Section 3: Leveraging Disruption for Society:
Vinod Khosla, Khosla Ventures: Non-Linear Futures and Disruptive Change
Vinod Khosla, Khosla Ventures
In your view, how does disruption generally happen?
There are two ways to predict the future. One way is to extrapolate the past – the domain of economists – and the other is to invent the future that you want, which is what technologists do. The future that you want is usually non-linear, different and full of possibilities, usually driven by the passion of imagining what the future might be by an entrepreneur, not an extrapolation of the past, which big corporations and incumbents tend to want to continue. An example is when AT&T was developing their 15-year plan in 1986, and so they asked McKinsey how many phones would exist in the year 2000. The prediction was around 1 million phones in the United States; the reality was actually 109 million phones. Academic prediction and forecasting does not work, especially when it comes to the future, because to have even a wild shot at predicting the number of phones in 2000, one would have to predict screen technology, microprocessor and sensor technology, battery technology and packaging technologies in 2000, along with their costs. What technology predictions and their integration can we make today for the year 2030, when we couldn’t predict the iPhone in 2006? Almost everyone assumed then that Motorola and Nokia would continue to be dominant players. They failed because they failed to imagine the future and to have the courage of their convictions to make it happen when they did imagine it.
Yogi Berra [retired American Major League Baseball player, coach and manager] was famous for saying that almost anything interesting has two parts. Imagine there’s the world where things happen according to a set of rules, and there’s the completely chaotic world with no order. A lot of research says that the most interesting developments happen at the edge of order or the edge of chaos. And as Karl Marx said, “When the train of history hits a curve, the intellectuals fall off”. That’s where good entrepreneurs operate and they, in fact, drive the curves, twists and turns when chaos and possibilities “un-gel” the possible future. That’s where they combine and recombine things to open up new possibilities, what I call “imagining the possible”. They imagine the possible in unusual combinations as opposed to big companies. And they use successive iteration, persistence and an ample dose of foolishness to get their longer-term vision right.
Getting it wrong isn’t proof that you’re not right; it’s just proof that your first attempt has something wrong with it. For example, there were six attempts at Facebook, from MySpace to Friendster. Once you have a hunch, and you refine it based on engaging in the field, successive learning and refinement, that’s the right way to change the world. There are no academic forecasts or big consultants who predict this. They are just completely clueless when change is happening and extrapolation of the past does not work. McKinsey will be too high-level to even remotely be relevant or actionable, because details are critical to iteration, which is critical to learning and refining views successively. Let me illustrate that by asking, who changed retail, Walmart or Amazon? Who changed media, NBC or YouTube? Who changed space launches, NASA or SpaceX? Who changed cars, General Motors or Tesla? Who changed car services, Uber or Hertz? Who might change driverless cars or payment systems? I’m hard pressed to find one major innovation from a big company, with the only exception being the iPhone, which was done by an entrepreneur, not a company, and that was entirely one person’s vision. Google also operates more entrepreneurial than most large companies. But almost all interesting launch phenomena come about from entrepreneurship and unusual ideas that were traditionally not accepted and with countless reasons why they wouldn’t work. Then, suddenly, the assumptions are falsified, often through multiple iterations, and one day it works, making the world a better place.
I’ll give you a stunning example which is not well understood. In 1996, we were starting a company called Juniper Networks. We were convinced that, in the future, most of the world’s networks would be run by internet protocols known as TCP/IP [Transmission Control Protocol/Internet Protocol], and not the existing ATM [Asynchronous Transfer Mode], the technology almost every carrier was planning to deploy because of their view that the average home needed no more than 64 kilobits per second of bandwidth, then called ISDN. Google, today, is planning on 1 million kilobits per second of bandwidth in every home. I went to every customer and, with one small start-up exception, all the clients said they wouldn’t adapt their protocols, as the internet was expected to be based on ATM. Even Cisco decided not to do TCP/IP for the public internet and bought an ATM company called Stratacom, because that is what the customers wanted. Looking back, you won’t find a single article saying that TCP/IP was going to be the future of the public communications world. No major company believed in it. At Juniper, we went ahead without the customer demand because we thought it was what the customers needed, even if they did not want it. Initially, we found a few smaller customers; but, most significantly, the internet grew faster than ATM could cope with, and eventually overcame the momentum of customer extrapolation of the past.
In challenging those assumptions, I remind people that when I started Sun Microsystems in 1982, I requested a minimum memory in each workstation of one megabyte at a time when the Intel microprocessors had a maximum of 640 kilobytes. That’s the nature of innovation. That’s the nature of imagining the future versus extrapolating the past, which Intel was doing. Today the typical memory size is one thousand times larger, not something Intel would ever have forecasted then.
In the end, the power of economics matters when new technologies make a new approach possible. The only question is, how can incumbents slow or stop fundamental economics?
You are very interested in healthcare. How will these insights be applied in that industry?
