Paul Graham at YC Startup School 2008

I sometimes re-listen to old talks by I find particularly helpful. While this talk was technically about why to make benevolent projects, I think Paul Graham is actually explaining the process of coming up with a startup idea worth pursuing. ( ) I wrote out the transcript here and organized the content so people can skim and google it easily.

Make something people want

About a month after we started Y Combinator, we sort of stumbled upon the phrase that became our motto.

We have changed our minds about a lot of things since then, but if I were choosing again today, this is still the one I would choose. I don’t think you can get it down to three words.

Don’t worry too much about the business model at the beginning 

The second most commonly thing we tell startups is to not worry too much about money. 

Not because it’s not important, but because it’s so much easier than making something great.

A couple weeks ago I realized that when you put these two ideas together you get something strange.

Make something people + Don’t worry too much about money = Nonprofit

When you make something people want and don’t worry too much about making money, what you’ve got is a description of a nonprofit.

When you get a weird result like this, there’s two possibilities. Either it’s a bug or a new discovery.

Either businesses are not supposed to be like charities and you’ve just proved by reductio ad absurdum that one or both of the principles you started with is false.

Or we have a new idea. I think actually it’s the latter.

When I thought of this a whole bunch of other things fell into place.

For example, Craigslist. 

It is not a charity, but they run it like one. They are amazingly successful. I haven’t actually done this math, but I suspect that they make more money per employee than Google does. 

When you look down the list of the most popular websites, the number of employees at Craigslist looks like a misprint. That’s very impressive. 

They don’t make as much money as they could, because they don’t concentrate that hard on making money.

But most startups would love to trade places with Craigslist.

In Patrick O’Brian’s novels, his captains are always trying to get the weather gauge of their opponents. They are trying to stay upwind of them. If you’re upwind, you decide when and if to engage the other ship.

Craigslist is effectively upwind of enormous revenues. 

They would have to face some challenges if they wanted to make more money, but they’d be like spin challenges. 

Not the kind of challenges you face when you’re tacking upwind. Trying to force a crappy product on ambivalent customers by spending 10 times as much on sales as you do on development.

I’d rather be in Craigslist position. 

I’m not saying that startups should try to end up like Craigslist. They are a product of fairly unusual circumstances. However, they are a really good model for the first phases.

Google looked a lot like a nonprofit in the very beginning. They didn’t have ads for the whole first year.

Google at year one was basically the limit in the calculus sense of what a nonprofit index of the web would have produced. 

Back when I was working on spam filters, I thought of making a sort of a free web-based email service with really good spam filters.

Not because I wanted to start another business, but just because it would be a good idea. But the more I thought about this project, the more I realized it would actually have to be a company to be done right. Since I didn’t want to start another company, I didn’t do it.

it was a realization to me that there were ideas that to be done right have be embodied as profit-making companies. 

There are often companies that claim to be benevolent.  But there are also purely benevolent projects. If you did them right, they would have to look like companies. I think if you had a purely benevolent project to index the web,  it could not really come out that much better than Google did.  

Notice the pattern here.  Whichever direction you start from, you end up at the same place.

Very successful startups often behave like nonprofits. When you think about nonprofit ideas, they make good companies surprisingly often. 

How wide is this territory? Would all good nonprofits make good companies? 

Possibly not. 

One reason Google is so valuable is that their users have money.

If you behave benevolently towards people who have money and make them fall in love with you, then you can probably get some of it.

Could you make money by behaving benevolently to people who have no money? 

You might be surprised.

Could you make money by trying to cure a deadly but unfashionable disease like malaria? 

Surely that wouldn’t work as a start-up idea.

Something like malaria must really be sort of a weight on the productivity of a region. Maybe if you could lift that weight off the people, you could benefit from the growth that would result. 

One way to tell how far an idea extends is to ask yourself at which point you’d be prepared to bet against it. 

The idea of betting against benevolence is really frightening. Exactly the same way as predicting that some sort of technology could never be developed.

You are just asking to be made a fool of when you do that kind of thing because these are such powerful forces.

Powerful forces are a good thing to have on your side in a start-up.

Initially I thought maybe this principle only applied to Internet startups. It applies to Google, but what about Microsoft?

Surely Microsoft isn’t benevolent. But when I think back to the very beginning, they were compared to IBM.

