pgstrata
Ideas for Startups
2

October 2005

3

(This essay is derived from a talk at the 2005 Startup School.)

4

How do you get good ideas for startups [blocked]?

5

That's probably the number one question people ask me.

6

I'd like to reply with another question: why do people think it's hard to come up with ideas for startups?

7

That might seem a stupid thing to ask.

8

Why do they think it's hard?

9

If people can't do it, then it is hard, at least for them.

10

Right?

11

Well, maybe not.

12

What people usually say is not that they can't think of ideas, but that they don't have any.

13

That's not quite the same thing.

14

It could be the reason they don't have any is that they haven't tried to generate them.

15

I think this is often the case.

16

I think people believe that coming up with ideas for startups is very hard-- that it must be very hard-- and so they don't try do to it.

17

They assume ideas are like miracles: they either pop into your head or they don't.

18

I also have a theory about why people think this.

19

They overvalue ideas.

20

They think creating a startup is just a matter of implementing some fabulous initial idea.

21

And since a successful startup is worth millions of dollars, a good idea is therefore a million dollar idea.

22

If coming up with an idea for a startup equals coming up with a million dollar idea, then of course it's going to seem hard.

23

Too hard to bother trying.

24

Our instincts tell us something so valuable would not be just lying around for anyone to discover.

25

Actually, startup ideas are not million dollar ideas, and here's an experiment you can try to prove it: just try to sell one.

26

Nothing evolves faster than markets.

27

The fact that there's no market for startup ideas suggests there's no demand.

28

Which means, in the narrow sense of the word, that startup ideas are worthless.

4–6

How do you get good startup [blocked] ideas? My number one question. I'd reply with another: why do people think it's hard?

11–17

People say not that they can't think of ideas, but that they don't have any-- maybe because they never tried to generate them. They assume ideas are like miracles: they either pop into your head or they don't.

18–24

My theory: they overvalue ideas. Since a successful startup is worth millions, they figure a good idea is a million dollar idea-- too hard to bother with. Our instincts say something so valuable wouldn't just be lying around.

25–28

An experiment to disprove it: just try to sell one. Nothing evolves faster than markets, and the absence of a market for startup ideas means no demand. In the narrow sense, startup ideas are worthless.

2–28

People think startup ideas are hard to get because they overvalue them, imagining each is a million-dollar secret. But there's no market for startup ideas, which means they're worthless.

30

Questions

31

The fact is, most startups end up nothing like the initial idea.

32

It would be closer to the truth to say the main value of your initial idea is that, in the process of discovering it's broken, you'll come up with your real idea.

33

The initial idea is just a starting point-- not a blueprint, but a question.

34

It might help if they were expressed that way.

35

Instead of saying that your idea is to make a collaborative, web-based spreadsheet, say: could one make a collaborative, web-based spreadsheet?

36

A few grammatical tweaks, and a woefully incomplete idea becomes a promising question to explore.

37

There's a real difference, because an assertion provokes objections in a way a question doesn't.

38

If you say: I'm going to build a web-based spreadsheet, then critics-- the most dangerous of which are in your own head-- will immediately reply that you'd be competing with Microsoft, that you couldn't give people the kind of UI they expect, that users wouldn't want to have their data on your servers, and so on.

39

A question doesn't seem so challenging.

40

It becomes: let's try making a web-based spreadsheet and see how far we get.

41

And everyone knows that if you tried this you'd be able to make something useful.

42

Maybe what you'd end up with wouldn't even be a spreadsheet.

43

Maybe it would be some kind of new spreadsheet-like collaboration tool that doesn't even have a name yet.

44

You wouldn't have thought of something like that except by implementing your way toward it.

45

Treating a startup idea as a question changes what you're looking for.

46

If an idea is a blueprint, it has to be right.

47

But if it's a question, it can be wrong, so long as it's wrong in a way that leads to more ideas.

48

One valuable way for an idea to be wrong is to be only a partial solution.

49

When someone's working on a problem that seems too big, I always ask: is there some way to bite off some subset of the problem, then gradually expand from there?

50

That will generally work unless you get trapped on a local maximum, like 1980s-style AI, or C.

31–32

Most startups end up nothing like the initial idea. Its main value is that, in discovering it's broken, you find your real idea.

