Course lectures "Startup". Peter Thiel. Stanford 2012. Session 15

In the spring of 2012, Peter Thiel ( Peter Thiel ), one of the founders of PayPal and the first investor of Facebook, held a course in Stanford - “Startup”. Before starting, Thiel stated: "If I do my job correctly, this will be the last subject you will have to study."

One of the students of the lecture recorded and laid out a transcript . In this case, astropilot translates the fifteenth lesson.

Session 1: Future Challenge
Activity 2: Again, like in 1999?
Session 3: Value Systems
Lesson 4: The Last Turn Advantage
Session 5: Mafia Mechanics
Activity 6: Thiel's Law
Activity 7: Follow the Money
Session 8: Idea Presentation (Pitch)
Lesson 9: Everything is ready, but will they come?
Lesson 10: After Web 2.0
Session 11: Secrets
Session 12: War and Peace
Lesson 13: You are not a lottery ticket
Session 14: Ecology as a Worldview
Session 15: Back to the Future
Session 16: Understanding
Session 17: Deep Thoughts
Session 18: Founder — Sacrifice or God
Session 19: Stagnation or Singularity?

Lecture 15 - Back to the Future

This time we were joined by four guests for a conversation after the end of the lecture:
')

1. Danielle Fong, Co-Founder and Principal Scientific Advisor, LightSail Energy;
2. Jon Hollander, Development Director, RoboteX,
3. Greg Smirin , Chief Operating Officer, The Climate Corporation;
4. Scott Nolan (Scott Nolan), head of the Founders Fund, previously an aerospace engineer at SpaceX (Elon Musk was going to come, but he is now busy launching missiles).

I. A look at the future from the past

Sometimes the best way to think about the future is to think about how the future was before. In the middle of the 20th century it was still possible to talk about the future, in which the weather could be accurately predicted or even controlled. Maybe someone would even find a way to predict a tornado. Or make it so that you can dispel the clouds. As for transport, people thought that they would move using flying machines or passenger submarines. Robotics also appeared to be another exciting and significant frontier.


But let's rewind the tape to today. Things have not changed as much as we had imagined in the 1950s and 1960s. The weather is still not too controllable, and people have already accepted this fact as inevitable. The main feeling about this is that attempts to control the weather are dangerous, and you should not get too hung up on it. With transport came the same disappointment. Forget about flying cars, we are still stuck in traffic jams. In robotics, however, there is some progress, but of course, not the kind that everyone expected. We were expecting a robot from the TV series “Lost in Space”, and in return received a robot vacuum cleaner.


All of the above contrasts sharply with computer technology. There is no such idea from the past concerning the development of computer technology that has not yet been brought to life. We saw Gordon Moore's Law in action, the persistent reduction in energy consumption, as well as the ever-increasing communication capabilities. Predictions about the future from the past have more or less come true in the computer field. Two-way wristwatches with a radio and a Dick Tracy TV are, in general, iPod nano. Arthur Clark practically predicted the appearance of the Internet in his 1956 book City and the Stars. So at least in the computer field it is quite possible that you will not find too much unfulfilled if you delve into the past.

/
But in most other industries, things did not go as expected. Expectations of the future did not materialize. Nuclear energy is another area where we were all confident that everything will work. Instead, everything turned out to be much more dangerous than expected. There were many assumptions about the possibilities of nuclear energy, which have not been realized, at least for now.

So what could work today? One way to find an answer to this question is to seriously consider the question of returning to the future. In the past, there may have been entire areas of research that were not considered promising, and it is worth returning to them. Not everything that has not been adequately investigated deserves oblivion. In the last session we talked about the energy of thorium. There are a number of different reasons why this idea may or may not work. But if you suspect that this topic is not sufficiently explored just because the military liked the plutonium / uranium topic more, then you should at least take another look at the idea of ​​using thorium.

Returning to the past is fun. Images of the future, born in the past - reasonable robots or space travel, which they dreamed of in the 60s - these are cultural symbols of that time. We cannot simply go back in time and copy them to achieve a better future. We must remember that the future from the past did not work. Therefore, today's goal is to look into the past, learn there everything that can be, and apply this knowledge in a new, meaningful way.

