd take my EKG when I got up in the morning. After I exercise, I'd take it again. After I ate, I'd take it. If I wasn't feeling well, I'd take it. If I sent— like you might say, take your blood sugar if you're diabetic. Because it doesn't cost anything. But if I sent 5 EKGs to my doctor every day, he'd stop dealing with me. He wouldn't know what to do. So now it's an IT problem. Now I need to do pattern recognition. So I told this company, I don't want them to think about an EKG machine. Just get me a million digital EKGs along with their associated case histories. Anybody in data science here? How many people in data science? Probably the single most important area the next 10 years. Single most important area. That combined with machine learning. But once we do that, once I have a million digital EKGs, I can do better than 99% of cardiologists. And that kind of medicine I'm happy to do.

So in the area of licenses, maybe mobile health is the sector that you would—

Well, mobile health, because it's driven by the consumer. And I don't care what the system thinks, what the healthcare system thinks. I'd like a reimbursement code, but I don't really care whether they have one or not for this. And so my view is very simple. If you aren't tied up in institutional bottlenecks, be they FDA or large hospital systems, then we're doing pretty damn good. And give you an example, since there's a lot of MDs here, one simple example of what data science can do. And I was with the CEO of Kaiser Permanente and their innovation team last week. And they took a very quantifiable thing, stroke patients coming into Kaiser Permanente. They have a large system, like 9 or 10 million patients. They get lots of stroke victims. By timing when they give them a statin, not part of normal medical protocol, they discovered from their data that they can cut death rate death rate by 40%. Now, isn't that stunning? There's no treatment for stroke that does that large a reduction, but data science can do that reduction. And they've done it. So if you get a stroke, go to Kaiser. They'll do much better than Stanford because they so rely on data. Yes, and then I'll go over there.

Can you say your name, affiliation, and briefly Take your question.

Uh, I'm working on two questions for you. First off, I'm actually back on diabetic. Is when you see actually blood sugar test results, any—

Um, so a company called Agamatrix has done a pretty dumb integration. I call it dumb because it's sort of just a data transfer. And good entrepreneurs do clever things. So this device is not connected electrically to this. So the FDA doesn't consider this device part of this device, so we can get certification for this independent of this device. Turns out this beeps at this device. And if anybody— nobody here is old enough. In the old days, you actually took your phone and put it on the modem. So this talks audio to this device through the microphone on this device. It's stunning how clever entrepreneurs can be. Yeah.

Yeah.

You know, there are so many. So again, since we're in the medical school, you wouldn't think social has a lot to do with medicine. But let me give you another stunning example. I love this. I love my job because I get to hear all these great stories. But the simple example is PatientsLikeMe. That's a great example of social media being applied and social data being applied to medicine. But take mental health. Fairly large problem. A typical mental health patient for Stanford will cost the, cost about $12,000 a year. And the patient comes in, sees their psychiatrist, they do whatever. Here's a prescription for bipolar disorder or manic depressive or take your favorite mental health condition. And the patient goes away. Psychiatrist may see him once a week, once a month, fairly infrequently. Has no idea what's happening to this mental health patient during those places. If they get depressed or bipolar, what's the first thing they'll likely do? Get off their meds. Which makes the condition even worse. You think the psychiatrist knows? There's no way of knowing. So we invested in a little company called Ginger Isle. A couple of machine learning and data science people out of MIT did his PhD in machine learning around mental health. Turns out, before you get really depressed, if you're getting depressed, You're having a bad day or a bad night. This thing knows that you aren't texting as often. Or you call your friends and you're not calling your friends. This thing knows if you stayed at home or went out for dinner like you normally do. This thing even knows you stayed in bed and didn't go to the kitchen. This thing is a way better predictor of your mental health. Tracking you passively. You don't have to fire up the app. Once installed by a psychiatrist, it knows more about your mental health than if you spent 2 hours trying to explain your week to your psychiatrist. It's a much better diagnostician. Now we've gotten rid of cardiologists and psychiatrists. What do we want to talk about next?

Yes. So my name is Colin Carpenter. I have a question. You touched on a little bit, but I wonder if you had any innovative strategies or you heard about any ways to get through industries that have incumbents that are really dug in. Healthcare is one that you mentioned.

