Viewpoints

Stanford President Emeritus Dr. John Hennessy Shares Insights with Mayfield


October 8, 2017 –  

 

 

Dr. John Hennessy, president emeritus of Stanford University, participated in a fireside chat with Mayfield Managing Director Navin Chaddha at a gathering of CXOs and entrepreneurs on September 6, 2017. Dr. Hennessy has been called “the godfather of Silicon Valley” for his work, including founding MIPS Computer (which was incubated at Mayfield), and his long-time board roles with Google, Cisco, and Atheros. This is an edited version of their conversation.

The importance of passion for work

I love what I do. I try only to do things that I’m really excited about, because I find that it’s very hard to maintain the energy to do something you’re not great at. There’s a great line in Steve Jobs’ commencement speech where he says, “When I get up in the morning, I look at the mirror and if I just had too many days in a row that I don’t like what I’m going to do that day, I know it’s time for change.” Well, that’s the way I feel. You should be really excited about what you’re doing and passionate. Whether it’s starting a company, building something at the university, whatever, you should be passionate about it.

Iconic companies at different stages of their lifecycle

Cisco and Google are two companies in completely different parts of their lifecycles and ways of doing innovation. Google is largely organic, drives innovation internally; Cisco has done a lot of its growth and expansion by acquisition, bringing in new teams. These are completely different ways to think about how to operate. Cisco benefitted from when the internet grew like a bandit for years and years, and now the number of people who are going to need new network equipment isn’t what it once was. Meanwhile, the spread of the internet, and the change in applications, and what Google is doing, is changing the world very fast.

The essence of Silicon Valley & Stanford

Silicon Valley is a constant hotbed of new injection of technology, new knowledge, new uses of technology — which is just as important as new technology. You take a company like Uber: it’s really about using a technology that’s well established to create a whole different business model. This is that constant injection of newness. Fifteen or twenty years ago, people were going to try to build “Silicon Valley Two” somewhere else. But what’s happened…is that the Valley’s lead as the tech center of the world has gotten bigger, not smaller. It’s partly our ability to bring in talent from around the world. If we ever get so stupid to stop importing the very best people from around the world, it will be over.  We have to continue to do that in order to thrive. In Stanford, and Berkeley, you’ve got the best private and public higher education institutions in the entire country, in the entire world.

Stanford is a place of optimism in three ways. First, it is a research-driven university. There’s a great belief that knowledge will have important practical impacts on our lives. You can’t always predict how long it’s going to take to have that impact — but there’s a real belief that search for knowledge is critical to progress for the entire human race. Second, how can you not be optimistic when every year you have this incredible 3,500 to 4,000 new people coming into the community, who are bright, ambitious, and want to make a difference. Then, we believe a university is also about figuring out how to put that knowledge to work. That’s why there’s so many startups at Stanford. People want to see their discovery, their invention go out and change the world, so they’re willing to start a company to take it out and really try to change the world with it.

Ultimately, people power technology

When I started MIPS, the truth is, I didn’t know what I was doing. The business plan that we brought to Mayfield … there are few slides saying, “Here’s this technology. We don’t know how to build a company. We need help.” I didn’t know how to recruit a CEO. If you asked me what fraction of the budget of the company should go to engineering, I would say, “at least half.” … We put that first business plan in the Computer History Museum in Mountain View because they asked us for early stuff. There’s no projection of market, revenue, nothing like that. Basically what it says is, “Here’s this great technology and if we build it, they will buy it.” That’s it. At the time I didn’t think you needed sales and marketing because I thought great products would sell themselves …. If you believe that, you should go do cold call sales for a month. Sales requires courage and commitment. That was a great learning lesson for me. The other thing about people is — you can’t do everything yourself. You’ve got to be part of a team, and you’ve got to align yourself with the success of the entire team. The notion that you’re going to be successful, and to hell with everybody else — that doesn’t work.

