Monopoly by Peter Thiel


繁體版本請點我

Introduction

The monopoly with the last mover and not first mover advantage often turns out to be the really valuable one. For example Google is the last search engine, Facebook turns out to be the last social networking company, PayPal – the last payments company, which are all very valuable. Even if you are the first in some sense, but if you get superseded by somebody else very quickly, your startup is often of much lesser value. When we think about this question of monopoly, there’s something I want you to always keep at the back of your mind: there’s always a time dimension. It’s not just about having a big monopoly and a lot market power, there’s a question of what period of time that gets maintained. If you have an enormous monopoly but it lasts for only six months or twelve months and then it’s gone, that is much less valuable than if it lasts for 10 or 20 years or even longer. Thus, as we talk about these different aspects of monopoly and how it works, I want you to keep this time dimension in mind.

What I thought we would do today is first to talk about 1) the different kinds of monopoly or different elements one can have in monopolies, and 2) I want to spend some time talking about this question of durability – the time dimension. And 3), I want to take a little bit of a bigger picture view and talk about how this question of monopoly and innovation have fit together in the last 200 and 250 years of business. Parts 1 and 2 today will be focused on the specifics of the companies, and then Part 3 will be more about the bigger picture of the history of the world, of how monopoly has worked in the history of innovation and technology over the last 200 or so years.

Part 1: Elements of Monopoly

There are many ways you can come up with a list of the different elements of monopoly. I’ve given a whole series here plus the other ones that one could add. I always think this is a very good representative list. We’ll talk a bit more about this list and enumerate some of them:

I – Proprietary Technology

Certainly a natural element in the classic zero to one companies is technology. It’s where you have a technology that’s better than anybody else in the world, and normally the relevant context of technology is global. So it’s not good enough if you have a better technology in San Francisco that is better than say the people in Texas. It has to be better than anywhere in the whole world. And even if your technology is the best in the world, it’s often not good enough if it’s just a little bit better. We are often presented with companies and ideas that are slightly better, often in the hard sciences where it’s very competitive. A researcher might have done something just a little bit better than everybody else in the world. However, I think what you want is technology that’s meaningfully better. The formula for the rule of thumb that I gave in Zero to Oneis that the technology has to be ten times better on some dimension.[1]As to what kind of dimension that is, for example once Amazon has ten times as many book titles as the next biggest bookstore in the world, it differentiates itself and gets people’s attention. Conversely, one of the challenges in clean technology in the United States from the business side is that even though people are innovating. Often, the very best technologies are only 3% or 5% better than the next best one. Minute differences could be very important for science because sometimes science only progresses in a very gradual way. However, for a business if you have something that’s 5% better than anybody else, that’s often not quite enough to garner attention. Therefore, it has to be somehow dramatically better in some particular dimension. For instance, if you have a disease that has no cure at all and someone discovers a cure, which is infinitely and dramatically better. Biotech is a classic example of something where you have this binary effect.

II – Network Effects

The second area is network effects which people talk about a great deal. The basic intuition with network effects is that it’s the value of a service which grows in some ways as more people are using it. Sometimes it’s described as Metcalfe’s law or a quadratic function where as you have more users you have a square number of possible users that have a connection between them:

There are all sorts of things with network effects where as more people use a technology, product, or service, the more valuable it becomes. Network effects are extremely valuable. It’s one that I’m always a little bit skeptical when people pitch it as a startup because it’s often juggled up in some of yesterday’s buzzwords. Network effects is like a marketplace or platform, and it often means the same thing – a buzzword that people use too often since everybody knows a network effect is great, everybody knows a marketplace is great, everybody knows a platform is great. Thus, there’s a temptation to try to incorporate it into a pitch. However, the challenge and tricky question you face if you are only getting started is why something has a network effect if you only have a few people on it. While they are extremely valuable at scale, people often obscure the very important question of how you get there. Sometimes when people talk about network effects, it’s not just a statement about a great business, but it’s often a statement about their own wishful thinking and how they don’t actually have a large network yet.