They were like Robin Hood when IBM produced the IBM PC. IBM thought they were going to make money by selling people hardware at really high prices. Microsoft had another idea. By gaining control of the PC standard, they enabled any manufacturer to compete in this market.

Prices plummeted and a lot of people got computers who otherwise could not have afforded them. 

It is just the sort of thing you would expect Google to do well. Microsoft isn’t so benevolent now. When one thinks of what Microsoft does to users, all the verbs that come to mind begin with F.

Like frustrate and foist. 

Yet it doesn’t seem to pay. Their stock price has been flat for years. Back when they were good like Robin Hood, their stock price went up like Google’s. 

Could there be a connection? You can see how there might be. 

When you are small, you can’t bully users. You have to charm them. 

You can’t say “This is Vista and we are not going to give you XP anymore. You’re going to take it and like it!”

When you’re big you can maltreat them at will and you tend to because that’s less work. 

You grow big by being nice, but you can stay big by being mean. 

Just ask the record companies. You get away with it till the underlying conditions change and then your victims all escape. 

Don’t be evil may actually be the most valuable thing that Paul Buchheit made for Google. More valuable than Gmail. It may turn out to be an elixir of corporate youth. 

I’m sure they find it constraining. But think how valuable it will be if it saves them from lapsing into the same laziness that Microsoft and IBM did. 

The curious thing is this elixir is freely available to any company that wants it. Any company can adopt “Don’t be evil” as their motto.

Google doesn’t have a any kind of IP protection on this one. The only catch is that people will hold you to it.

I don’t think you’re going to see record labels or tobacco companies taking advantage of this. 

There’s a lot of evidence that benevolence works. How exactly does it work?

One of the advantages of investing in a huge number of startups is that you get a lot of data about what happens inside them. 

I figured out that there’s

Three ways the benevolence will help your startup:

It improves your morale 

It makes other people want to help you. Above all, it helps you be decisive. 

Morale is tremendously important to a startup. Morale is so important, that it is almost enough to determine your success alone. 

You will hear people describe startups as an emotional rollercoaster. One minute you feel like you’re going to take over the world, the next minute you feel like you’re doomed. 

The problem with feeling that you’re doomed is not just that it makes you unhappy. 

I could deal with that. The problem is that it makes you stop working. The downhill parts of the emotional rollercoaster are actually more of a self-fulfilling prophecy than the uphill parts. 

If feeling like you’re going to succeed makes you work harder that will probably improve your chances of succeeding.

But if feeling like you’re going to fail makes you stop working, that practically guarantees that you will fail.

Here is where benevolence comes in. 

If you feel like you’re really helping people, you will keep working. Even when you feel like you are doomed. Most of us have some amount of natural benevolence. The mere fact that someone needs your help makes you inclined to help them. 

Let’s say you start built something. The kind of startup where users come back every day. Basically you build yourself a giant Tamagotchi. 

You have made something that you need to take care of.

Blogger is a famous example of a startup that went through very low lows and survived. At one point, they ran out of money and everyone left. The founder, Evan Williams, came back the next day and he was the only one there. 

What keeps you going in a situation like this? 

One thing that must have kept him going was simply that users needed him.

Beneath everything else. Like the hope of making money from a startup. Or the fear of looking like a failure to your friends. You have this server with all these users that need you. You know the response time has to be good. You can’t let it just die.

There are many advantages of launching quickly. The most important may be that as soon as you launch, the tamagotchi effect kicks in. 

Once you have users to take care of you’re forced to figure out what they want and that information is very valuable. 

One of the unexpected side effects of helping people is that it makes you more confident. That turns out to be very helpful with investors. 

One founder told me that at one point the two of them had sat down and discussed if what they were building was important. So important that they were going to build it no matter what. Even if they had to go back to live to move back to Canada and live in their parents basements.

They kept going to investor meetings. But the funny thing was once they stopped caring about what investors thought of them, investors got a lot more interested strangely enough. 

They could tell that these guys didn’t need them. They were going to do this startup with them or without them. That’s what investors are looking for. 

That’s even what we’re looking for even though we’re so you know on the benevolent end of the spectrum. 

We are looking for startups that would do better with us. Not startups that would die without us. 

Frankly, if they’re that close to dying, they’ll probably die with us too. If you’re really committed and your startup is cheap to run, you become very hard to kill. 

This should be your model for a startup. Not Webvan. 

Craigslist and cockroaches.