In the end, the power of economics matters when new technologies make a new approach possible. The only question is, how can incumbents slow or stop fundamental economics? Imagine for a moment that the cell phone can do more than the doctor – and that’s easy to believe. If you look at the developing world, many countries don’t have a sufficient doctor-patient ratio. Former New York City Mayor Bloomberg recently spoke with me about the fact that while there are 500 people to a doctor in the US, it’s over 50,000 people to a doctor in Tanzania. If a patient needs treatment, that can be a death sentence; and so, high-school graduates are being trained for the medical profession. However, almost all diagnoses can be done much better by machines. Once machines are better, why would you need a doctor for diagnosis, prescription or monitoring? Perhaps mentoring, motivating and coaching, but that’s a different kind of job. I jokingly say that the best doctors may come out of the UCLA [University of California, Los Angeles] Film School, since they have more empathy, rather than from the schools that select for the highest IQ.
Contrastingly, the American Medical Association (AMA) is campaigning for what’s called a reduced scope of practice, whereby they are trying to get nurses to do less, they are fighting prescriptions for antibiotics by nurses, and a doctor can’t give an eyeglass prescription without an optometrist, when an IPhone can do that prescription easily – and hundreds of millions of people who need these prescriptions in India cannot afford an optometrist. I think that the unbundling of healthcare will happen one piece at a time, once computers can do it better, using the traditional slippery slope to innovation. IBM Watson is a start, but I see much more impactful and sophisticated approaches from start-ups in every area, from cardiology to psychiatry. It is hard for me to imagine that a psychiatrist can do better than a smartphone in diagnosing, treating or monitoring mental disease.
Developing countries are becoming flexible on healthcare patent enforcement to help increase access to medicine. What role will patents play here?
Patents or no patents, it comes down to economics. Patents will remain important, but not as we currently see, where 95% gross margins can be charged on a drug because of patents. Consider a cancer-screening test like BRCA 1 or BRCA 2, which can cost $4,000 per test. Very few people have access to it, and many insurance companies won’t cover it. When generic providers can supply that test for $200, then the population economics become very different – both our approach and practice will change, patents or not. So I think those kinds of patents will be important, but I don’t think that’ll be the driving variable in disruptive change.
In the 1990s and 2000s, the internet was the key driver of disruption across industries. You seem to think technology will be the key driver across healthcare. What will be the important technologies that will disrupt other currently inefficient industries?
All sorts. There are a lot of interesting technologies emerging. 3D printing might allow us to print new replacement organs 10-15 years from now, and, in fact, a 3D trachea has already been printed and put to use in a human being. That’s pretty stunning. 3D printing and cloud computing will play a role, but so will non-linear developments that are not yet understood. Take machine learning: for the first time, this is supplanting human learning, and that’s never been done before. The last time we replaced a significant human capability was when the steam engine replaced human muscle power. I suspect that machine-learning technologies driven by data from very inexpensive sensors – many of them wearable – will swamp what we know from today’s medical equipment, tests and physiologic measurements, though some of these, like MRI machines, will still be relevant. Already we have a company that can take 400,000 vital-sign readings a day at near zero cost. When data becomes immense, humans will not be able to process it, and machines will develop insights based on this data’s complex patterns, and make diagnoses, prescriptions and do monitoring. The 30-lab-test technology for blood may still be relevant, but it won’t be as precise as reading 300,000 biomarkers from every blood sample, as Applied Proteomics is doing.
How will this impact society at a larger level?
Income disparity will increase dramatically, and that is structural, but that will happen in conjunction with good GDP growth and an era of relative abundance. The old economic argument that “for every job that technology replaces, new jobs are created”, will not be valid forever. While economists happily measure productivity, the more relevant driver of change will likely be machines surpassing human judgement in more and more areas over time. I think we will see a fundamentally different type of change. We may have to support much of the public who choose to pursue other passions that are not regarded as work as we define it today.
There are already new kinds of jobs emerging, as we are seeing in entertainment, for example, but it is hard to predict what new types of jobs will emerge. Some may be productive for society, and others may not. Others will emerge from new possibilities opened up by technology.
We will see an increasing abundance, with the worst-off being in a better position than they are today, but with much larger income disparity. This will happen simultaneously and contrary to today’s traditional economic models, which are based on extrapolation of the past. While economists twiddle, the classic Karl Marx “train of history” could hit a curve that renders traditional arguments around “labour versus capital economics” obsolete in the face of more fundamental new drivers of economics – the economics of ideas and technology innovation, driven by entrepreneurial energy. It could drive the need for a new kind of capitalism of innovation, instead of the incumbency capitalism we have today.
The traditional economic idea of “capital versus labour” is broken. Now we have a third factor in the economy of ideas and entrepreneurship: generating wealth and ideas that increase the average while decreasing the median, usually through some sort of economic efficiency.