Ii. Where the future failed

There are many industries that can be said that the future has failed there. Let's look at four of them: energy storage, weather, robotics, and space. After a short abstract excursion into each of them, we will talk with people who work in companies involved in these industries.

A. Energy Storage

The main problem associated with energy is that the cost of its production is variable. The cost of energy increases significantly during hours of maximum use. But since it is very difficult to save energy produced not during peak hours, acceptable solutions are the most difficult to reach.

The original technology of energy storage and storage was the use of batteries. But there is a reason to think that we are moving towards critical and possibly unchanged limits on the technology of manufacturing batteries. This invention is 200 years old. Of course, throughout this period there have been significant improvements in technology, but now it seems that it is striving for the limit of its capabilities. In a limited space, you can put a finite number of batteries. Batteries rust. Since most batteries contain particles with opposite charges, there are chemical limits. At the moment, the emergence of improved technology for the production of batteries could be something like the discovery of a new chemical element.

Therefore, the future of energy storage technology is unclear, and it is very interesting. The chemical energy repository paradigm worked wonders, but miracles may have ended. Now we are interested in the question, will reflections on completely different ways of storing energy be productive? Are there any other ways to solve the problem besides chemical?


And here LightSail Energy, which is developing a method for more efficient energy storage, deserves attention. Again, this method provokes an incredible debate about whether it can benefit from the difference between the cost of energy during peak hours and the rest of the time. Batteries and hydroelectric technologies are expensive and very limited. The approach of this company is radically different from the generally accepted one and treats energy storage not as a chemical, but as a physical task.


At one level, this is the basic Boyle-Marriott law. You use energy to force air under pressure into steel tanks. Later, when you need energy, you reduce the air pressure to release it. The main problem is that the air is very hot under pressure. Because of this, the amount of energy decreases. As a solution to this problem, it is proposed to inject water into the air in order to cool it. Naturally, there are a lot of complex nuances on which this system works, but in general, this is a simple idea implemented at a high level. LightSail is said to have swung billions of dollars to the market. And we must push them to find out if this is a big opportunity or a highly competitive environment where it’s impossible to defend. But the physical approach to the problem of storing energy instead of chemical is a very original approach, and the technology is also very promising.

B. Weather

People have always been interested in predicting the weather. It's amazing how much we still do not know how to do it. Even short-term forecasts are traditionally inaccurate. There are plenty of excuses for this. We know that forecasting systems are disordered and complex, but there is a feeling that people just gave up and stopped trying. If at the moment weather forecasting looks like a fortune telling, perhaps we can make it less similar to that. Maybe we could be more accurate in our forecasts, and this would undoubtedly be valuable.

There is also the question whether we can control the weather. This is almost no one does, as it is considered that it is dangerous. The scattering of clouds seems to some even an insane idea than the terraformation of Mars. The negative reaction to trying to control the weather or even trying to think about it is strong enough. You can choose your favorite combination of impossible / undesirable / unnecessary to present this reaction.


An interesting company that struggles with this reactionary consensus is The Climate Corporation, formerly known as Weatherbill. The company is trying to improve the situation in the crop loss insurance market by accurately predicting the weather conditions on a particular piece of land, and then making quick corrections based on state-of-the-art computer simulations. This is a statistical approach, so some congenital skepticism is appropriate. Thus, this is a statistical approach to solving problems in an industry where people have developed distrust of statistical approaches, because all of them have failed before. So unusual statistical approaches seem to be underestimated because people put a cross on them. Another secret is hidden even deeper - hardly anyone sees agrotechnology as a sector - because, by definition, this is what happens outside of Silicon Valley, that is, events like this are usually ignored and underestimated.

C. Robotics

In the 1950s and 1960s, we had a ton of ideas on how robots would improve our future. Mostly, of course, people imagined a home cook / driver / housekeeper / servant raising their standard of living by doing all these routine duties for free. Now it is obvious that this did not happen. Some progress has been made in this direction, but robotics has almost reached its limit in recent decades. It is rather expensive to create humanoid robots, and their capabilities are rather limited. The latest development is something like a robot that can fold laundry, which can fold one thing in 45 minutes. If you ask the experts when we should expect the appearance of a robot from “Lost in Space”, they usually say that it’s not in the next 25-50 years. This may well be true, but thinking in perspective is a useful exercise. But the period of 25-50 years is really too long. It's just beyond the responsibility of people issuing such deadlines. Such a prediction is an encrypted message like “It will happen, but I will not have to do anything to do it. Someone else will do it. ”


There are many ways to rethink what robots can or cannot.
RoboteX is a company located in Silicon Valley and is doing just that.

