pgstrata
The Future of Web Startups
2

October 2007

3

(This essay is derived from a keynote at FOWA in October 2007.)

4

There's something interesting happening right now.

5

Startups are undergoing the same transformation that technology does when it becomes cheaper.

6

It's a pattern we see over and over in technology.

7

Initially there's some device that's very expensive and made in small quantities.

8

Then someone discovers how to make them cheaply; many more get built; and as a result they can be used in new ways.

9

Computers are a familiar example.

10

When I was a kid, computers were big, expensive machines built one at a time.

11

Now they're a commodity.

12

Now we can stick computers in everything.

13

This pattern is very old.

14

Most of the turning points in economic history are instances of it.

15

It happened to steel in the 1850s, and to power in the 1780s.

16

It happened to cloth manufacture in the thirteenth century, generating the wealth that later brought about the Renaissance.

17

Agriculture itself was an instance of this pattern.

18

Now as well as being produced by startups, this pattern is happening to startups.

19

It's so cheap to start web startups that orders of magnitudes more will be started.

20

If the pattern holds true, that should cause dramatic changes.

4–12

Startups are undergoing the transformation technology does when it gets cheaper: an expensive device made in small quantities becomes, once someone makes it cheaply, something built in volume and used in new ways. Computers went from machines built one at a time to a commodity we stick in everything.

13–17

This pattern is very old—most turning points in economic history are instances. It happened to steel, to power, to cloth manufacture in the thirteenth century, generating the wealth that brought about the Renaissance.

18–20

Now it's happening to startups. It's so cheap to start web startups that orders of magnitude more will be started—and that should cause dramatic changes.

2–20

Startups are undergoing the same transformation technology does when it gets cheap—the oldest pattern in economic history, now happening to startups themselves.

22

1. Lots of Startups

23

So my first prediction about the future of web startups is pretty straightforward: there will be a lot of them.

24

When starting a startup was expensive, you had to get the permission of investors to do it.

25

Now the only threshold is courage.

26

Even that threshold is getting lower, as people watch others take the plunge and survive.

27

In the last batch of startups we funded, we had several founders who said they'd thought of applying before, but weren't sure and got jobs instead.

28

It was only after hearing reports of friends who'd done it that they decided to try it themselves.

29

Starting a startup is hard, but having a 9 to 5 job is hard too, and in some ways a worse kind of hard.

30

In a startup you have lots of worries, but you don't have that feeling that your life is flying by like you do in a big company.

31

Plus in a startup you could make much more money.

32

As word spreads that startups work, the number may grow to a point that would now seem surprising.

33

We now think of it as normal to have a job at a company, but this is the thinnest of historical veneers.

34

Just two or three lifetimes ago, most people in what are now called industrialized countries lived by farming.

35

So while it may seem surprising to propose that large numbers of people will change the way they make a living, it would be more surprising if they didn't.

23–28

My first prediction: there will be a lot of them. When startups were expensive, you needed investors' permission; now the only threshold is courage—and even that is falling as people watch others survive the plunge.

29–31

Starting a startup is hard, but a 9 to 5 job is hard too, in some ways a worse kind of hard: worries, but not that feeling that your life is flying by. Plus you could make much more money.

32–35

A job at a company seems normal, but it's the thinnest of historical veneers. Two or three lifetimes ago most people in industrialized countries lived by farming. It would be more surprising if large numbers didn't change how they make a living.

22–35

When starting a startup was expensive you needed investors' permission; now the only threshold is courage, and even that is falling as people watch friends survive the plunge.

37

2. Standardization

38

When technology makes something dramatically cheaper, standardization always follows.

39

When you make things in large volumes you tend to standardize everything that doesn't need to change.

40

At Y Combinator we still only have four people, so we try to standardize everything.

41

We could hire employees, but we want to be forced to figure out how to scale investing.

42

We often tell startups to release a minimal version one quickly, then let the needs of the users determine what to do next.

43

In essense, let the market design the product.

44

We've done the same thing ourselves.

45

We think of the techniques we're developing for dealing with large numbers of startups as like software.

46

Sometimes it literally is software, like Hacker News and our application system.

47

One of the most important things we've been working on standardizing are investment terms. Till now investment terms have been individually negotiated.