33–36

The initial idea is a starting point-- not a blueprint, but a question. Instead of "my idea is a collaborative, web-based spreadsheet," say: could one? A few grammatical tweaks turn a woefully incomplete idea into a promising question.

37–38

An assertion provokes objections a question doesn't. Say you'll build a web-based spreadsheet, and critics-- the most dangerous in your own head-- reply you'd be competing with Microsoft, couldn't give the expected UI, and users wouldn't trust your servers.

39–44

A question is just: let's try and see how far we get. You'd make something useful-- maybe not a spreadsheet, but some collaboration tool you'd reach only by implementing toward it.

45–50

A blueprint has to be right; a question can be wrong, so long as it leads to more ideas. One valuable way to be wrong is to be a partial solution. When a problem seems too big, I ask: can you bite off a subset, then expand? That works unless you get trapped on a local maximum, like 1980s-style AI, or C.

30–50

Most startups end up nothing like their initial idea, so treat the idea as a question, not a blueprint. A question provokes no objections and can be productively wrong.

52

Upwind

53

So far, we've reduced the problem from thinking of a million dollar idea to thinking of a mistaken question.

54

That doesn't seem so hard, does it?

55

To generate such questions you need two things: to be familiar with promising new technologies, and to have the right kind of friends.

56

New technologies are the ingredients startup ideas are made of, and conversations with friends are the kitchen they're cooked in.

57

Universities have both, and that's why so many startups grow out of them.

58

They're filled with new technologies, because they're trying to produce research, and only things that are new count as research.

59

And they're full of exactly the right kind of people to have ideas with: the other students, who will be not only smart but elastic-minded to a fault.

60

The opposite extreme would be a well-paying but boring job at a big company.

61

Big companies are biased against new technologies, and the people you'd meet there would be wrong too.

62

In an essay [blocked] I wrote for high school students, I said a good rule of thumb was to stay upwind-- to work on things that maximize your future options.

63

The principle applies for adults too, though perhaps it has to be modified to: stay upwind for as long as you can, then cash in the potential energy you've accumulated when you need to pay for kids.

64

I don't think people consciously realize this, but one reason downwind jobs like churning out Java for a bank pay so well is precisely that they are downwind.

65

The market price for that kind of work is higher because it gives you fewer options for the future.

66

A job that lets you work on exciting new stuff will tend to pay less, because part of the compensation is in the form of the new skills you'll learn.

67

Grad school is the other end of the spectrum from a coding job at a big company: the pay's low but you spend most of your time working on new stuff.

68

And of course, it's called "school," which makes that clear to everyone, though in fact all jobs are some percentage school.

69

The right environment for having startup ideas need not be a university per se.

70

It just has to be a situation with a large percentage of school.

71

It's obvious why you want exposure to new technology, but why do you need other people?

72

Can't you just think of new ideas yourself?

73

The empirical answer is: no. Even Einstein needed people to bounce ideas off.

74

Ideas get developed in the process of explaining them to the right kind of person.

75

You need that resistance, just as a carver needs the resistance of the wood.

76

This is one reason Y Combinator has a rule against investing in startups with only one founder.

77

Practically every successful company has at least two.

78

And because startup founders work under great pressure, it's critical they be friends.

79

I didn't realize it till I was writing this, but that may help explain why there are so few female startup founders.

80

I read on the Internet (so it must be true) that only 1.7% of VC-backed startups are founded by women.

81

The percentage of female hackers is small, but not that small.

82

So why the discrepancy?

83

When you realize that successful startups tend to have multiple founders who were already friends, a possible explanation emerges.

84

People's best friends are likely to be of the same sex, and if one group is a minority in some population, pairs of them will be a minority squared. [1]

55–56

To generate such questions you need two things: familiarity with promising new technologies, and the right kind of friends. New technologies are the ingredients; conversations with friends are the kitchen.

57–61

Universities have both: new technologies, because only new things count as research, and the right people-- students, smart but elastic-minded to a fault. The opposite is a well-paying but boring big-company job, biased against new technologies and full of the wrong people.

62–63

In an essay [blocked] for high schoolers I said: stay upwind-- work on things that maximize your future options. For adults: stay upwind as long as you can, then cash in the potential energy when you need to pay for kids.