That is really just about the most valuable quality you can have in a startup. Because practically all

startups, even the most successful ones, come close to death at some point.

In fact, you’re lucky if it’s only one point. If doing good for the world gives you a sense of mission that makes you harder to kill, that alone may more than compensate whatever you lose by not choosing more selfish projects.

It makes people want to help you

That too, is an inborn trait that probably goes back before humans were really human. One of the startups we have funded, octopart, is currently locked in a classic battle of good versus evil. They are a search site for industrial components. Sounds boring, but a lot of people need to search for industrial components. Before octopart, there was no good way to do it that. 

It turns out that was no coincidence. Octopart has built the right way to search for components. Users like it. They’re growing rapidly. 

And yet, for most of octoparse life, the biggest distributor digikey has been trying to force them to take their prices off the site.

Octopart is sending them traffic for free. Customers who buy things. And Digikey wants to force them to stop sending them people.


Because their current business model depends on overcharging people who have incomplete information about prices. They don’t want search to work. The Octoparts are the nicest guys in the world. They dropped out of the PHD program in physics at Berkeley to do this.

They just wanted to solve a problem that they were encountering in their own research. Imagine how great it would be if all the world’s engineers could do part searches online. When I hear that a big evil company is trying to stop them in order to keep search broken, it makes me want to do everything I can to help them. 

It just made it makes me spend more time on them than I do on most of the startups we funded. It just made me spend several minutes telling an entire roomful of people how great they are.


Because they’re good guys and they’re doing a good project and I want to help them.

If you’re a benevolent, people will rally around you. Investors. Customers. Other companies and

potential employees. 

In the long term, that may be the most important. The potential employees. 

I think everyone probably knows that good hackers are much more valuable pound for pound than mediocre ones.

If you can attract the best hackers to work for you (like Google) that is a huge advantage. 

The thing about the very best hackers is they tend to be idealistic. 

They’re not desperate for a job. They can work anywhere. 

They get to pick and choose. They’re going to choose the kind of things that make the world better. 

So if you have a project that makes the world better, you’re like a magnet for all the smart people to come and work for you. 

The most important thing of all is that it acts as a compass

One of the hardest parts about a start-up is that there tend to be thousands of things to do and only two or three of you to do them. 

How do you decide what to do?

Well you got to decide because the one thing you can’t do is vacillate. 

Here’s the answer. 

Just do whatever is best for your users. 

You can hold on to that like a rope in a hurricane and it will save you if anything can. You just pull on that rope and it will guide you through everything you have to do. 

Figure out what they want. Make them happy. Keep doing more of that.

It’ll even give you the answer to questions it doesn’t seem to be the answer to. Like how to raise money from investors. If you’re a good salesman you try to just talk them into it.

But you know most hackers aren’t that great salesman. The straightforward approach is just to go to investors through your customers.

If you make something that users like enough to tell your friend about, that is a recipe for exponential growth.

Being good is a particularly useful strategy for making decisions in situations where things change fast. 

It’s stateless. It’s like telling the truth. The trouble with lying as I discovered when I was a kid is that you have to remember everything you’ve said in the past in order not to say anything that will contradict yourself. 

If you tell the truth, you don’t have to remember anything. That is a really useful property in domains where things change fast. 

For example, Y Combinator now has invested in 80 startups. 57 of which are still alive for some definition of alive. 

The rest have died or merged or been acquired.

When you’re trying to advise 57 startups at once, it turns out that you have to have a stateless algorithm.

You can’t have ulterior motives because you can’t remember them. Our rule is just try and do whatever is best for the startups. 

This may not be optimal for us, but it’ll be reasonably close. It’s not because we’re particularly benevolent, but it’s just the only algorithm that works on that scale.

It’s the same reason probably that Google doesn’t munge search results. Once they start like messing around with search results, how could they possibly scale that. 

When you give a talk telling people to be good you seem to be claiming to be good yourself. 

In fact, that’s probably the reason most such talks happen. I want to say explicitly that I am not a particularly good person.

When I was a kid, I was firmly in the camp of bad. The way adults used the word good seemed to be synonymous with the word quiet. I grew up very suspicious of it.

You know how there are some people that people talk about and say, “he’s such a great guy”? People don’t say that about me.

The best I get is “He means well.”

I’m not claiming to be good. I’m like about average. At best, I speak good as a second language. 