48

This is a problem for founders, because it makes raising money take longer and cost more in legal fees.

49

So as well as using the same paperwork for every deal we do, we've commissioned generic angel paperwork that all the startups we fund can use for future rounds.

50

Some investors will still want to cook up their own deal terms. Series A rounds, where you raise a million dollars or more, will be custom deals for the forseeable future.

51

But I think angel rounds will start to be done mostly with standardized agreements.

52

An angel who wants to insert a bunch of complicated terms into the agreement is probably not one you want anyway.

53

3. New Attitude to Acquisition

54

Another thing I see starting to get standardized is acquisitions.

55

As the volume of startups increases, big companies will start to develop standardized procedures that make acquisitions little more work than hiring someone.

56

Google is the leader here, as in so many areas of technology.

57

They buy a lot of startups— more than most people realize, because they only announce a fraction of them.

58

And being Google, they're figuring out how to do it efficiently.

59

One problem they've solved is how to think about acquisitions.

60

For most companies, acquisitions still carry some stigma of inadequacy.

61

Companies do them because they have to, but there's usually some feeling they shouldn't have to—that their own programmers should be able to build everything they need.

62

Google's example should cure the rest of the world of this idea.

63

Google has by far the best programmers of any public technology company.

64

If they don't have a problem doing acquisitions, the others should have even less problem.

65

However many Google does, Microsoft should do ten times as many.

66

One reason Google doesn't have a problem with acquisitions is that they know first-hand the quality of the people they can get that way.

67

Larry and Sergey only started Google after making the rounds of the search engines trying to sell their idea and finding no takers.

68

They've been the guys coming in to visit the big company, so they know who might be sitting across that conference table from them.

38–46

When technology gets dramatically cheaper, standardization follows. At Y Combinator we still have only four people, so we standardize everything, forcing ourselves to figure out how to scale investing. We treat our techniques as software—sometimes literally, like Hacker News.

47–52

The most important thing we've standardized is investment terms, till now individually negotiated, which makes raising money slower and costlier. So we've commissioned generic angel paperwork startups can reuse for future rounds. Series A rounds will stay custom, but angel rounds will mostly use standardized agreements.

54–58

Acquisitions are getting standardized too. As startups multiply, buying one will be little more work than hiring someone. Google is the leader; they buy more startups than people realize, and being Google, they're doing it efficiently.

59–65

For most companies acquisitions still carry a stigma of inadequacy—a feeling their own programmers should build everything. Google's example should cure that: they have by far the best programmers of any public technology company, so if they have no problem with acquisitions, others should have less. However many Google does, Microsoft should do ten times as many.

66–68

Google has no problem because they know first-hand the quality of people they get that way. Larry and Sergey started Google only after making the rounds of the search engines and finding no takers. They've been the guys coming in to visit the big company.

37–68

Cheapness brings standardization—of investment terms, and of acquisitions. Google leads on buying startups, having cured the old stigma; Larry and Sergey were once the guys across the table.

70

4. Riskier Strategies are Possible

71

Risk is always proportionate to reward.

72

The way to get really big returns is to do things that seem crazy, like starting a new search engine in 1998, or turning down a billion dollar acquisition offer.

73

This has traditionally been a problem in venture funding.

74

Founders and investors have different attitudes to risk.

75

Knowing that risk is on average proportionate to reward, investors like risky strategies, while founders, who don't have a big enough sample size to care what's true on average, tend to be more conservative.

76

If startups are easy to start, this conflict goes away, because founders can start them younger, when it's rational to take more risk, and can start more startups total in their careers.

77

When founders can do lots of startups, they can start to look at the world in the same portfolio-optimizing way as investors.

78

And that means the overall amount of wealth created can be greater, because strategies can be riskier.

79

5. Younger, Nerdier Founders

80

If startups become a cheap commodity, more people will be able to have them, just as more people could have computers once microprocessors made them cheap.

81

And in particular, younger and more technical founders will be able to start startups than could before.

82

Back when it cost a lot to start a startup, you had to convince investors to let you do it.

83

And that required very different skills from actually doing the startup.

84

If investors were perfect judges, the two would require exactly the same skills.