Any other that you can think of? I'll tell you a general principle. I don't— every industry is different, so I can't answer that question. By and large, people take large industries like healthcare. And they try and reform them. Hillary Clinton did and broke her pick on it. Very, very hard to take anything head on. The best way to think of this strategy is find a tiny little niche where you can get started but grow exponentially. Once you're on the exponential growth curve, you'll hit an industry before they know what happened. By and large, that's the strategy. The math of exponential growth is compelling. So don't look for large markets, look for small markets growing exponentially. Other questions? Yes. Follow up on, uh, Are you a data scientist? I'm always looking for— send me your resume. Is anybody with data science or machine learning background? Excuse me, let me— I thought I'd turn this off. Sorry, go ahead.

Difficulty in getting data out of— are you with me on that? Hospitals will have thousands and thousands of MRI scans.

And they won't give it to you.

You just go—

well, so you get creative. So Ginger.io is the company I just mentioned for mental health. By passively installing an app on consumers' things, and once installed, it does the tracking and gives you the data. Tells you a lot about mental health. That is why what I said is when I showed you this EKG machine in my pocket. If it's $10, don't try and sell it. Give 10,000 of them away to cardiology patients at UCSF. I said to them, in the first year, if you only get me— if all you do is give 10,000 of these away, because that's $100,000, not a lot of money, and get me 100 EKGs per, so get me a million EKGs, I don't care. Why? Because the data matters. Mental health data matters. CellScope's doing this little derm thing. What they do is it's really a controlled— lighting and imaging system that attaches to the normal cell phone camera on your iPhone. They control the light, they control the magnification so you can get medical quality images. And they can go from 3x magnification to 300 depending upon the application. 300 is not needed in most applications. They'll be doing simple things like letting you take a picture and send it to your doc. But they're also collecting the data. So you have to find excuses to collect data, by and large, in medicine. A few people are creative. I found Kaiser Permanente pretty innovative. And so they have pretty good groups there that are looking at data seriously. Most medical institutions have this linear view of data. I think Amir is the CEO of Stanford Hospital. He came by to see me. They're looking at Stanford 2020, this large building project, couple of billion dollars. What should we be thinking about? And they also sent IDEO, which is doing consulting. I said, you are off by at least a factor of 100 maybe a thousandth on the amount of data you will have on every patient. A cardiology patient, you shouldn't have 2 EKGs a year. You should have 200 or 500 EKGs a year or, or 1,000 EKGs a year. You're just way off in the amount of data. I think it'll happen. It'll accelerate only when the first few examples of good use of data become sort of storytelling. The statin example I gave you out of Kaiser Permanente, that story should spread. And if a startup had done that, it'd be in the New York Times. Let me go back there. Yeah. Um, what I'd say to you is there's every flavor of VC. There's no such thing as VC. There's every flavor of VC. There's VCs for every area, whether it's social or medicine or pharma or biotech or robotics, pick your favorite. There's also VCs with very, very different time outlooks. You know, we have a different fund to do what we call science experiments. So some things are very popular on Sand Hill Road. Internet startups that sort of can be sold in 2 years, that's pretty popular. But if you look hard enough, there's VCs that do almost every kind of investing. So don't make the mistake of categorizing it. It's sort of saying all doctors do X. But there's really great doctors, there's really horrible doctors, there's primary care physicians, and there's cardiologists, and every esoteric specialty you can imagine. And so I look at the world that way. Yeah, you had a question. I love the idea of patients owning their own data. I think it'll happen. It'll happen in Europe probably before it happens in the US. I think it's further along in Europe. But I haven't followed it too closely, so I may be ignorant. I think it's a great idea. I think institutions will resist it. I'm enrolled in a study at Stanford. And it's a study where every 6 to 8 weeks I get all my omics. My proteomics, my metabolomics, my microbiomics, my— probably 10,000 data points every 6, 8 weeks. I love the data.

What do you do with it?