III – Scale

A third category of a monopoly involves natural monopolies and economies of scale. The larger the scale, the lower the marginal costs per customer and the lower the average costs end up becoming. If you have sufficiently large economies of scale, where you have just need one company for a given market, then that company ends up in a natural monopoly. Hence, if you can get to economies of scale, that’s very valuable. The challenge with economies of scale for startups tends to be that you have to get very big. Like network effects, economies of scale are very obvious but quite hard. We’ll talk about this a little bit more. There’s something about China where people are very focused on economies of scale even though paradoxically, China is one of the places where it seems to be getting harder because it’s so big. But if you’re living in Sweden with 8-9 million people, you can get to an economy of scale that’s relevant on the scale of Sweden, perhaps much more easily than on the scale of China. But certainly if you get to a monopoly on scale in a country as big as China or the United States, the economies of scale are incredibly valuable. IV – Distribution

The fourth dimension is what I describe as distribution. This is always a big one in sales, marketing, viral marketing, in all these different channels that get the product out. I often said that in every technology business, there’s always a combination of two things: the combination of a technology and the selling of the technology. You can often think of it as a very hard technical engineering oriented piece where the people that are building the technology, and there’s often another piece where people that are trying to figure out how to distribute the technology and get it out. Certainly, historically in Silicon Valley, you often had this pattern where you had two co-founders for a business. One person is more technical and one was more on the business side. The reason you had this division of labor was that you actually had to always solve these two very difficult questions together: getting a new technology that works by partnering with really good engineers. And secondly, convincing people to use it, which often requires a very good salesperson.

Now there are a lot of permutations on this, but the distribution problem is always a big one. I think people are aware in Silicon Valley, that they need and admire technology, network effects, and economies of scales. They might underestimate how hard they are but they are aware of them. Distribution is the one people always underestimate. And there’s always this bias, especially from people that are more technically trained, that if you have something that’s good it will distribute itself automatically. On the same note, it you are a great scientist, you will be rewarded by the natural goodness of the universe. You don’t need to tell anybody about anything, and you just tell people a tiny bit, and the entire world will line up at your door step. Sometimes that can happen, but that’s not actually a common pattern, and very often there’s a very tricky question about how to get things out. And so at certain times the distribution channels themselves can be a source of monopoly. If you have a distribution channel or if you have a way of solving distribution problems that is much better than anybody else, then even if the product is fairly simple or not complicated as an idea, it can be an extremely valuable company because once you distribute it, it becomes very hard for anybody to copy. It’s not that they can’t copy the product or the technology, but they can’t copy the distribution channel. While this can be a big challenge, it can become a big source of monopoly and comparative advantage if you can solve it.

V – Brand

The fifth element of monopoly is something that can be incredibly understated or overstated – is the brand. The classic example in the United States is Coke versus Pepsi. I think they’ve done these tests of asking people to blindly taste them. It’s almost impossible to tell which is which, so there’s no big difference. But people have been brainwashed into thinking that “Oh! I like Pepsi” or “I like Coke”. Even though they’re both sugar water with some carbonation in it, people are willing to pay a lot more for one or another because people have convinced themselves that they like one more than the other. So there is this incredibly ephemeral kind of thing known as brand where if you can get it, it is extremely valuable. There are entire industries where there is nothing but brand going on like in the fashion industry, which is very noticeable for this. Even in very high end fashions, the underlying quality of the products are not necessarily that high or not that expensive to make. Almost 80% or more of the price that consumers pay for is this very intangible thing called brand. It’s a very serious thing how it gets created and generated. I will be a little bit modest here, I always like to tell people I don’t know how to create a brand. Certainly if someone has a business with no technology, no network effects, no economies of scale and no distribution yet say they will create a great brand, I will passively always say that I don’t understand brand. But more precisely, I think it’s very hard to start with brand. If you’re starting a business maybe you can build a brand over time where you leverage one of these other things, and sometimes people can create brands from scratch, such as new fashion companies. But it is a very mysterious yet important process of how this happens.