85

But unfortunately most investors are terrible judges.

86

I know because I see behind the scenes what an enormous amount of work it takes to raise money, and the amount of selling required in an industry is always inversely proportional to the judgement of the buyers.

87

Fortunately, if startups get cheaper to start, there's another way to convince investors.

88

Instead of going to venture capitalists with a business plan and trying to convince them to fund it, you can get a product launched on a few tens of thousands of dollars of seed money from us or your uncle, and approach them with a working company instead of a plan for one.

89

Then instead of having to seem smooth and confident, you can just point them to Alexa.

90

This way of convincing investors is better suited to hackers, who often went into technology in part because they felt uncomfortable with the amount of fakeness required in other fields.

71–72

Risk is always proportionate to reward. The way to get really big returns is to do things that seem crazy, like starting a new search engine in 1998, or turning down a billion-dollar acquisition offer.

73–78

Founders and investors differ on risk: investors like risky strategies, knowing reward tracks risk on average, while founders lack the sample size to care about averages. If startups are easy to start, the conflict goes away—founders can start them younger and start more over a career, optimizing a portfolio like investors do, so more wealth gets created.

80–86

Cheap startups also mean younger, more technical founders. When it cost a lot, you had to convince investors—different skills from doing the startup—and most investors are terrible judges. The selling required in an industry is always inversely proportional to the judgement of the buyers.

87–90

If startups get cheaper, there's another way to convince investors: launch on a few tens of thousands in seed money and approach them with a working company. Instead of seeming smooth, you just point them to Alexa—a method better suited to hackers, who dislike the fakeness required in other fields.

70–90

Risk is proportionate to reward, so investors like risk and conservative founders don't. Cheap startups close that gap—and let younger, more technical hackers convince investors with a working company instead of a pitch.

92

6. Startup Hubs Will Persist

93

It might seem that if startups get cheap to start, it will mean the end of startup hubs like Silicon Valley.

94

If all you need to start a startup is rent money, you should be able to do it anywhere.

95

This is kind of true and kind of false.

96

It's true that you can now start a startup anywhere.

97

But you have to do more with a startup than just start it.

98

You have to make it succeed.

99

And that is more likely to happen in a startup hub.

100

I've thought a lot about this question, and it seems to me the increasing cheapness of web startups will if anything increase the importance of startup hubs.

101

The value of startup hubs, like centers for any kind of business, lies in something very old-fashioned: face to face meetings.

102

No technology in the immediate future will replace walking down University Ave and running into a friend who tells you how to fix a bug that's been bothering you all weekend, or visiting a friend's startup down the street and ending up in a conversation with one of their investors.

103

The question of whether to be in a startup hub is like the question of whether to take outside investment.

104

The question is not whether you need it, but whether it brings any advantage at all.

105

Because anything that brings an advantage will give your competitors an advantage over you if they do it and you don't.

106

So if you hear someone saying "we don't need to be in Silicon Valley," that use of the word "need" is a sign they're not even thinking about the question right.

107

And while startup hubs are as powerful magnets as ever, the increasing cheapness of starting a startup means the particles they're attracting are getting lighter.

108

A startup now can be just a pair of 22 year old guys.

109

A company like that can move much more easily than one with 10 people, half of whom have kids.

110

We know because we make people move for Y Combinator, and it doesn't seem to be a problem.

111

The advantage of being able to work together face to face for three months outweighs the inconvenience of moving.

112

Ask anyone who's done it.

113

The mobility of seed-stage startups means that seed funding is a national business.

114

One of the most common emails we get is from people asking if we can help them set up a local clone of Y Combinator.

115

But this just wouldn't work.

116

Seed funding isn't regional, just as big research universities aren't.

117

Is seed funding not merely national, but international?

118

Interesting question.

119

There are signs it may be.

120

We've had an ongoing stream of founders from outside the US, and they tend to do particularly well, because they're all people who were so determined to succeed that they were willing to move to another country to do it.

121

The more mobile startups get, the harder it would be to start new silicon valleys.

122

If startups are mobile, the best local talent will go to the real Silicon Valley, and all they'll get at the local one will be the people who didn't have the energy to move.

123

This is not a nationalistic idea, incidentally.