Here's the problem. The IRB at Stanford won't let me have my own data. Won't let me look at it, afraid that I might not be able to interpret it. I know I can interpret it better than most people on the IRB, but you'll see that kind of resistance. But Michael Snyder is doing a study. He's a very progressive, data-oriented guy. I'm great, I love being part of that kind of study. In fact, what's funny is if I get sick, which I hope I don't, I'll have a baseline. Most of you don't have a baseline. It's silly in this day and age to not know what your normal was to know what abnormal is. My platelet count at 120 is normal. That's abnormal for almost everybody else. Over the last 5 years I've tracked it. I can't get it to move. So if I suddenly had an event and I didn't know what my baseline was, they'd think, there's a problem with my platelets. That's the kind of stuff. Most patients discover they have a heart attack after they have a heart attack, or a heart disease after they have a heart attack. How silly is it to only discover a problem after you have it? There's huge value in knowing 2 years in advance you might have a heart attack or you're in the 90th percentile of risk. Which is what my genomics says I'm in, the 90th percentile of risk. Then you can actually measure and do some things to reduce your risk. Yes. Sorry, did you have a follow-on? Well, it is what it is. It's a reasonable system. It's not perfect. It's sort of like democracy. It's a bad system, but every other system is much worse. So I prefer democracy to any system that's worse.

Yes. I have a question.

This is pretty naive.

So let me answer that question in a more general way. This is a pretty common problem entrepreneurs have. First, I'd say go back to my earlier answer on VCs, that every type of VC. I prefer to hear about it early because then I can help brainstorm a solution. Many VCs, 80% of VCs would like to see a buttoned-up business plan. But that leaves 20% of VCs who are open to more and more flexible. So I'm sort of always an optimist. About these things. Some are essentially risk-averse investors, and they like to see things buttoned up. I think they eliminate, in my view, all the interesting possibilities by doing that. What But it's not what VCs want. What I'd say to you first is think about what's right for you. People want to rush to get funding, develop a business plan, and pursue it. No matter what your area, it always makes time to think about what you're trying to do. And explore the vicinity, if I might. Know what the good things are. Know what the critique is of this, of your idea. I like to say, think of yourself at one of these British roundabouts, but go around in circles for a while. Investors don't like that, but go around in circles because if there's 5 roads going off, this roundabout in various directions. Once you pick a road, you will only get to where the road leads. You spend some time going around in circles and trying to scope out where the road goes, you're much better off. You'll pick a better road to start down. Then you can execute well. While you're in this exploratory mode, don't get your burn rate high. Don't get too dependent on timelines because you are exploring and you can't have a predictable schedule, and take that time because a lot of good evolution happens of your plan during this exploratory phase. Once you sort of start taking lots of money and start spending lots of money, you don't have the time to, to explore. So you end up in this execution mode, and very often if you're executing on the wrong thing, you'll keep executing on the wrong thing. You'll do very well the wrong thing. You'll get to the wrong place. So, yes. And then I'll come back there.

I wanted to circle back. My name is Shubhra Jain. I trained as a physician and as an engineer. I wanted to circle back.

I love that combination.

Thank you. I wanted to circle back. Going back to the question of data and related to developing countries or more so underdeveloped countries where in countries like India and places in Africa where there is not that much smartphone penetration and it's not so easy to collect data by just doing it via app. How would you say, what are some of the good measures of collecting data over there and what relevance So let me answer it again more generally.

If you go on our website, there's a paper there on entrepreneurial thinking written by somebody at a French business school. And they say what entrepreneurs do is a factual reasoning. I didn't write this, so I can speak about it. Not every problem is solvable, at least not today. And so what entrepreneurs do is look around and say, what problem can they solve today? Start on that path and evolve to doing the full thing. There's well enough smartphone penetration to get started in India. Okay? Enough. Not at the grassroots. Depends on what kind of— Let me disagree. At the grassroots level, there isn't enough reason to own a smartphone. If you had a great medical diagnostic system, people would buy a smartphone. It's cheaper than buying a doctor. Even— look, if you're in rural India and your choice is every time you need to consult a doctor, you have to take a bus or a bicycle 30, 40, 50 miles away and waste a whole day away from work, you're going to get a smartphone. Maybe it's a shared smartphone among a few families. That's what happens all the time. But if you have no reason to own one and you can't save money owning one, you won't get one. In any reasonable time frame, like 5 years, almost everybody will have access, even in rural India, to a smartphone. So think of the world as it will be, not what it is today. And again, if you want to invent the future you want, you create a great medical app on a smartphone. People will start getting smartphones for healthcare. Yes? Push stock in. Yeah. Okay. Well, so first thing I'd say to you is the amount of fundamental research I think has gone up, not down. I'd like a lot more. Everybody would like a lot more, but you have to deal with the reality of affordability. And if the economy does poorly, there'll be larger cuts, and if the economy does better, there'll be more resources. It just, you know, you have to deal with the facts of life. It's one of these facts, these research funding goes up and down. But by and large, if you take a long-term trend, especially including productivity in research, It is going up. The amount of venture funding is going up. But none of this is perfect, but you deal with it. I mean, I really wish we'd allocate more of our resources into fundamental research.