64–70

Downwind jobs like churning out Java for a bank pay well precisely because they're downwind: more pay for fewer future options. Exciting new work pays less, because part of the compensation is the skills you learn-- grad school being the extreme: low pay, mostly new stuff. The right environment need not be a university, just a large percentage of school.

71–74

Why other people? Can't you think alone? Empirically, no-- even Einstein needed people to bounce ideas off. Ideas develop in explaining them; you need that resistance, as a carver needs the resistance of the wood.

75–77

This is one reason Y Combinator won't fund single-founder startups. Practically every successful company has at least two, and since founders work under great pressure, it's critical they be friends.

78–83

That may explain why there are so few female startup founders. Only 1.7% are women, but the share of female hackers isn't that small. Since founders were already friends, and best friends tend to be the same sex, pairs from a minority are a minority squared. [1]

52–84

To generate good questions you need familiarity with new technologies and the right kind of friends. Universities have both; you need other people because ideas develop in explaining them.

86

Doodling

87

What these groups of co-founders do together is more complicated than just sitting down and trying to think of ideas.

88

I suspect the most productive setup is a kind of together-alone-together sandwich.

89

Together you talk about some hard problem, probably getting nowhere.

90

Then, the next morning, one of you has an idea in the shower about how to solve it.

91

He runs eagerly to to tell the others, and together they work out the kinks.

92

What happens in that shower?

93

It seems to me that ideas just pop into my head.

94

But can we say more than that?

95

Taking a shower is like a form of meditation.

96

You're alert, but there's nothing to distract you.

97

It's in a situation like this, where your mind is free to roam, that it bumps into new ideas.

98

What happens when your mind wanders?

99

It may be like doodling.

100

Most people have characteristic ways of doodling.

101

This habit is unconscious, but not random: I found my doodles changed after I started studying painting.

102

I started to make the kind of gestures I'd make if I were drawing from life.

103

They were atoms of drawing, but arranged randomly. [2]

104

Perhaps letting your mind wander is like doodling with ideas.

105

You have certain mental gestures you've learned in your work, and when you're not paying attention, you keep making these same gestures, but somewhat randomly.

106

In effect, you call the same functions on random arguments.

107

That's what a metaphor is: a function applied to an argument of the wrong type.

108

Conveniently, as I was writing this, my mind wandered: would it be useful to have metaphors in a programming language?

109

I don't know; I don't have time to think about this.

110

But it's convenient because this is an example of what I mean by habits of mind.

111

I spend a lot of time thinking about language design, and my habit of always asking "would x be useful in a programming language" just got invoked.

112

If new ideas arise like doodles, this would explain why you have to work at something for a while before you have any.

113

It's not just that you can't judge ideas till you're an expert in a field.

114

You won't even generate ideas, because you won't have any habits of mind to invoke.

115

Of course the habits of mind you invoke on some field don't have to be derived from working in that field.

116

In fact, it's often better if they're not.

117

You're not just looking for good ideas, but for good new ideas, and you have a better chance of generating those if you combine stuff from distant fields.

118

As hackers, one of our habits of mind is to ask, could one open-source x?

119

For example, what if you made an open-source operating system?

120

A fine idea, but not very novel.

121

Whereas if you ask, could you make an open-source play? you might be onto something.

122

Are some kinds of work better sources of habits of mind than others?

123

I suspect harder fields may be better sources, because to attack hard problems you need powerful solvents.

124

I find math is a good source of metaphors-- good enough that it's worth studying just for that.

125

Related fields are also good sources, especially when they're related in unexpected ways.

126

Everyone knows computer science and electrical engineering are related, but precisely because everyone knows it, importing ideas from one to the other doesn't yield great profits.

127

It's like importing something from Wisconsin to Michigan.

128

Whereas (I claim) hacking and painting [blocked] are also related, in the sense that hackers and painters are both makers [blocked], and this source of new ideas is practically virgin territory.

87–91

The most productive setup is together-alone-together: together you talk about a hard problem, getting nowhere; then one of you has an idea in the shower; then together you work out the kinks.

92–97

What happens in that shower? A shower is like meditation: alert, but nothing distracts you. It's where your mind is free to roam that it bumps into new ideas.