Innovation matters for startups

The first lesson for entrepreneurs is that innovation has to be the driving goal. You’ve got to be thinking, “How am I going to innovate? How am I going to do something different? How am I going to break out from the pack?” Especially today, when so many companies are funded with somewhat overlapping business strategies.

A lot of students who come to see me say, “I want to start a company.” I say, “Tell me about your technology.” “Well, I’m not sure what it is yet —but I want to start a company.” I say, “How many photo-sharing sites got venture funding?” The answer is over 100 — and you all know about Instagram. That’s a lesson to everybody. You can say, “I’m going to be the one to succeed. I’m going to be the Facebook, not the MySpace.” But you’ve got to think about how you build a team that’s really going to do that. How do you get focused? What I tell people about startups is, you do a great job on your first product because that enables you to do a second product. You do a great job on your second product and maybe you have a company then. That’s what helps people think about getting focus and drive.

Innovation matters for established companies

I think about this core question: how are you going to keep internal innovation going in your company, your university, your country — how are you going to make change the order of the day, something that you’re constantly focusing on? It’s so easy to slip into just doing the next incremental step. As you get bigger as an institution, all those new business opportunities look like a drop in the bucket. After all, if you’re a $10 or $20 billion operation, why do you want to start a business, which for the next few years will probably struggle to get to $100 or $200 million? Because someday, that business is a billion, $2 billion, or $10 billion business.

You’ve got to constantly be thinking about better organic growth. You almost can’t start it too early. I think what’s happened in many companies is that they wait too long. They wait till the company has plateaued in its initial core technology that made them great. Then they start to look for these new businesses even knowing that new business won’t be interesting for five years. That’s too late. You had to start five years earlier when you’re still growing your core business.

Finding the right people

One of the biggest challenges any young growing company has is getting good people. When we were growing MIPS early on, we felt so pressured — I realized we were sacrificing quality because we were desperate to get people. It turned out to be a mistake almost every single time. In hiring, I think you’re looking for integrity, passion, commitment, some sense of being a team player, courage.

The role VCs play with the team can be important. As VCs, you don’t want to subtract value— you want to add value. Obviously, good venture firms like Mayfield have a network. Using the network that Mayfield has to help find and locate people is a critical asset. Coaching through financing — a critical issue for companies, just asking good strategic questions. I think the best board members I’ve known, they’re not the people who spent all their time studying the financials and looking at the balance sheet. They’re the people who asked the good strategic questions: “What are you going to do about an emerging competitor? What are you going to do about this market opportunity?” That’s where the real value comes.

The lesson I’ve learned is if you really want to develop people, you’ve got to think about creating opportunities for them to go up that ladder. Don’t take somebody really promising who doesn’t have the experience and try to force fit them into a job they’re not ready for — you’re going to end up doing damage to their career. If you take somebody who’s really talented and you jump them up, skipping a few rungs, the failure rate is remarkably higher. It’s very hard to jump over rungs, because you haven’t learned certain skills, like how to deliver bad messages. How to coach an employee. How to deal with the crisis. How to make a tough decision. Had you prepared them in a more logical way, they would have been incredibly successful.

The importance of succession planning

First, you have to build a great team. You try to hire people who are more talented than you are. I have noticed that people who are really capable and feel comfortable in their own skin will hire people better than they are. People who are worried and nervous about their position in the world hire weak people so they’re not challenged. That’s a real mistake, because it creates a real succession planning issue

As a board member, we want to do a yearly review. I want to know who is your number one or number two person under each major position inside the company. We started doing this at Stanford because universities typically don’t do it. That’s a disaster. I don’t want a Dean who gets ready to step down and says there’s nobody who could be my successor. They failed as Dean because they missed building up the opportunity.