But can the VC do that?

VCs can do some, but not everything. I mean, you can't eliminate the need for fundamental research. And you do need institutions and funding sources for those institutions that'll encourage that fundamental research. But at some point, people can take it— so take an area like synthetic biology. I think most of the research has happened not in universities, but in private companies and venture-funded and Other questions? Yes. Mechanical engineering. What I would say to you first is do what you're passionate about. The thing about startups people don't realize— there's a talk on my website called Entrepreneurial Roller Coaster, where the highs are high and the lows low. It really is a roller coaster. You have great days and you have horrible days. You have really depressing periods. You start saying, why did I do this? You have to be prepared. Now, if you have religion about what you're trying to do, if you have passion, because you get past the downs, and you should. If you don't, you tend to abandon them at the first sign of trouble. And so passion is a really important piece of what you want to pursue, I believe. Having said that, I think it's fair to say If you tested those 3 ideas with, say, investors, you could get a lot of good critique of each of those ideas. I wouldn't go to people and say, I'm doing 3 of them, and don't do this to me. But you can waste other people's time taking one at a time and say, what about this? And then somebody else, what about this? You'll get good critique. It's important to look at what's great about an idea, but it's equally important to look at what might fail, what its liabilities are, what are the problems with it. And as many of the negatives that you can flush out early will help you make a decision.

Yes. And is it good to give up an idea just because you've fleshed out too many negatives?

No, there's a difference between understanding your negatives or liabilities and having them deter you. If there's enough negatives and not enough positives, then in fact it should deter you. If there's a few negatives and the upside is great, the consequences of success is large, then it's worth taking the risk of, can I solve these negatives or not? And the bigger the upside, the more the negatives you take on. I mean, take something like ground beef I mentioned. Seems unlikely, but if it works, the consequences is so large it'd be criminal to not try it. So we said, sure. And Pat Brown gave up his Howard Hughes Fellowship because he said it'd be criminal to not try it. That's passion. No matter what the problems, he's going to keep at it and get creative. Most creativity happens when you run into brick walls. And that's when you give up if you don't have passion. That's why passion for what you're doing is so important.

Yes. So, my name is Dylan Mondego. I think I'm a little younger than this crowd. I just finished undergrad. And you have a pretty cool job. You get to talk to a whole bunch of different people and learn a ton about a bunch of different things, right? And I was wondering what, in your experience, was the most exciting company that you came across, or just the one that you fell in love with the most, like this, Molly's Bar and Grill?

Well, there's way too many companies I fall in love with, so I can't tell you one thing. But what I love is diversity. I spent time on robotics today. I spent time on this beef idea. I spent time on a medical idea that's pretty breakthrough if it works. You know, somebody said they were diabetic. We just agreed to fund— a good spectrometer is $30,000, $50,000, 1-nanometer resolution, wide spectrum. If this one's right, it'll cost $5 for that 1-nanometer resolution. I don't even know what you do with such a device, but sounded pretty damn cool. That, you know, to your idea of do you believe in the application, they thought what you should do is attach it to your iPhone and zap your food and see if the— what the fat content is and what the protein percentage is. 15% protein, 30% fat, throw that meal out. I don't know if it's practical or not, but it just was cool technology, fundamental innovation in how optics is done. Worth doing. We might collect a lot of data. There was a question here, and then I'll come to you.

My name is Mana Lewis. I trained in statistics in France. And now I have started a company told me that their passion is art, that deals with art. The question is, I have to evaluate a team member because I'm building the team and I feel difficult understanding how to evaluate a team member with very valuable long term, but in the short and medium term, really able to impact towards a very high level from equity and all those things, how would you advise?

So you have a hard question.