98–103

A wandering mind may be like doodling. People doodle in characteristic, unconscious but not random ways: mine changed after I studied painting, becoming atoms of drawing arranged randomly. [2]

104–107

Letting your mind wander is doodling with ideas: you keep making the mental gestures learned in your work, but randomly-- calling the same functions on random arguments. That's what a metaphor is: a function applied to an argument of the wrong type.

108–114

As I wrote this my mind wandered: would metaphors be useful in a programming language? A perfect example of a habit of mind. If ideas arise like doodles, this explains why you must work at something before you have any: you won't even generate ideas without habits of mind to invoke.

115–121

The habits needn't come from the field you're in-- often better if they don't, since you want good new ideas. As hackers we ask, could one open-source x? An open-source OS is fine but not novel; an open-source play might be onto something.

122–128

Harder fields may be better sources, since hard problems need powerful solvents-- math is worth studying just for its metaphors. Related fields help only when related unexpectedly: importing from CS to EE, which everyone knows are related, is like importing from Wisconsin to Michigan. Whereas hacking and painting [blocked] are related because both are makers [blocked]-- virgin territory.

86–128

Ideas pop into your head when your mind is free to wander, like doodling. You doodle with the mental gestures learned in your work, so you need habits of mind first-- and combining distant fields yields the newest ideas.

130

Problems

131

In theory you could stick together ideas at random and see what you came up with.

132

What if you built a peer-to-peer dating site?

133

Would it be useful to have an automatic book?

134

Could you turn theorems into a commodity?

135

When you assemble ideas at random like this, they may not be just stupid, but semantically ill-formed.

136

What would it even mean to make theorems a commodity?

137

You got me.

138

I didn't think of that idea, just its name.

139

You might come up with something useful this way, but I never have.

140

It's like knowing a fabulous sculpture is hidden inside a block of marble, and all you have to do is remove the marble that isn't part of it.

141

It's an encouraging thought, because it reminds you there is an answer, but it's not much use in practice because the search space is too big.

142

I find that to have good ideas I need to be working on some problem.

143

You can't start with randomness.

144

You have to start with a problem, then let your mind wander just far enough for new ideas to form.

145

In a way, it's harder to see problems than their solutions.

146

Most people prefer to remain in denial about problems. It's obvious why: problems are irritating.

147

They're problems!

148

Imagine if people in 1700 saw their lives the way we'd see them.

149

It would have been unbearable.

150

This denial is such a powerful force that, even when presented with possible solutions, people often prefer to believe they wouldn't work.

151

I saw this phenomenon when I worked on spam filters.

152

In 2002, most people preferred to ignore spam, and most of those who didn't preferred to believe the heuristic filters then available were the best you could do.

153

I found spam intolerable, and I felt it had to be possible to recognize it statistically.

154

And it turns out that was all you needed to solve the problem.

155

The algorithm I used was ridiculously simple.

156

Anyone who'd really tried to solve the problem would have found it.

157

It was just that no one had really tried to solve the problem. [3]

158

Let me repeat that recipe: finding the problem intolerable and feeling it must be possible to solve it.

159

Simple as it seems, that's the recipe for a lot of startup ideas.

131–135

In theory you could stick ideas together at random. Could you turn theorems into a commodity? Assembled this way they may be not just stupid but semantically ill-formed-- I didn't think of that one, just its name.

136–138

You might get something useful, but I never have. It's like knowing a sculpture is hidden in marble and you need only remove the rest-- encouraging, but useless, because the search space is too big.

139–141

To have good ideas I need to be working on some problem. You can't start with randomness; start with a problem, then let your mind wander just far enough for new ideas to form.

142–148

It's harder to see problems than solutions, because people prefer denial-- problems are irritating. Imagine people in 1700 seeing their lives as we would; unbearable. This denial is so powerful that, even shown solutions, people prefer to believe they wouldn't work.

149–152

I saw this with spam filters. In 2002 most preferred to ignore spam, or believed the heuristic filters then available were the best possible. I found spam intolerable, and felt it must be recognizable statistically.

153–157

That was all you needed. The algorithm was ridiculously simple-- anyone who'd really tried would have found it. No one had. [3] The recipe: finding a problem intolerable and feeling it must be solvable. That's the recipe for a lot of startup ideas.