Modern robots fall into one of the four quadrants of this 2 by 2 matrix:



The idea behind RoboteX is that people now have to do a lot of dangerous things. Police assault teams, hazardous materials specialists, sappers teams — all of these people are doing extreme dangerous work. Instead of sending people to an unknown territory, you can first send a robot there to see what is happening there and, possibly, begin to control the situation. The robots of RoboteX, for example, have proven their effectiveness in hostage-taking situations. Bad guys, wound up and tuned to a fight with the police, to some extent receive a psychological shock when they see a small robot, and suddenly give up. It turns out that many organizations are willing to pay a sizeable amount of money in order to avoid the potential harm that their employees may incur. That is why the company's approach to robots as intermediaries, and not as humanoids performing tasks for humans, is valuable enough.

D. Cosmos

Cosmos became the last frontier for Star Trek in 1967. But since then, space has been perceived as a frontier, which is almost closed for us. Space museums seem frozen in time, almost like historical museums. The landing of Apollo 11 on the moon occurred in July 1969, and the landing of Apollo 17 - the last in US history - took place in December 1972. It turns out that the entire period of flights to the moon lasted only 3.5 years. Since then, more than 40 years, no one has been on the moon. Mars now seems a terribly ambitious project, given that our generation has never flown even to the moon.

Cosmos has always been a cult symbol of the future. But in the past few decades, a lot has gone wrong. Costs have increased dramatically. The program to create a space shuttle space shuttle has lost its original significance. The cost has increased, the return has fallen, the risks have been higher than in the project of the Saturn-V rocket. The recent decommissioning of the project was perceived as a departure from the border. Although it may not be so. New interesting perspectives and approaches to solving problems suggest that the new era of space technology is not far off. Can we develop more advanced telemetry software? Or radically new types of missiles? Will new technologies be able to take us back to the future?


For more than half a century, we have launched rockets into space, but the cost of launching 1 kg of cargo into orbit has not changed in the past 40 years. There has always been a lot of different reasons for this. The main reason for this is that most of the financial calculations for the traditional space industry were made on the basis of the principle of “costs plus fixed profit”. Contractors owning an aerospace vehicle billed for a rocket. This amount was laid profitable contractor profit margin. Since people always have a lot of incentives to increase costs, the cost is constantly growing. There is also an argument that the launch has become more dangerous.

An important question arose: was it possible at all to reconsider the approach to the problem from a radically different point of view? This is exactly what SpaceX is doing: it focuses on properly structuring costs, reducing the launch cost, revising the main project documentation and introducing new technologies and materials into it. Judging by the main criteria, their radical improvements were quite successful. SpaceX, perhaps, opens for us a new era of space flight.

E. The goal is to return to the future.

The idea of ​​returning to the future is this: you need to figure out what we were wrong in the past, learn the lessons and make sure that this time everything worked. But what is important is that you need to resist the movement by inertia, which imposes on us a cult vision of the future from the past. It makes no sense to repeat the same mistakes again. The key point is to approach the solution of old problems from a new point of view.

Iii. Prospects - a conversation with representatives of the companies LightSail Energy, The Climate Corporation, RoboteX and SpaceX

Peter Thiel : Your companies demonstrate different ways of thinking about the future. In this lesson, we talked about the fact that radical uniqueness can be a very good business idea. But does not such an original approach complicate the search and recruitment of employees?

Scott Nolan : SpaceX made me a job offer when I was working on one of the aerospace programs. In fact, everything was simple: “You can work for Boeing or Lockheed. Well, or come to us in SpaceX. ”The well-established players in this market worked according to the“ cost plus fixed profit ”model. I had a feeling that there would be no full-fledged opportunity to make some significant contribution to the development of the project. Solving problems there was basically just adding more people to the project. But young engineers seek responsibility. So you look at a startup type and think, whoa, I will have 30-50 cool guys on the project, and I will be responsible for this project. I really can change something.