VI – Complex Coordination

A sixth one that I didn’t talk about that much in the Zero to One book but I allude to in different contexts is what I describe as complex coordination.[1]This is where what you’re actually doing is not creating specific new technology, network effect, scale or distribution, but taking a lot of pieces that already exist and combining them in a specific way. And combining them in just the right way creates a new product and it turns out that getting this sort of complex coordination to work is often harder than it sounds. There are a lot of people who are not trying to do anything like this so if you are able to pull it off, you’ll often have some very significant barriers to entry. We’ll go through some examples of this, but I think complex coordination is one of the forms of monopoly that’s the most underrated in Silicon Valley. I think people understand technology, network effects, economies of scales, distribution engineers underestimate but understand generally and brand is something people ‘understand’ too well. Complex coordination is actually not even a category people think of, but one I always pay a lot of attention to. As an investor, the contrarian question for me is which companies have value that is underestimated by other investors. When I find companies that require a certain amount of complex coordination, I often think that these are things that people are undervaluing to some extent.

VII – Government

And there’s this whole government layer one can add on top of this in all sorts of different ways where government can help people create monopolies or prevent them from doing so, create legal monopolies in the form of patents or intellectual property,

etc. And there are all kinds of ways that all these things intersect with the government, which will be discussed later. But these are the elements, I don’t think you need all seven of them. My rule of thumb is if you have one out of seven, that’s enough. It’s not like an academic class where one out of seven is not that good. In business if you get one out of seven, that’s like an A+. Most of the time people get zero out of seven. So it’s like a very hard math exam where the median score is zero, so one out of seven is a very good score. And if you get just one out of these seven, then you have a pricing power and a valuable company. Next, let me say more about these categories and illustrate them with some concrete examples.


Part 2: Deep Discussion

I – Proprietary Technology

Technology is always the question: Can one have a real difference in technology? I think when I started PayPal, our initial idea was just a disturbingly simple idea to link money with email. Nobody had thought of it, it was a new idea, but we were worried that it would be copied very quickly, and at the end of the day, it would not be that valuable. One of the things that turned out to be unfortunate but also fortunate was that if you just linked money with email, that would not be enough because there were a lot of people on the Internet who were doing criminal things, stealing your money, and you needed to find ways to detect fraud. The non-technological solution to fraud is that you find somewhat cumbersome and difficult ways to know a lot about your user. So if we thought that Blake here was a fraudulent person, we could ask ask him to send us a picture, ask him to physically show up at our office and get his fingerprints. You can always prevent fraud if you are willing to put enough work. The critical thing is to somehow make a product that is somehow still very easy to use. The challenge was to detect fraud so it was not a burden on anybody who was a good customer. There were some algorithms and set of technologies developed at PayPal to detect fraud that were not cumbersome to the users, and this turned out to be a very big advantage against the startups we competed against who got wiped out by fraudsters. As for the banks we competed against, they were very worried about fraud, but they had such clumsy means of stopping it that they ended up with no customers. So either they had very heavy-handed ways to stop fraud that made their product too hard to use, or they had nothing and the company got wiped out. There was a somewhat humorous way to describe PayPal in 2001 and 2002: we had a very strange relationship with these criminal organizations some of which were based in Russia and other countries. Even though they were stealing money from us, we were developing this anti-fraud technology so they had to get better and better at stealing money, and they were using their improved abilities on all our competitors. So we were sort of training them to wipe out our competitors, and we had this strange symbiotic relationship with the gangsters. And so, over time this technology developed into a significant barrier that gave us a real advantage.


Now of course, a lot of people have heard about Google but in the late 90s, they were just another search engine. The critical thing is that they had this PageRank algorithm where they could organize far more web pages far more efficiently and there were ways that technology enabled them to get to scale. The technology for PayPal was quite valuable for a long time. On the other hand eventually other people learned the PageRank algorithm from Google so it was not a permanent advantage, but it was an advantage that at least gave them time by which they could build out network effects, economies of scale and brand. Often these things had a time dimension where the technology gives you head start, and then you have to turn it into one of these other things which Google succeeded in doing.