130–159

You can't generate ideas from randomness; the search space is too big. You have to start with a problem-- and seeing problems is harder than solving them, because most people prefer denial. Find one intolerable that must be solvable.

161

Wealth

162

So far most of what I've said applies to ideas in general.

163

What's special about startup ideas?

164

Startup ideas are ideas for companies, and companies have to make money.

165

And the way to make money is to make something people want.

166

Wealth is what people want.

167

I don't mean that as some kind of philosophical statement; I mean it as a tautology.

168

So an idea for a startup is an idea for something people want.

169

Wouldn't any good idea be something people want?

170

Unfortunately not.

171

I think new theorems are a fine thing to create, but there is no great demand for them.

172

Whereas there appears to be great demand for celebrity gossip magazines.

173

Wealth is defined democratically.

174

Good ideas and valuable ideas are not quite the same thing; the difference is individual tastes.

175

But valuable ideas are very close to good ideas, especially in technology.

176

I think they're so close that you can get away with working as if the goal were to discover good ideas, so long as, in the final stage, you stop and ask: will people actually pay for this?

177

Only a few ideas are likely to make it that far and then get shot down; RPN calculators might be one example.

178

One way to make something people want is to look at stuff people use now that's broken.

179

Dating sites are a prime example.

180

They have millions of users, so they must be promising something people want.

181

And yet they work horribly.

182

Just ask anyone who uses them.

183

It's as if they used the worse-is-better approach but stopped after the first stage and handed the thing over to marketers.

184

Of course, the most obvious breakage in the average computer user's life is Windows itself.

185

But this is a special case: you can't defeat a monopoly by a frontal attack.

186

Windows can and will be overthrown, but not by giving people a better desktop OS.

187

The way to kill it is to redefine the problem as a superset of the current one.

188

The problem is not, what operating system should people use on desktop computers? but how should people use applications?

189

There are answers to that question that don't even involve desktop computers.

190

Everyone thinks Google is going to solve this problem, but it is a very subtle one, so subtle that a company as big as Google might well get it wrong.

191

I think the odds are better than 50-50 that the Windows killer-- or more accurately, Windows transcender-- will come from some little startup.

192

Another classic way to make something people want is to take a luxury and make it into a commmodity.

193

People must want something if they pay a lot for it.

194

And it is a very rare product that can't be made dramatically cheaper if you try.

195

This was Henry Ford's plan.

196

He made cars, which had been a luxury item, into a commodity.

197

But the idea is much older than Henry Ford.

198

Water mills transformed mechanical power from a luxury into a commodity, and they were used in the Roman empire.

199

Arguably pastoralism transformed a luxury into a commodity.

200

When you make something cheaper you can sell more of them.

201

But if you make something dramatically cheaper you often get qualitative changes, because people start to use it in different ways.

202

For example, once computers get so cheap that most people can have one of their own, you can use them as communication devices.

203

Often to make something dramatically cheaper you have to redefine the problem.

204

The Model T didn't have all the features previous cars did.

205

It only came in black, for example.

206

But it solved the problem people cared most about, which was getting from place to place.

207

One of the most useful mental habits I know I learned from Michael Rabin: that the best way to solve a problem is often to redefine it.

208

A lot of people use this technique without being consciously aware of it, but Rabin was spectacularly explicit.

209

You need a big prime number?

210

Those are pretty expensive.

211

How about if I give you a big number that only has a 10 to the minus 100 chance of not being prime?

212

Would that do?

213

Well, probably; I mean, that's probably smaller than the chance that I'm imagining all this anyway.

214

Redefining the problem is a particularly juicy heuristic when you have competitors, because it's so hard for rigid-minded people to follow.

215

You can work in plain sight and they don't realize the danger.

216

Don't worry about us.

217

We're just working on search.

218

Do one thing and do it well, that's our motto.

219

Making things cheaper is actually a subset of a more general technique: making things easier.

220

For a long time it was most of making things easier, but now that the things we build are so complicated, there's another rapidly growing subset: making things easier to use.

221

This is an area where there's great room for improvement.

222

What you want to be able to say about technology is: it just works.

223

How often do you say that now?

224

Simplicity takes effort-- genius, even.

225

The average programmer seems to produce UI designs that are almost willfully bad.