So I think that SpaceX companies just need to attract young engineers who are set up to work in a startup. It was more difficult to attract more experienced veterans of the aerospace industry. Everything becomes much more interesting when people from this industry earned and earn more than hundreds of thousands of dollars a year. To exchange a good salary for a share in a startup is a big risk for people who are not accustomed to it.

Daniel Fong : We had no big problems with recruiting at LightSail. Silicon Valley has many talented electronics and hardware engineers. All members of our founding team joined us quite easily. We started the project in the garage, in the repair shop. However, we could not just take and get us to join us experienced in this industry professionals, in fact, not so many people specializing in reciprocating air compressors. Therefore, we had to thoroughly search. We reviewed over 1000 resumes and conducted about 400 interviews. Many of our top employees came to us from the auto racing industry. These are people compatible with the startup model. They understand what hard work is, small teams and deadlines.

If your car is not ready to start the race, you have already lost. The rules in motor sports are constantly changing, so people are accustomed to the annual redesign of engines. Another challenge was to study how the racing industry works - all this in order to hire people who work in it. But in the end, we succeeded by asking these incredibly talented people if they really wanted to drive cars in circles for the rest of their days.

Greg Smirin : This approach can be very convincing. Our business requires many analysts and experienced programmers. But Facebook and Google and all such advertising optimization companies absorb a lot of talented people. So we created a rather provocative statement: solving problems for predicting the weather is more important than helping middle-aged women send each other virtual pigs. The gaming industry is really very cool. But a lot of people are not content with developing games, but want to do something more meaningful.

John Hollander : The creators of RoboteX, Adam and Nathan Götingi, began by assembling robots in their garage. Adam was in charge of the mechanics, and Nathan was in charge of the software. These modest first steps were important. Many make the mistake of trying to do something very advanced right away. It is much more important to start with simple mechanisms that work according to the laws of physics. A robot that cannot climb stairs or move around a territory is useless, no matter how perfect its computer or cooling system is. Many of our competitors look very cool, but they have problems with basic things - the caterpillars fall off as they go.

As soon as we had the first prototype, it was not so difficult to recruit engineers. Robotics are very attractive. Our main slogan is convincing: we create robots that save lives. Customers applaud us every day. Right today, one of our robots concluded some of those talks about hostages you mentioned. The man who put the knife to the throat of his wife was forced to surrender in peace. We are developing humanoid robots or C-3PO . We create effective and inexpensive robots for the police or for work at the power station - for those places where there is a danger to humans or some tactical needs.

Peter Thiel : Let's talk about the essence of your core technology, because at least on a superficial level, you are all trying to solve problems that have been around for a long time. Returning to the future involves 2 questions: What went wrong in the past? Why are you sure you can solve these problems now?

Daniel Fong : It turned out that for a long time the idea of ​​using air as a storage of energy was literally “in the air”. It was very popular in the 1870s, about 10 years before the power grid. And throughout this “golden age of compressed air,” people tried to increase efficiency by squirting water into the air! They knew that the heat capacity of the water is very high. But this technology is completely forgotten. All that we were able to dig out about this is that "problems have arisen." We know nothing more. Perhaps they did not have the right materials or they were faced with the problem of corrosion. It is very difficult to eliminate defects in the conclusions of inventors who have long gone into another world. But what is truly amazing is that if everything was done then in the correct sequence — if they correctly used the principle of air compression — the history of technology would have developed in a completely different way. The relationship with the history of energy development would be very powerful.

Greg Smirin : People have long been interested in predicting the weather. The concept of insurance in the agricultural sector is also not new. It seems that in one form or another this problem has existed since biblical times. People diligently tried to calculate the chances and prescribe the strategy of behavior. There are many newspaper articles on crop insurance in the late 1800s - early 1900s.

But then people did not have computing power or the necessary data for high-quality processing. There was only fragmentary one-sided information. Today we have the computing power and the necessary data. We have a thousand processors that quickly collect and process data from remote sensors that perform detailed measurements at various points on a particular piece of land. One of the challenges is how to reduce all these data so that they are clear and accessible to the farmers themselves - in a sense, this is the opposite of the problem that people faced 100 years ago.