II – Network Effects

If we go from technology to network effects, certainly you have a lot of the classic consumer internet businesses such as Facebook and Linkedin which rely on network effects. One of the things that’s characteristic of a lot of network effect businesses is that there’s a very big challenge in how you get these businesses started on day one. Facebook was able to do it by starting at very small colleges. Yesterday I said you have to start small and scale, and if you want to build network effects, you have to start very small because you have to get them to work even when you have a small amount of people. So, somehow Facebook could get everyone in the college to adopt it, and they got enough of a network effect with even a few thousand to 10,000 people that it was valuable. LinkedIn had a somewhat stranger pattern where initially it was about putting your resume on the Internet, which was valuable to you even if you were the only user using LinkedIn. But then it scaled and turned out to have more network-like effects. A lot of these businesses begin with a bootstrapped quality, then scales into a network effect, but you have to start with it being somehow valuable to users from day one.

Payment businesses have always had network effects. The big challenge of a payment system is that it’s only valuable if other people are using it. For example the credit card companies got started in the United States such as MasterCard, it was very difficult because you need both the customers with credit cards and the merchants with credit card machines. This was a chicken and egg problem. How do people get the credit cards? How do the merchants get the machines? Visa, I believe started in Fresno, this town in central California, and it took them almost ten years of inundating the town, giving out credit card machines to the merchants and handing out credit cards to people. Eventually everyone got used to it, and we gradually got this new payment network effect. So the network effects of payment systems are incredibly powerful, yet also incredibly hard to get them started.

When I started PayPal in 1998, our very ambitious goal was to create a new currency / payment system on the Internet. The Internet already took off in 1993 and 1994. In those four-five years, there had been a whole generation of companies that had tried to start new payment systems on the Internet and they all had this network effect story. The problem was they all started too big and shrank. They had a story about how they were going to create a small amount of value to a billion people, through digicash and cybercash, all sorts of companies like this. But the critical thing was even though you’re trying to get to the network effects, how do you get to something that’s valuable to the very first person? This was the single question I spent so much time thinking about at PayPal as we were getting started. Yes we want to get to a network effect, but we couldn’t get there on day one. So how could it be valuable to the first person, or the 10th person, or the 100th person? Once you have a million people it’s valuable, but there’s no path to a million people if there’s no value to the 1st or the 10th or the 100th person. The idea we ended up coming up with was linking payments with email, and we already had the network effect of email because in 1999 there were already 300 million people with email addresses. We enabled users to send payments to anyone with an email address, and even though they were not yet part of our payment system, they would be invited to join. It had a one-way, open-ended network effect. You didn’t need to get both sides onboard.

There was a similar history with Netscape four years earlier. One of the legendary companies in Silicon Valley was called Xanadu, which was started in 1960. It had this idea of creating something like the Internet years ahead of its time to connect all these computers, and it would only work once everybody is connected both ways. So you had to get both the client and server connected all at once. It was a cult-like company that kept on getting funding for 29 years, never made any money because they were trying to create this network effect that would only work if the whole network effect was switched on, on the same day. In 1992 they finally ran out of money. The next year, Marc Andreessen’s Mosaic and Netscape Browsers took off, and the way that worked is that it was one way and one direction, just the client - server that you responded to. As more people got onboard, you could get onboard. You didn’t have to get everyone in the world to join for it to work. Therefore, the one way network effect can be a very powerful way of bootstrapping. One funny story was in late 98 and early 99 as I was starting PayPal and I was thinking about the network effect and talking to many people about it. I was talking to some of the people who had stopped working at Xanadu after 20 or more years. After Netscape took off they decided to start an online payment system. The company they were working on was called Digicash, and it was just like Xanadu. It was both ways and it was going to work incredibly well once everyone in the world joined. I realized that they were making the exact same mistake as the Xanadu people had made, and we had to figure out a way to make it one way just like Netscape. So Paypal was to Digicash like Netscape was to Xanadu. It was an ironic history where it was literally the same people who were too early both times, making the exact same mistake of thinking that they could get to a network effect on day one, instead of thinking really hard about how to build it up organically. So, network effects are very important, but you have to somehow start very small or start with something that’s even different from a network effect altogether that somehow gets transformed into that over time.