226

I was trying to use the stove at my mother's house a couple weeks ago.

227

It was a new one, and instead of physical knobs it had buttons and an LED display.

228

I tried pressing some buttons I thought would cause it to get hot, and you know what it said?

229

"Err."

230

Not even "Error."

231

"Err." You can't just say "Err" to the user of a stove.

232

You should design the UI so that errors are impossible.

233

And the boneheads who designed this stove even had an example of such a UI to work from: the old one.

234

You turn one knob to set the temperature and another to set the timer.

235

What was wrong with that?

236

It just worked.

237

It seems that, for the average engineer, more options just means more rope to hang yourself.

238

So if you want to start a startup, you can take almost any existing technology produced by a big company, and assume you could build something way easier to use.

162–166

What's special about startup ideas? Companies make money by making something people want. Wealth is what people want-- not philosophy, but a tautology.

167–176

Wouldn't any good idea be something people want? Not quite: no demand for new theorems, great demand for celebrity gossip. Wealth is defined democratically. Still, valuable ideas are very close to good ideas in technology-- close enough to chase good ideas, so long as you finally ask: will people pay for this?

178–183

One way to make something people want: look at stuff people use now that's broken. Dating sites are a prime example-- millions of users, yet they work horribly, as if they used worse-is-better but stopped after the first stage and handed it to marketers.

184–189

The most obvious breakage is Windows, but a monopoly can't be beaten by frontal attack. You kill it by redefining the problem as a superset: not what OS desktops should run, but how should people use applications-- and some answers don't involve desktop computers.

190–191

Everyone thinks Google will solve this, but it's so subtle a company that big might get it wrong. Odds are better than even the Windows transcender comes from some little startup.

192–199

Another classic move: take a luxury and make it a commodity. People must want something if they pay a lot, and few products can't be made dramatically cheaper. Henry Ford's plan-- but much older: water mills made mechanical power a commodity in the Roman empire.

200–206

Made dramatically cheaper, things change qualitatively, because people use them in new ways-- cheap computers became communication devices. Often you must redefine the problem: the Model T dropped features and came only in black, but solved what people cared most about, getting from place to place.

207–218

The most useful habit I learned from Michael Rabin: the best way to solve a problem is often to redefine it. You need a big prime? Expensive. How about a number with only a 10-to-the-minus-100 chance of not being prime? Redefining is especially juicy against competitors, who can't follow: you work in plain sight and they miss the danger.

219–223

Making things cheaper is a subset of making things easier-- and now that what we build is so complicated, of making things easier to use. What you want to say about technology is: it just works. How often do you say that now?

224–236

Simplicity takes effort-- genius, even. My mother's new stove had buttons and an LED instead of knobs; I pressed some and it said "Err." You can't say "Err" to the user of a stove. Design so errors are impossible-- they even had a model: the old stove, one knob for heat, one for the timer. It just worked.

237–238

For the average engineer, more options just means more rope to hang yourself. So take almost any big-company technology and assume you could build something way easier to use.

161–238

A startup idea is an idea for something people want. The ways to find one: fix things that are broken, turn a luxury into a commodity, redefine the problem, and above all make things easier to use.

240

Design for Exit

241

Success for a startup approximately equals getting bought.

242

You need some kind of exit strategy, because you can't get the smartest people to work for you without giving them options likely to be worth something.

243

Which means you either have to get bought or go public, and the number of startups that go public is very small.

244

If success probably means getting bought, should you make that a conscious goal?

245

The old answer was no: you were supposed to pretend that you wanted to create a giant, public company, and act surprised when someone made you an offer.

246

Really, you want to buy us?

247

Well, I suppose we'd consider it, for the right price.

248

I think things are changing.

249

If 98% of the time success means getting bought, why not be open about it?

250

If 98% of the time you're doing product development on spec for some big company, why not think of that as your task?

251

One advantage of this approach is that it gives you another source of ideas: look at big companies, think what they should be doing, and do it yourself.

252

Even if they already know it, you'll probably be done faster.

253

Just be sure to make something multiple acquirers will want.

254

Don't fix Windows, because the only potential acquirer is Microsoft, and when there's only one acquirer, they don't have to hurry.

255

They can take their time and copy you instead of buying you.

256

If you want to get market price, work on something where there's competition.