124

It's cities that compete, not countries.

125

Atlanta is just as hosed as Munich.

93–99

It might seem cheap startups mean the end of hubs like Silicon Valley. Kind of true, kind of false. You can now start a startup anywhere. But you have to make it succeed, and that's more likely in a hub.

100–102

Cheapness will if anything increase the importance of hubs, whose value lies in something old-fashioned: face-to-face meetings. No technology will soon replace walking down University Ave and running into a friend who tells you how to fix a bug that's plagued you all weekend.

103–106

Whether to be in a hub is like whether to take outside investment: not whether you need it, but whether it brings any advantage—because anything that brings an advantage gives competitors one over you if they do it and you don't.

107–112

Hubs are as powerful as ever, but the particles they attract are getting lighter. A startup now can be a pair of 22-year-old guys—able to move far more easily than a company of 10, half of whom have kids. We make people move for Y Combinator, and it isn't a problem.

113–120

The mobility of seed-stage startups makes seed funding a national, maybe international, business. People ask us to help set up a local clone of Y Combinator, but seed funding isn't regional, just as big research universities aren't. Our founders from outside the US do particularly well—determined enough to move countries to succeed.

121–125

The more mobile startups get, the harder to start new Silicon Valleys: the best local talent goes to the real one, and the clone gets those who lacked the energy to move. It's cities that compete, not countries. Atlanta is just as hosed as Munich.

92–125

You can now start a startup anywhere, but you have to make it succeed, and that still happens in hubs. Cheapness makes startups lighter and more mobile—which only strengthens the real Silicon Valley.

127

7. Better Judgement Needed

128

If the number of startups increases dramatically, then the people whose job is to judge them are going to have to get better at it.

129

I'm thinking particularly of investors and acquirers.

130

We now get on the order of 1000 applications a year.

131

What are we going to do if we get 10,000?

132

That's actually an alarming idea.

133

But we'll figure out some kind of answer.

134

We'll have to.

135

It will probably involve writing some software, but fortunately we can do that.

136

Acquirers will also have to get better at picking winners.

137

They generally do better than investors, because they pick later, when there's more performance to measure.

138

But even at the most advanced acquirers, identifying companies to buy is extremely ad hoc, and completing the acquisition often involves a great deal of unneccessary friction.

139

I think acquirers may eventually have chief acquisition officers who will both identify good acquisitions and make the deals happen.

140

At the moment those two functions are separate.

141

Promising new startups are often discovered by developers.

142

If someone powerful enough wants to buy them, the deal is handed over to corp dev guys to negotiate.

143

It would be better if both were combined in one group, headed by someone with a technical background and some vision of what they wanted to accomplish.

144

Maybe in the future big companies will have both a VP of Engineering responsible for technology developed in-house, and a CAO responsible for bringing technology in from outside.

145

At the moment, there is no one within big companies who gets in trouble when they buy a startup for $200 million that they could have bought earlier for $20 million.

146

There should start to be someone who gets in trouble for that.

128–135

If startups multiply, the people who judge them—investors and acquirers—must get better at it. We now get on the order of 1000 applications a year; what do we do at 10,000? We'll figure out some answer—probably by writing software, which fortunately we can do.

136–144

Acquirers do better than investors, picking later with more performance to measure, but even the most advanced are extremely ad hoc. They may eventually have chief acquisition officers who both find good acquisitions and make them happen—functions now split between developers who spot startups and corp dev guys who negotiate.

145–146

At the moment, no one in big companies gets in trouble for buying a startup for $200 million that they could have bought earlier for $20 million. There should start to be someone who gets in trouble for that.

127–146

With far more startups, investors and acquirers will have to get much better at judging them. Acquirers are still ad hoc; PG imagines a chief acquisition officer who gets in trouble for buying late and expensive.

148

8. College Will Change

149

If the best hackers start their own companies after college instead of getting jobs, that will change what happens in college.

150

Most of these changes will be for the better.

151

I think the experience of college is warped in a bad way by the expectation that afterward you'll be judged by potential employers.

152

One change will be in the meaning of "after college," which will switch from when one graduates from college to when one leaves it.

153

If you're starting your own company, why do you need a degree?