III - Scale

There’s a lot one can say about scale monopolies and natural monopolies: basically the larger the size, the cheaper the cost. The big examples on the Internet tend to be Ecommerce sites and Amazon is probably the one that has emerged in the US as this relentless scale monopoly. It’s a very strange business. Even today, it’s been over 20 years since Amazon started in 1995, and it’s still barely making money. You scale it, and every dollar gets reinvested into scaling it even more. The mathematical theory behind Amazon is that they will make a tremendous amount of money at some point in the future, but you have to have to get it to a bigger and bigger scale and wipe everybody else out first, and then you get to an enormous scale which is enormously profitable in the very long run. I think scale is very important and interesting in a lot of ways. For example in China, they have this strange attribute that because it’s better to be bigger than smaller, then your incentive is to put all their effort into scaling even more. If the key monopoly featured in your business is scale then that becomes an argument for not making profits initially, and so it becomes a big mistake to make profits too early because if you’re trying to make profits too early, you’re not investing in scaling even faster. One of the ways I described this at PayPal when we were losing a lot of money early on is, we had to take over the world, and once we took over the world, we could make money. If we first tried to make money and prove that the business model worked, then we lose time and someone else would scale more quickly. So we couldn’t make money and then take over the world, we could only take over the world and then make money. This sounded very crazy, but I think there is an element of truth to it with scale, especially in the Chinese context.

There’s a strange history where Ebay had this big advantage in China, but then Alibaba was able to beat Ebay. One of the reasons was Ebay started charging money way too early and Alibaba kept everything free, was able to scale more and get the economies of scale that Alibaba used to beat Ebay in late 2004 and early 2005. I think this is understood much better inside China, where the general view was very obvious that what Alibaba did was right and that Ebay was an incredibly shortsighted company. So in China this is very clear. Outside of China, it’s generally not quite as clear but there are many instances where premature profitability has been a big mistake. MySpace tried to become prematurely profitable, and it made money in 2006. By 2007, the growth rate was negative, whereas Facebook did less on advertising, kept reinvesting and scaling even faster which was the right decision. Although the way it was talked about outside of China, it was much less clear-cut. In the US, many people thought that Facebook wasn’t profitable and wasn’t making enough money, and MySpace was considered the more responsible company at least for a short period of time. So I think this scale question is somehow appreciated more in China than outside of China even though it’s important in many other places.


IV – Distribution

What was critical to Snapchat and Whatsapp was a viral form of distribution where just using the product gets other people to use it. This is where you have these very thin products which do not have much technology or anything else but gets distributed so quickly that nobody can ever catch up. On our website because we are trying to push hard and breakthrough technologies, we have this tagline: they promised us flying cars, and all we got is only 140 characters. This is sort of a little negative statement about Twitter. It’s because Twitter is this a very thin product which doesn’t have much technology and we can debate about how much it is changing our society or civilization. The critique that we have of Twitter as a technology is that it’s not as exciting as a flying car or something like that. It is, nevertheless, a very good business because once they figured out how to get it to work, it grew so quickly that nobody could catch up. Thus, if you have a distribution plan which is exponential, and you have something that is growing exponentially, even if someone else copies everything you do, you then just generate a second exponential curve. If there’s a time delay, it’s just a diverging curve. Let’s say Twitter grew from 100,000 to a million people from month 3 to month 6, and then a million to ten million from month 6 to month 9. If you took 3 months to copy Twitter and at month 6 you have the exact clone of Twitter you have 100,000 followers but Twitter has a million, you’re 900,000 behind and just a diverging function. At month 9 you’re at one million, and Twitter has ten million, so you’re nine million behind. So the exponential curves are fast enough even when you are able to copy something fairly quickly, copycats end up following further and further behind. If you have very fast and powerful distribution channels, these can in and of themselves become very powerful barriers to entry and make a company much more valuable.

Of course there are all sorts of slightly different channels of one sort of another and not just viral distribution. One of the companies I’m invested in in Silicon Valley is called Stripe, which is a new payment processing solution. The big thing that they figured out is that when new startup companies accept payments, the payment solution is often decided on by the web designers or the people at the company who built the web page. Therefore the payment companies need to be optimized for the people who are designing the web page and be very friendly to the web designers as they will adopt your product. Historically, you have a payment solutions and you need to talk to the vice president of the company with a salesperson. Stripe figured out that no we need a product that is designed for the people designing the web page. Hence there was a different distribution channel and figuring it out was a powerful comparative advantage.

