Paul Graham - Essays

In 1994 my friend Koling wanted to talk to his girlfriend in Taiwan, and to save long-distance bills he wrote some software that would convert sound to data packets that could be sent over the Internet. We weren't sure at the time whether this was a proper use of the Internet, which was still then a quasi-government entity. What he was doing is now called VoIP, and it is a huge and rapidly growing business.

If you want to know what ordinary people will be doing with computers in ten years, just walk around the CS department at a good university. Whatever they're doing, you'll be doing.

In the matter of "platforms" this tendency is even more pronounced, because novel software originates with great hackers, and they tend to write it first for whatever computer they personally use. And software sells hardware. Many if not most of the initial sales of the Apple II came from people who bought one to run VisiCalc. And why did Bricklin and Frankston write VisiCalc for the Apple II? Because they personally liked it. They could have chosen any machine to make into a star.

If you want to attract hackers to write software that will sell your hardware, you have to make it something that they themselves use. It's not enough to make it "open." It has to be open and good.

And open and good is what Macs are again, finally. The intervening years have created a situation that is, as far as I know, without precedent: Apple is popular at the low end and the high end, but not in the middle. My seventy year old mother has a Mac laptop. My friends with PhDs in computer science have Mac laptops. [2] And yet Apple's overall market share is still small.

Though unprecedented, I predict this situation is also temporary.

So Dad, there's this company called Apple. They make a new kind of computer that's as well designed as a Bang & Olufsen stereo system, and underneath is the best Unix machine you can buy. Yes, the price to earnings ratio is kind of high, but I think a lot of people are going to want these.

[1] These horrible stickers are much like the intrusive ads popular on pre-Google search engines. They say to the customer: you are unimportant. We care about Intel and Microsoft, not you.

[2] Y Combinator is (we hope) visited mostly by hackers. The proportions of OSes are: Windows 66.4%, Macintosh 18.8%, Linux 11.4%, and FreeBSD 1.5%. The Mac number is a big change from what it would have been five years ago.

This summer, as an experiment, some friends and I are giving seed funding to a bunch of new startups. It's an experiment because we're prepared to fund younger founders than most investors would. That's why we're doing it during the summer-- so even college students can participate.

We know from Google and Yahoo that grad students can start successful startups. And we know from experience that some undergrads are as capable as most grad students. The accepted age for startup founders has been creeping downward. We're trying to find the lower bound.

The deadline has now passed, and we're sifting through 227 applications. We expected to divide them into two categories, promising and unpromising. But we quickly saw that we needed a third: promising people with unpromising ideas. [1]

We should have expected this. It's very common for a group of founders to go through one lame idea before realizing that a startup has to make something people will pay for. In fact, we ourselves did.

Viaweb wasn't the first startup Robert Morris and I started. In January 1995, we and a couple friends started a company called Artix. The plan was to put art galleries on the Web. In retrospect, I wonder how we could have wasted our time on anything so stupid. Galleries are not especially excited about being on the Web even now, ten years later. They don't want to have their stock visible to any random visitor, like an antique store. [2]

Besides which, art dealers are the most technophobic people on earth. They didn't become art dealers after a difficult choice between that and a career in the hard sciences. Most of them had never seen the Web before we came to tell them why they should be on it. Some didn't even have computers. It doesn't do justice to the situation to describe it as a hard sell ; we soon sank to building sites for free, and it was hard to convince galleries even to do that.

Gradually it dawned on us that instead of trying to make Web sites for people who didn't want them, we could make sites for people who did. In fact, software that would let people who wanted sites make their own. So we ditched Artix and started a new company, Viaweb, to make software for building online stores. That one succeeded.

We're in good company here. Microsoft was not the first company Paul Allen and Bill Gates started either. The first was called Traf-o-data. I don't know exactly what happened to Traf-o-data, but it does not seem to have done as well as Micro-soft, as Microsoft was called then.

In Robert's defense, he was skeptical about Artix. I dragged him into it. [3] But there were moments when he was optimistic. And if we, who were 29 and 30 at the time, could get excited about such a thoroughly boneheaded idea, we should not be surprised that hackers aged 21 or 22 are pitching us ideas with little hope of making money.

Why does this happen? Why do good hackers have bad business ideas?

Let's look at our case. One reason we had such a lame idea was that it was the first thing we thought of. I was in New York trying to be a starving artist at the time (the starving part is actually quite easy), so I was haunting galleries anyway. When I learned about the Web, it seemed natural to mix the two. Make Web sites for galleries-- that's the ticket!

If you're going to spend years working on something, you'd think it might be wise to spend at least a couple days considering different ideas, instead of going with the first that comes into your head. You'd think. But people don't. In fact, this is a constant problem when you're painting still lifes. You plonk down a bunch of stuff on a table, and maybe spend five or ten minutes rearranging it to look interesting. But you're so impatient to get started painting that ten minutes of rearranging feels very long. So you start painting. Three days later, having spent twenty hours staring at it, you're kicking yourself for having set up such an awkward and boring composition, but by then it's too late.

Part of the problem is that big projects tend to grow out of small ones. You set up a still life to make a quick sketch when you have a spare hour, and days later you're still working on it. I once spent a month painting three versions of a still life I set up in about four minutes. At each point (a day, a week, a month) I thought I'd already put in so much time that it was too late to change.

So the biggest cause of bad ideas is the still life effect: you come up with a random idea, plunge into it, and then at each point (a day, a week, a month) feel you've put so much time into it that this must be the idea.

How do we fix that? I don't think we should discard plunging. Plunging into an idea is a good thing. The solution is at the other end: to realize that having invested time in something doesn't make it good.

This is clearest in the case of names. Viaweb was originally called Webgen, but we discovered someone else had a product called that. We were so attached to our name that we offered him 5% of the company if he'd let us have it. But he wouldn't, so we had to think of another. [4] The best we could do was Viaweb, which we disliked at first. It was like having a new mother. But within three days we loved it, and Webgen sounded lame and old-fashioned.