So please, don’t tune this out. The way you’d automatically tune out someone standing up on a stage and telling you to be good. I’m not suggesting you be good in the usual sanctimonious goody two-shoes way.

I’m suggesting it because it works.

it will work not just as a corporate statement of corporate values. 

It will work as a guide to corporate strategy. It will work as a design spec for software. 

Don’t just not be evil, be good.

Q & A

How many startups have mottos?

Xobni: There was a book written once here at Stanford called “Built to last”. 

It was looking at companies that make it really big and are successful and do good things. One of the things they had in common was that they all had sort of credos or mission statement. It seems like you have one at Y Combinator. How many startups that you found or fund have credos? 

PG: How many startups have mottos?

Xobni: Google does. doesn’t it?

PG: I think that’s probably a good idea. I think you may be onto something. Maybe we should try and encourage people to make up mottos. The thing is they barely have time though. If making up a motto makes you delay launching for a couple hours, then it’s probably not worth it. Maybe after launching people can come up with mottos. 

Xobni: Inbox backwards didn’t work well. 

PG: You could make a motto out of that…How about organizing the world’s information?

Xobni: Good idea we’ll think about it. 

Should you even go to college?

Guy 1: You recommend doing an undergraduate degree as opposed to a graduate degree, saying more wealthy people tend to have undergraduate degrees because they spend more time making money. The general idea there is – is it better to start a startup first? Should you even get the undergraduate degree? 

If you have a good idea, should you just like run with it and get the heck out of there and start making money?

PG: It might not be such a great idea as you think. 

The question is, you have a good idea and you’re in college. Should you stay in college or should you just like go do the startup right? 

Because a lot of successful founders never finished college or never went to it.

Stuff like that is a good point, but making money and doing startups is not the only thing in your life. 

You can have a lot of fun in college. Maybe even learn something.

There’s no huge rush. 

You could just hang out in college. Learn stuff. Meet people. Have friends, 

That idea that you think is so great that you want to drop out of school might not turn out to be that great. It’s certainly even more likely that if you wait till you finish school there’ll be some other idea you’ll think of.

It will be even better. 

I wouldn’t hurry. Don’t hurry. 

Guy 2: I’ve got to disagree with you firmly and vigorously. 

Most of the companies we work with. 

Apple. Cisco. Nokia. 

We have a lot of very intelligent people. Very well educated. The top advisors for Apple and others said (And I agree, I’ve seen it over 15 years) 

The people who get undegraduate degress and graduate degrees work for people who drop out and do the business and execute

PG: Yeah maybe…

Guy 2: No. Uniformly. 

At the top level. Uniformly.

We can’t be too down on school. 

It’s good to go back and give some money and learn by teaching other people. But uniformly. 

Because school is always going to be there. You will have better expertise and experience to apply the research. 

You will have the passionate thing. Follow your passion.

Screw the parents. Screw everybody else. Follow your passion.

PG: Maybe.

How do you come up with startup ideas?

Person: I have a question regarding ideas. Usually when I think about ideas. Looking at stuff that’s happening today. You almost end up coming up with ideas that somebody has already ran with or already is halfway building a prototype for. If you’re going to think today, you’re going to come up with something that’s already done.

How do you think about stuff that is going to happen in three or four years?

For example, how do you think about what’s going to happen in Web 3.0? What kind of stuff you need to know? What are the some of the materials that you read?

Is it just being in the right company? 

PG: I think the best way to look for idea is to first look for problems. 

Don’t try and think of solutions. Just try and find good problems and let them drive the solution.

Just look for stuff that seems broken. 

Components search is broken. Octoparts noticed it because they had to buy a lot of components in their research. They did Applied Physics with components. 

If they’d been Theoretical Physicists, they would have just been like starting a website to buy legal pads.

How do you invest?

Q: When deciding which startups to fund, does this benevolence factor really come into play?

PG: It is.

We found that we try to fund people who do things we like. Actually, it was really self-indulgence at first.

Q: What other factors – if you want to just summarize? 

Besides benevolence?

Really, we look for people who seem like they’re going to succeed. 

There’s so many reasons. I mean it would seem obvious to any other investor that they should fund startups. that are going to succeed.

If you’re playing a game of soccer, you want to kick the ball into your opponent’s goal. Not the one behind you. 

We really just try to fund the kind of founders we like and we think are going to succeed. 

It turns out that they tend to do benevolent things. 

Being good works.