John Hollander : One of the reasons why robots did not appear even earlier in the past is that no one focused on mechanics. Science fiction very well fascinated people with robots, but also created some prejudices. People see humanoid robots on the screen and begin to create robots that have legs. This makes no sense. Crawler travel is much better. But more importantly, the very high cost of production did not allow the industry to grow. Cheap robots would be ubiquitous, like laptops now. But only a few companies can create them. To reduce costs, our company takes advantage of the computer industry. There is already a huge and efficient Asian infrastructure created for the production of computer components. Therefore, we benefit from it and supply our robots with ready-made computer components. There are other innovations that we use. For example, instead of cutting components from aluminum blocks, we use durable plastic to make the same part for 1% of the cost of the aluminum equivalent.

Scott Nolan : Since the launch cost did not decrease before entering SpaceX, it was clear that there was no progress in rocket engineering yet. The reason why a startup like SpaceX could succeed (and hopefully continue to dominate) is that we restructured the development process. Large, well-established players developed projects according to the specifications that were already available and gave some processes to outsourcing. During the design process, there was a lot of friction, and the maintenance of the high cost of the project was strongly encouraged. SpaceX has changed everything. The company completely destroyed the industry by developing projects from scratch, combining functionally interconnected departments within the company and the culture of Silicon Valley. We brought with us a new engine design, a new design and technology approach and a new on-board electronics. We have developed the concept of using composite materials to the limit. Lighter constructive elements and new welding technologies were used. Few forgotten fuel injectors were revived again. The number of innovations or radically rethought things is simply amazing.

John Hollander : Despite the fact that RoboteX has started a horizontal cost reduction, it is currently following the path of vertical integration, uniting functionally interconnected units. Now we have our own molds, models of chips and radio cards and much more. It is possible to slightly reduce the cost of production, if you yourself are developing some things that perform strictly defined functions and nothing else. No need to pay any margin.

Daniel Fong : In this respect, SpaceX was an incredibly brave company. Typically, the cost structure has little to do with the real cost of components, and more depends on how much people are used to taking money for components. The shake-up of the entire industry, which operates on the principle of "cost plus", of course, caused someone to be very annoyed. But it was a great opportunity. One of the techniques that I learned is simply to tell any suppliers who roll out the monstrous price tag that I can just do this part myself. Then they very quickly switch to the negotiation mode and reduce the price. Then you take this price and call it to another supplier, and in general you provide the suppliers with each other to fight for the delivery. The goal is to achieve efficiency where you do not have vertical integration and where you have it.

Audience question : What is your experience hiring employees or collaborating with academia?

Greg Smirin : We have about 20 employees with doctoral degrees at Climate Corporation, and for many of them this is their first experience outside the scientific community. Of course, there is a transitional period. We must help people learn to use their talent in such a way as to really benefit the organization in which they work, and not just apply for research grants or similar programs. Before we take someone to ourselves, there are many preparatory procedures on both sides. Everything is looking for a suitable person who would successfully adapt to the company. Of course, not everyone will do. But those who fit into our environment turn out to be incredibly energetic enthusiasts. The fact that they can contribute and influence the development of a product in their field brings them immense satisfaction and motivation.

Daniel Fong : We are constantly working with professors on many research topics. Their input can be very valuable in terms of changing the course of the discussion and selecting new ideas that we are considering. Professors are also fairly honest with us when we check the recommendations and ask them how good the candidates are.
Question from the audience: How do you solve the distribution problem if you need to provide supplies to the government or very large enterprises?

John Hollander : In general, there are two main approaches to government supplies. The traditional way is to go into the military industry and enter into transactions or conduct research in this area. The unconventional path that RoboteX has taken is to start moving in the direction of small, narrowly targeted institutions, like local police stations, and then grow upwards. It was important for us to remain a completely private company. If you accept government funding, then often this will lead to serious restrictions. By selling our products to a multitude of small institutions, we retain a sense of freedom, get feedback from enthusiasts, and can rebuild a good flow of income. And, of course, there are plans for expansion, and now we take on any orders of large commercial corporations, which have some life-threatening tasks.