257

If an increasing number of startups are created to do product development on spec, it will be a natural counterweight to monopolies.

258

Once some type of technology is captured by a monopoly, it will only evolve at big company rates instead of startup rates, whereas alternatives will evolve with especial speed.

259

A free market interprets monopoly as damage and routes around it.

241–243

Success for a startup approximately equals getting bought. You need an exit, because you can't get the smartest people without options worth something-- so getting bought or going public, and very few go public.

244–246

Should getting bought be a conscious goal? The old answer was no: pretend to want a giant public company, act surprised at an offer. Really, you want to buy us? Well, for the right price.

247–250

Things are changing. If 98% of the time success means getting bought, why not be open about it, and think of yourself as doing product development on spec? It even gives you ideas: do what big companies should be doing yourself-- you'll be done faster.

251–254

Just make something multiple acquirers will want. Don't fix Windows: the only acquirer is Microsoft, and a lone acquirer needn't hurry-- they can copy you instead. For market price, work where there's competition.

255–257

More startups doing development on spec would counterweight monopolies. Captured by a monopoly, technology evolves at big-company rates while alternatives evolve fast. A free market interprets monopoly as damage and routes around it.

240–259

Success for a startup approximately means getting bought, so make that a conscious goal-- do product development on spec for big companies. Just make something multiple acquirers will want.

261

The Woz Route

262

The most productive way to generate startup ideas is also the most unlikely-sounding: by accident.

263

If you look at how famous startups got started, a lot of them weren't initially supposed to be startups.

264

Lotus began with a program Mitch Kapor wrote for a friend.

265

Apple got started because Steve Wozniak wanted to build microcomputers, and his employer, Hewlett-Packard, wouldn't let him do it at work.

266

Yahoo began as David Filo's personal collection of links.

267

This is not the only way to start startups.

268

You can sit down and consciously come up with an idea for a company; we did.

269

But measured in total market cap, the build-stuff-for-yourself model might be more fruitful.

270

It certainly has to be the most fun way to come up with startup ideas.

271

And since a startup ought to have multiple founders who were already friends before they decided to start a company, the rather surprising conclusion is that the best way to generate startup ideas is to do what hackers do for fun: cook up amusing hacks with your friends.

272

It seems like it violates some kind of conservation law, but there it is: the best way to get a "million dollar idea" is just to do what hackers enjoy doing anyway.

262–265

The most productive way to generate startup ideas is also the most unlikely-sounding: by accident. Many famous startups weren't supposed to be. Lotus began with a program Mitch Kapor wrote for a friend; Apple, because HP wouldn't let Steve Wozniak build microcomputers at work; Yahoo, as David Filo's collection of links.

266–269

You can also consciously come up with an idea; we did. But measured in market cap, build-stuff-for-yourself may be more fruitful, and it's certainly the most fun. Since founders should already be friends, the surprising conclusion: the best way to generate ideas is to cook up amusing hacks with your friends.

270

It seems to violate some conservation law, but there it is: the best way to get a "million dollar idea" is just to do what hackers enjoy doing anyway.

261–272

The most productive way to get startup ideas is the most unlikely-sounding: by accident, building stuff for yourself with friends. The best way to get a million-dollar idea is to do what hackers enjoy anyway.

274

Notes

275

[1] This phenomenon may account for a number of discrepancies currently blamed on various forbidden isms. Never attribute to malice what can be explained by math.

276

[2] A lot of classic abstract expressionism is doodling of this type: artists trained to paint from life using the same gestures but without using them to represent anything. This explains why such paintings are (slightly) more interesting than random marks would be.

277

[3] Bill Yerazunis had solved the problem, but he got there by another path. He made a general-purpose file classifier so good that it also worked for spam.

278

One Specific Idea [blocked]

275

The minority-squared math may account for discrepancies blamed on various forbidden isms. Never attribute to malice what can be explained by math.

276

Classic abstract expressionism is doodling of this type: artists trained to paint from life, using the same gestures but not to represent anything-- which is why it beats random marks.

277

Bill Yerazunis solved the spam problem by another path: a general-purpose file classifier so good it also worked for spam.

274–278

Footnotes on the math behind minority co-founder pairs, abstract expressionism as doodling, and the independent inventor of statistical spam filtering.