So let me say one of the hardest things an entrepreneur does is decide whose judgment to trust on what topic. Most people ask people around them, their friends. They're not qualified to help you answer these questions. And so whose judge— whose judgment to trust on what topic is one of the most important things entrepreneurs do. So when you get feedback on your idea, Spend time thinking about how valid that answer is. How qualified is the person to answer that question? And people never examine this. And the way human psychology works, they get biased a certain way. So let me be a little more helpful. Go to my website and— There's two papers. It's almost a book. I wrote them personally based on lots of experience with this hard problem. One's called How Do You Engineer the Gene Pool of a Startup? It should be under entrepreneurial resources on our website. How do you really engineer a gene pool? And you have to engineer the gene pool both for your success factors, what your assets are, what your strengths are, but also for your weaknesses. What's your liabilities? What are the negatives? You can engineer a team to minimize the impact of your negatives. And that's what I mean in this gene pool engineering paper talks about that. Unfortunately, it's about 20, 25 pages long. It's boring reading, but I wrote it, and it took a lot of time and effort to write it, so. The other paper I'd refer you to is how do you recruit a partner, especially in an area you don't understand? If you're a neurobiology person, you're trying to recruit an optics person, how do you judge that person? So there's a paper that's also 15 or 20 pages long written on that topic. There's a lot of resources on our website meant to answer— these are real hard questions entrepreneurs face. This is the hard part about being an entrepreneur. There's no source of information. And so I've committed to keep writing these. I probably write about 100 pages a year.

When you decide to work with a company, do you provide this type of mentoring?

Absolutely. That's most of what I do. So the last thing I'd say is very tactical. If it's very high level, I wouldn't get him as a partner. I'd get him as an advisor or some informal relationship. You want people who do stuff. That's what entrepreneurs do. Try, fail, iterate. That's what entrepreneurship is about. That's the kind of team I'd build. Without knowing the person or anything, that'd be my bias.

Yes?

I think it's going to be the last question so that people have time to network with you at the end. OK.

Um, I'm Boris Coleman. I'm a neuroscientist here with a background in physics. You expressed a lot of optimism that technology could be used to solve almost every problem out there. I agree with you.

I didn't say every, but most large problems technology is a significant contributor. So, but I would add one more thing. But just because you get the technology right doesn't mean everything works because sometimes politics gets in the way, sometimes big companies, somebody else block you. So there are lots of hurdles. But if you look at the fundamentals of, of sort of the planet, I look at the world as 500 million, maybe a billion people, but really 500 million people have an energy and resource-rich lifestyle. That means they have instant access to a doctor, they have air conditioning in their homes, they have cars and fuels and all that kind of stuff. 5 billion people want it. We need 10 times more resources on this planet. That gap can't be met physically by more mining of copper or more minting of doctors. It's technology that's the solution. But technology, I believe, is necessary. It isn't always sufficient. Anyway, sorry.

But yeah, my question was, uh, what pessimism— or put another way, this technology is going to become a much larger part of our life. It already has, clearly, over the last 10 or 15 years. What systematic problems— or rephrase this, opportunities— problems to be solved? Well, you know, you cut off your finger, you're system doesn't break. And in general, that's not true of technology.

Yeah, some technology.

Some technology. So you— are there— can you think of problems that this—

well, problems are problems. I think we have enough of them. We don't need to find new ones. We can apply technology to most of them. Politics may be one that's harder to do. But even that, I think, will gradually and is starting to change. What I would say to you—

Is—

so I think over time, when we solve our current problems, will other things that don't appear like big problems will start to appear bigger problems. There's always a queue of problems. Today we worry about the top 5 or top 10. If you look at all the polling that was done for the presidential election and the elections in general, people worried about number 1, 2, and 3 of their problems and voted on them. When you solve 1, 2, 3, 4, 5, and 6 will show up. And once you solve 4, 5, and 6, 7, 8, and 9 will show up. And every time you solve a problem, you often create another problem or a different problem. And so, look, this, this is sort of the evolution of problems, but From my point of view, it keeps life interesting. One thing I'd say is if there was no problem, life would be pretty damn boring. You'd have nothing to work on, nothing to be proud of solving, nothing to make a difference with. And every large problem to me is an opportunity. Because if there wasn't a large problem, nobody's going to pay you to solve it. Then it isn't an opportunity. And all large problems are large problems because somebody hasn't solved it, which means they are hard. And so all large problems are hard, but not all hard things are large problems. You can work on hard things that are inconsequential too. So probably a good place to stop. Thank you all for coming. Thank you very much.

Before you all go, I have two quick announcements. You are all invited to attend M's PBL next Monday. A former postdoc from Stanford is going to share his experience about starting a medical company. And you're also invited to attend EM Social Events that will be co-organized with the GSBE club on Thursday. So thank you very much and see you soon.

Thank you.