154

We don't encourage people to start startups during college, but the best founders are certainly capable of it.

155

Some of the most successful companies we've funded were started by undergrads.

156

I grew up in a time where college degrees seemed really important, so I'm alarmed to be saying things like this, but there's nothing magical about a degree.

157

There's nothing that magically changes after you take that last exam.

158

The importance of degrees is due solely to the administrative needs of large organizations.

159

These can certainly affect your life—it's hard to get into grad school, or to get a work visa in the US, without an undergraduate degree—but tests like this will matter less and less.

160

As well as mattering less whether students get degrees, it will also start to matter less where they go to college.

161

In a startup you're judged by users, and they don't care where you went to college.

162

So in a world of startups, elite universities will play less of a role as gatekeepers.

163

In the US it's a national scandal how easily children of rich parents game college admissions.

164

But the way this problem ultimately gets solved may not be by reforming the universities but by going around them.

165

We in the technology world are used to that sort of solution: you don't beat the incumbents; you redefine the problem to make them irrelevant.

166

The greatest value of universities is not the brand name or perhaps even the classes so much as the people you meet.

167

If it becomes common to start a startup after college, students may start trying to maximize this.

168

Instead of focusing on getting internships at companies they want to work for, they may start to focus on working with other students they want as cofounders.

169

What students do in their classes will change too.

170

Instead of trying to get good grades to impress future employers, students will try to learn things.

171

We're talking about some pretty dramatic changes here.

149–155

If the best hackers start companies after college instead of getting jobs, college changes—mostly for the better, since the experience is warped by the expectation of being judged by employers. "After college" will switch from when you graduate to when you leave: if you're starting your own company, why do you need a degree? Some of our most successful companies were started by undergrads.

156–159

I grew up when degrees seemed important, so I'm alarmed to say it, but there's nothing magical about one—nothing changes after that last exam. Degrees matter solely because of the administrative needs of large organizations, and tests like this will matter less and less.

160–165

Where you go will matter less too. In a startup you're judged by users, who don't care where you went, so elite universities play less of a gatekeeper role. It's a national scandal how easily rich children game US admissions—but this gets solved not by reforming universities but by going around them. In technology we know that move: you don't beat the incumbents, you redefine the problem to make them irrelevant.

166–170

A university's greatest value is the people you meet. Students may start to maximize this—working with future cofounders instead of chasing internships—and in class, instead of chasing grades, they'll try to learn things.

148–171

If the best hackers start companies instead of getting jobs, college changes for the better. Degrees and brand-name schools matter less; what students maximize shifts from impressing employers to finding cofounders and actually learning.

173

9. Lots of Competitors

174

If it gets easier to start a startup, it's easier for competitors too.

175

That doesn't erase the advantage of increased cheapness, however.

176

You're not all playing a zero-sum game.

177

There's not some fixed number of startups that can succeed, regardless of how many are started.

178

In fact, I don't think there's any limit to the number of startups that could succeed.

179

Startups succeed by creating wealth, which is the satisfaction of people's desires.

180

And people's desires seem to be effectively infinite, at least in the short term.

181

What the increasing number of startups does mean is that you won't be able to sit on a good idea.

182

Other people have your idea, and they'll be increasingly likely to do something about it.

183

10. Faster Advances

184

There's a good side to that, at least for consumers of technology.

185

If people get right to work implementing ideas instead of sitting on them, technology will evolve faster.

186

Some kinds of innovations happen a company at a time, like the punctuated equilibrium model of evolution.

187

There are some kinds of ideas that are so threatening that it's hard for big companies even to think of them.

188

Look at what a hard time Microsoft is having discovering web apps.

189

They're like a character in a movie that everyone in the audience can see something bad is about to happen to, but who can't see it himself.

190

The big innovations that happen a company at a time will obviously happen faster if the rate of new companies increases.

191

But in fact there will be a double speed increase.

192

People won't wait as long to act on new ideas, but also those ideas will increasingly be developed within startups rather than big companies.

193

Which means technology will evolve faster per company as well.

194

Big companies are just not a good place to make things happen fast. I talked recently to a founder whose startup had been acquired by a big company.

195

He was a precise sort of guy, so he'd measured their productivity before and after.