Candor and failure

All my direct reports know that they can say to me, “Look, Hennessy, that’s a really stupid idea. Do not do it.” I need that. You need people who push back on you. People who shoot the messenger become very lonely, very isolated and make more mistakes. And when you do make mistakes, or admit failure, you have to acknowledge it. A number of years ago, we were invited to propose building a campus in New York. We made a strong pitch. What happened in the end was we couldn’t get on the same page with some aspects of it — we would not agree to grow faster than we thought we could hire high quality people, because I wasn’t going to compromise the quality of the faculty. Too many universities that have two different campuses have an A campus and a B campus and none of the people at the A campus would ever go to the B campus. I didn’t want that. But New York City wanted a different set of things in terms of guarantees for a number of students and a number of faculty. I made the decision overnight to pull the deal. We had spent at least a million bucks, had the team lined up to do it. But we had failed to convince the city of the importance of the model that we had.

AI and e-learning

Here’s what we’ve learned. First of all, the problem with an online learning system is it’s not very personal, not very motivational. Some people are good enough that that doesn’t matter. The e-learning stuff works really well for highly motivated people who are looking for specific skills, and talent typically related to their career. An executive MBA, or a sequence in cybersecurity, that works really well for somebody who’s already got a college degree. It’s not a replacement for the undergraduate experience at all. But e-learning will play a bigger role in the developing world for the simple reason that there are fewer teachers there. If you don’t have educated teachers, then the online experience is better. I think we still haven’t really brought our knowledge and machine learning to bear on it to build adaptive learning systems.

Imagine an e-learning system with a lot of built-in machine learning that’s learned from lots of students. When you get something wrong, it has a pretty good idea of what you don’t understand about a particular topic and can direct you. We’re not there yet, but I think a few more years of research and development and we can begin to build systems. At least primitive things. Teach kids algebra, basic math skills, things like that. Harder for other things.

Healthcare and machine learning

Three things are driving what’s happening in healthcare. First, we’ve got to deal with cost issues. Our system is fundamentally broken. The rise of big data and machine learning is going to mean that we can build computer systems within a short amount of time that will out-diagnose 90% of the physicians in the world. A recent Stanford study [August 2017] had machine learning doing EKGs against a board-certified panel of cardiologists; the machine diagnosed heart arrhythmia better. Why? Well, look at an EKG. There’s a lot of information there. It’s fairly complex. You’re looking for complex patterns. You do still need the expert; you just need them once to get the system knowledgeable, and then you move on. That’s A.

B, we need real focus on the illnesses we are increasingly going to face in this country, namely neurodegenerative diseases and cancer. Your odds of dying of one of those two diseases is going up year by year. If you’re under 40, it’s going to go up pretty quickly, because we have pretty good ways of dealing with most other conditions. The third thing I think we see happening is a real push on precision health: getting data, getting your DNA sequencing. Look at chemotherapy: roughly half the chemotherapy administered in the U.S. does no good on the cancer it’s attacking. And meanwhile, it’s killing you. We’ve just got to advance that technology much faster so that we can target drugs much more to the individual involved. That’s going to happen over time. We’ve got a lot of work to do, though.

A decade ago, my view was that biological and biomedical sciences were going to supplant information technology as king. Then along came machine learning and AI. All of a sudden, there’s at least parity. Now the overlap area is trying to take deep learning, monstrous databases, and apply them to the healthcare challenges we face. Suppose you could get the digital patient records for everybody in China. You then get all their DNA sequenced, and now you do the big computation and begin to really understand the genetic basis of human health at scale — this has got to be at a big scale. You need millions and millions of patient records because lots of diseases are fairly rare in terms of their genetic basis.

The responsibility of Silicon Valley companies

I am worried about the hollowing-out of middle class jobs which require at least some college level education, but which don’t really have deep skills and knowledge. Those are jobs that are very susceptible to replacement with tech, especially as you advance AI and machine learning. That hollowing out is going to increase the bi-modality of income distribution in the US. And that’s a real challenge. If the question is, do large tech companies like Google and Facebook have a responsibility to help in this, they have a responsibility to three communities first: shareholders, employees, and customers. Community is in fourth place. But it’s a very good question.