Peter Thiel : The basic distribution rule says that the higher the cost of a product, the slower the process. The classic mistake people make is to satisfy a fantasy that you can just get a single contract for $ 100 million, and everything will be in chocolate. In practice, it almost never works. Theoretically, for a large nuclear power plant, it makes sense to introduce the use of robots. RoboteX robots would be very helpful during the Fukushima accident. But basically, you can't just go and sell robots to a theoretically ideal buyer. There are a lot of different processes. When thinking about sales and distribution of products, you must remember that people often do not know what they want. There is never a complete objectivity. And even if your technology is a whole order of magnitude better than others, you can’t just take and make a gold sale. The person who will write you a check for $ 100 million will ask who else was buying your product. And if the answer is no one, then the check may remain unsigned, because its owner may be suspicious that something is wrong with the product.

The most successful startups usually use a phased sales model. They reach 50-100% growth gradually over several years. They can reach a sales level of $ 5 million in the first year, and then, if things go well, double the result every subsequent year for 10 years. You may wonder why the income simply does not increase 10 times in the fourth year, when everyone already understands the benefits of the product. But it does not work like that. It usually turns out that not a single buyer wants to make a deal that is 10 times higher than the sum of the largest deal that you have so far concluded. Perhaps double the maximum amount of the transaction - it is more realistic hope.

There is also an investor view of things. An investor wants to find one really rich limited partnership that will invest its money in an investment fund. Find such a gold investor - and no longer have to talk to anyone. But it never works. Rich companies, like everyone else, are grouped together. Everyone wants to act as if they know what they are doing. But in reality, no one has an understanding. Everyone succumbs to someone else’s influence in some hidden, unspoken way.
Therefore, the strategy should be such: to get a small client who can give a good review about you. Move fast and gain good reviews. RoboteX sells robots to local police stations. The sales themselves are not as large-scale, but the company can use the success of any individual implementation. If the robot is good enough to prevent the killing of hostages, then perhaps it is suitable for other purposes.

Daniel Fong : The only exceptions are cases when someone just desperately needs what you offer.

Peter Thiel : Yes. Always start with such customers.

Daniel Fong : People who desperately need something, in conditions where other suppliers do not fulfill the conditions or simply disappear, often become excellent customers.

Greg Smirin : Many sales models fail because people miscalculate the sales cycle. They underestimate how long it takes to get the perfect customer. Therefore, a good recipe for success is to find very small and best customers, and quickly. Take the low barriers first. Then you will know more about how to handle larger customers.

Scott Nolan : And you can appeal to the sense of urgency that people have. SpaceX approaches clients with such a question: do you want to launch your satellite in 4 years? Or maybe you want to launch it at a much lower price and fly next year?

Audience question : What do you think of a long-term care strategy? Are you planning to take over your company or transform it into a public company?

Daniel Fong : In theory, we would consider the absorption option. The question is whether the company absorbing us would not hurt what we are doing. This is quite likely because we are talking about a unique approach to energy storage. But no one knows. , , - .

: , , . . . , , , , .

: , , . , - . - , , – , . : , . , . : , . – , .

. – . , , . , . – , . . , , . , PayPal eBay . , . , . , . , , .

Audience question : Can you comment on the process of raising funds for conservative technology companies?

Daniel Fong : It's damn hard!

Scott Nolan : Elon had to invest $ 100 million from his own funds in SpaceX. I doubt a little that he would spend everything he has, if necessary. This did not happen. But this example shows that the founders of such companies are ready or should be ready to go all the way. They do not rely on investors to achieve their goals.

Peter Thiel: NASA to some extent demanded that SpaceX accept third-party funding in 2008. Then the financing went through the Founders Fund. Various Silicon Valley investment companies have expressed concerns about this. They warned us that investing in SpaceX was a risky and perhaps even crazy step. And it was not even at a very early stage, but after the company built rockets and made several launch attempts.

Daniel Fong : , - . , . , , - . , .

: . . . , , , .

, , . , - . , , : « ?» : «». , SpaceX. , , . : « ?», «». 40 .

, iPhone . , , . , , .

, - , , , , , , . . , , . , , , .

Note :
. astropilot , .