196

He counted lines of code, which can be a dubious measure, but in this case was meaningful because it was the same group of programmers.

197

He found they were one thirteenth as productive after the acquisition.

198

The company that bought them was not a particularly stupid one.

199

I think what he was measuring was mostly the cost of bigness.

200

I experienced this myself, and his number sounds about right.

201

There's something about big companies that just sucks the energy out of you.

202

Imagine what all that energy could do if it were put to use.

203

There is an enormous latent capacity in the world's hackers that most people don't even realize is there.

204

That's the main reason we do Y Combinator: to let loose all this energy by making it easy for hackers to start their own startups.

174–180

Easier startups mean easier competitors too, but that doesn't erase the advantage of cheapness—it's not zero-sum. I don't think there's any limit to the number that can succeed: startups create wealth, the satisfaction of people's desires, and desires seem effectively infinite.

181–182

What more startups does mean is you won't be able to sit on a good idea. Others have it, and they'll be increasingly likely to do something about it.

184–190

For consumers, the good side is faster technology. Some innovations happen a company at a time—ideas so threatening big companies can't even think of them. Look at Microsoft's hard time discovering web apps: like a character in a movie everyone in the audience can see something bad is about to happen to, but who can't see it himself.

191–193

And a double speed increase: people won't wait as long to act, and ideas will increasingly be developed within startups rather than big companies—so technology evolves faster per company too.

194–201

Big companies are just not a good place to make things happen fast. A founder whose startup was acquired measured productivity by lines of code—the same programmers—and found them one thirteenth as productive afterward. What he measured was the cost of bigness. There's something about big companies that just sucks the energy out of you.

202–204

Imagine what all that energy could do if put to use. There's enormous latent capacity in the world's hackers—and letting it loose is the main reason we do Y Combinator.

173–204

Easier startups mean more competitors, but it isn't zero-sum—wealth, and human desire, are effectively unlimited. You just can't sit on an idea anymore. And technology evolves faster, because big companies suck the energy out of everything.

206

A Series of Tubes

207

The process of starting startups is currently like the plumbing in an old house.

208

The pipes are narrow and twisty, and there are leaks in every joint.

209

In the future this mess will gradually be replaced by a single, huge pipe.

210

The water will still have to get from A to B, but it will get there faster and without the risk of spraying out through some random leak.

211

This will change a lot of things for the better.

212

In a big, straight pipe like that, the force of being measured by one's performance will propagate back through the whole system.

213

Performance is always the ultimate test, but there are so many kinks in the plumbing now that most people are insulated from it most of the time.

214

So you end up with a world in which high school students think they need to get good grades to get into elite colleges, and college students think they need to get good grades to impress employers, within which the employees waste most of their time in political battles, and from which consumers have to buy anyway because there are so few choices.

215

Imagine if that sequence became a big, straight pipe.

216

Then the effects of being measured by performance would propagate all the way back to high school, flushing out all the arbitrary stuff people are measured by now.

217

That is the future of web startups.

207–210

Starting startups now is like the plumbing in an old house: pipes narrow and twisty, leaks at every joint. In the future this mess becomes a single, huge pipe—the water still gets from A to B, but faster and without spraying out through some random leak.

211–214

In that big, straight pipe, the force of being measured by performance propagates back through the whole system. There are so many kinks now that most are insulated from it—high schoolers chasing grades for colleges, college students for employers, employees wasting time in political battles, consumers buying anyway because there are so few choices.

215–217

Imagine that sequence became a big, straight pipe. Then being measured by performance would propagate all the way back to high school, flushing out all the arbitrary stuff people are measured by now. That is the future of web startups.

206–217

Starting startups is now like the plumbing in an old house—narrow, twisty, leaky. Replace it with a single huge pipe and the force of being measured by performance propagates all the way back to high school.

219

Thanks to Brian Oberkirch and Simon Willison for inviting me to speak, and the crew at Carson Systems for making everything run smoothly.

219

Thanks to Brian Oberkirch and Simon Willison for inviting me to speak, and the crew at Carson Systems for making everything run smoothly.

219

Thanks to Brian Oberkirch and Simon Willison for the FOWA invitation, and the Carson Systems crew for running it smoothly.