Locally, if we don’t solve the issue of housing and transportation in this area, we will strangle what has become the technology center of the world and it will go somewhere else. And I think for better or worse, the lead is going to have to come from the company leaders around the Valley, because our local governments are semi-dysfunctional with respect to solving it. It’s going to have to be people who come from the tech side and say, “This is critical to the future of what we try to build in this Valley and we have to solve the problem.”

On AI and jobs

All technologies we developed can be used for good or for bad. That’s always true. I don’t believe you can put the cap on a technology. You put a cap on AI development here in the U.S., I guarantee you China won’t have a cap. I think you can see the difficulties of that. We don’t want to supplant the human role in the world by building AI, which replaces core human functions. You want to enhance what humans can do with AI, as a way to leverage human brain power. Think about any big machine learning system. Think about ImageNet, recognizing images. Where’s ground truth? How did you program the system to begin with? It had to come from people. There’s going to be an ongoing role for people in terms of developing systems, and then using that technology in an appropriate fashion. The point is, it’s going to change job makeup in the U.S. What we ought to be thinking about is how to help people find fulfilling career opportunities that will make them a good living and engage them. That’s what we ought to be thinking about.

On the future of digital currency

Clearly, there are some motivations to create a currency other than the dollar that have some digital basis and a level of interchangeability that is immune to governmental control. But I don’t know why you need 20 of those currencies. A viable digital currency could become a universal currency then if you could really make that happen. But to have 20 universal currencies is not going to work.

Early lessons from Google

Search turned out to be the killer app. That was a bit of luck, that they were building a really great search engine. At the time, Alta Vista was great, but when I saw Google search, I said, “Forget it guys. This is so much better.” Google was highly quantitatively driven, with constant improvement. The mantra was – make it better. Make it better. Make it better. One of the amazing things about Google is they were able to get ad placement to be better, so the ads are more interesting for the user, which also works for revenue. Eric Schmidt once said, “We’re trying to build something that operates a bit like a university, a bit like Stanford operates, that has that kind of innovative spirit.” They’re willing to try new things and experiment and try them out, and that makes a big difference.

On genetics and ethics

With recombinant DNA, where there was a coming together of the technical community. There was a famous meeting in Asilomar where the scientists came together and decided, “Here are the boundaries on recombinant DNA. We are not going to take genes from pigs and mix them in with the human genome and implant them in the gene line.” I think we need that around CRISPR. What exactly should this technology be used for? What’s its real role in the world? How does it compare with alternatives? If you can test for Down’s syndrome, there are probably better ways to fix it than try to build a CRISPR fixer. The other thing that’s clear is that a CRISPR may be doing a little more editing than you want sometimes. What’s the implication of that? Are you going to allow people to edit their gene line so all of a sudden a CRISPR-edited gene becomes inherited? There are a lot of things here that we need to figure out much more carefully. We haven’t had those ethical decisions and those absolutely need to be had.

On US innovation and global competition

Since the Second World War, the U.S. has the best innovation system in the world. It happened by aligning things — incentives for startups, federal investment in universities where most of the research went on. This produced many of the great things that have happened here in the Valley and around the country.

I think we all worry now about the inability to deal with entitlement spending and the anti-elitist thing that seems very popular now in Congress. It’s kind of crazy to be anti-smart. I worry about that. We’re not going to be a great country by doing lots of incredible real estate deals around the world; we’re going to be a great country by continuing to lead in innovation. Look at AI. How many years of investment have gone into building the breakthrough that occurred with deep learning. 40, 50, 60 years — ? That’s how far back AI goes. The government kept investing, kept investing, kept investing in it. AI’s breakthrough came because we built all those layers of knowledge and capability. If we lose that, it will be the beginning of our trajectory towards being a second-status country in the world for innovation.