I think distribution, as I already said, people with an engineering or science background always tend to underestimate. They tend to always think if you have something better, the product sells itself, which is often untrue. One way I often describe this is that if you’re a good engineer, you always have an incentive to exaggerate how hard engineering with many different gradations of difficulty is because you want to differentiate yourself from other people. Whereas if you’re a salesperson who works on distribution, you never want to tell people how good you are at sales. You always want to leave it very understated because if it requires tremendous sales to sell a product, that must be a sign that the product is not good. In the United States, the people one thinks of as the archetypes/stereotypes of the salesperson is a bad salesperson. This is a used car dealer with a silly tie with polka dots on it who is very extroverted. Conversely with a good salesperson, you don’t even notice that he/she is a salesperson. In many of these companies, the good salespeople often have titles other than sales, you might call them corporate development, business development and the CEO is really just the salesperson. But you never call the CEO the chief salesperson although there are certainly technology companies where this is certainly true. You always have this division of labor in the early Apple between Steve Wozniak who is a hardcore engineer and Steve Jobs who was the CEO and the marketing salesperson. Jobs would never describe himself as the chief salesperson, you would only do that if you were bad. There’s a tendency for us to overestimate engineering challenges because the engineers have the incentive to exaggerate how hard it is because they get rewarded the harder it is, whereas the salesperson has an incentive to understate and to hide it, “I know it sort of sells itself and I do not have anything to do it, but if I am not there then it does not get sold”. So when you look at it from the outside, one tends to underestimate distribution as a problem. Anything that has to do with bigger distribution is quite critical.

V – Brand

I’m going talk a little bit about this very mysterious thing called brand. The archetypical brand company is Apple, an iconic brand that people have learned to recognize all over the world. However, I want to suggest that the starting point for Apple was not just creating the brand - it started from technology or economies of scale or getting distribution channels, and then the brand was gradually built around these. If you try to copy Steve Jobs, the temptation is just to copy the branding side, by wearing a black sweater and jeans. There is something about their branding that’s very good, but that is not quite the place to start. One of the ways to describe the mystery of branding in the words of my friend is, “everybody wants what everyone wants”, which is an ambiguous statement. On one level, it’s like a tautology in mathematics, it’s a circular argument and it doesn’t say anything at all. In a static world this statement is completely meaningless as nothing changes. But it can also describe a very powerful dynamic process where you want things that other people want and there’s a dynamic flow where everybody is copying everybody else, and everybody wants what everybody else wants which has a snowball and escalating dynamic. If I had to say there’s a key to advertising, it’s to create brand around this idea of everybody wants what everybody wants. On one level it is circular, and on the other level is this very powerful dynamic. It’s extremely easy to say, but extremely hard to do. As a venture capitalist, once every few years I have to try to raise money from LPs for my venture capital fund. Two and a half years ago, I was going through this process of a roadshow where I meet 5 or 6 prospective investors every day for 2 to 3 weeks. At every single meeting, the investors would ask all these different questions. One day I thought to myself, there’s only one question that everybody really has and that nobody ever asks: Are other people investing in you? Nobody would ask that question because people were too embarrassed – they want to say that they are thinking for themselves and are not influenced by anybody else. But the one question at the back of their minds is always this question: What does everyone want? Is this what everyone wants? If you could convey that, it is very powerful. If you actually told people that everybody wants to invest, and you probably are not going to get in. That doesn’t work because then they would think, why are you even talking to me if everybody wants to invest in your business? There are all these subtle ways people try to convey it, and they tend to backfire because if you do it the wrong way, it does not show scarcity but instead shows your desperation. So branding and advertising can be very powerful, and can backfire in all sorts of ways. My advice is not to underestimate it, but also not to focus on it. Be aware of it, when you get it to work it is super powerful, but don’t start with brand.

[1] This lecture is Peter’s most extensive discussion on monopolistic elements with new categories and examples. Cf, Zero to One chapter 5 and http://blakemasters.com/post/21169325300/peter-thiels-cs183-startup-class-4-notes-essay

[2] https://genius.com/Peter-thiel-lecture-5-business-strategy-and-monopoly-theory-annotated

#monopoly #startup #economiesofscale #brand #technology #distribution #networkeffects #PeterThiel