How do you avoid copying the wrong things? Copy only what you genuinely like. That would have saved me in all three cases. I didn't enjoy the short stories we had to read in English classes; I didn't learn anything from philosophy papers; I didn't use expert systems myself. I believed these things were good because they were admired.

It can be hard to separate the things you like from the things you're impressed with. One trick is to ignore presentation. Whenever I see a painting impressively hung in a museum, I ask myself: how much would I pay for this if I found it at a garage sale, dirty and frameless, and with no idea who painted it? If you walk around a museum trying this experiment, you'll find you get some truly startling results. Don't ignore this data point just because it's an outlier.

Another way to figure out what you like is to look at what you enjoy as guilty pleasures. Many things people like, especially if they're young and ambitious, they like largely for the feeling of virtue in liking them. 99% of people reading Ulysses are thinking "I'm reading Ulysses" as they do it. A guilty pleasure is at least a pure one. What do you read when you don't feel up to being virtuous? What kind of book do you read and feel sad that there's only half of it left, instead of being impressed that you're half way through? That's what you really like.

Even when you find genuinely good things to copy, there's another pitfall to be avoided. Be careful to copy what makes them good, rather than their flaws. It's easy to be drawn into imitating flaws, because they're easier to see, and of course easier to copy too. For example, most painters in the eighteenth and nineteenth centuries used brownish colors. They were imitating the great painters of the Renaissance, whose paintings by that time were brown with dirt. Those paintings have since been cleaned, revealing brilliant colors; their imitators are of course still brown.

It was painting, incidentally, that cured me of copying the wrong things. Halfway through grad school I decided I wanted to try being a painter, and the art world was so manifestly corrupt that it snapped the leash of credulity. These people made philosophy professors seem as scrupulous as mathematicians. It was so clearly a choice of doing good work xor being an insider that I was forced to see the distinction. It's there to some degree in almost every field, but I had till then managed to avoid facing it.

That was one of the most valuable things I learned from painting: you have to figure out for yourself what's good. You can't trust authorities. They'll lie to you on this one.

In a few days it will be Angel Day, when the startups we funded this summer present to investors. Y Combinator funds startups twice a year, in January and June. Ten weeks later we invite all the investors we know to hear them present what they've built so far.

We call it Angel Day because most of the investors are individual "angels." But there have been an increasing number of VCs, and recently some potential acquirers.

Ten weeks is not much time. The average startup probably doesn't have much to show for itself after ten weeks. But the average startup fails. When you look at the ones that went on to do great things, you find a lot that began with someone pounding out a prototype in a week or two of nonstop work. Startups are a counterexample to the rule that haste makes waste.

(Too much money seems to be as bad for startups as too much time, so we don't give them much money either.)

A week before Angel Day, we have a dress rehearsal called Demo Day. At other Y Combinator events we allow outside guests, but not at Demo Day. No one except the other founders gets to see the rehearsals.

The presentations on Demo Day are often pretty rough. But this is to be expected. We try to pick founders who are good at building things, not ones who are slick presenters. Some of the founders are just out of college, or even still in it, and have never spoken to a group of people they didn't already know.

So we concentrate on the basics. On Angel Day each startup will only get ten minutes, so we encourage them to focus on just two goals: (a) explain what you're doing, and (b) explain why users will want it.

That might sound easy, but it's not when the speakers have no experience presenting, and they're explaining technical matters to an audience that's mostly non-technical.

This situation is constantly repeated when startups present to investors: people who are bad at explaining, talking to people who are bad at understanding. Practically every successful startup, including stars like Google, presented at some point to investors who didn't get it and turned them down. Was it because the founders were bad at presenting, or because the investors were obtuse? It's probably always some of both.

At the most recent Demo Day, we four Y Combinator partners found ourselves saying a lot of the same things we said at the last two. So at dinner afterward we collected all our tips about presenting to investors. Most startups face similar challenges, so we hope these will be useful to a wider audience.

Investors' main question when judging a very early startup is whether you've made a compelling product. Before they can judge whether you've built a good x, they have to understand what kind of x you've built. They will get very frustrated if instead of telling them what you do, you make them sit through some kind of preamble.

Say what you're doing as soon as possible, preferably in the first sentence. "We're Jeff and Bob and we've built an easy to use web-based database. Now we'll show it to you and explain why people need this."

If you're a great public speaker you may be able to violate this rule. Last year one founder spent the whole first half of his talk on a fascinating analysis of the limits of the conventional desktop metaphor. He got away with it, but unless you're a captivating speaker, which most hackers aren't, it's better to play it safe.

A demo explains what you've made more effectively than any verbal description. The only thing worth talking about first is the problem you're trying to solve and why it's important. But don't spend more than a tenth of your time on that. Then demo.

When you demo, don't run through a catalog of features. Instead start with the problem you're solving, and then show how your product solves it. Show features in an order driven by some kind of purpose, rather than the order in which they happen to appear on the screen.

If you're demoing something web-based, assume that the network connection will mysteriously die 30 seconds into your presentation, and come prepared with a copy of the server software running on your laptop.

One reason founders resist describing their projects concisely is that, at this early stage, there are all kinds of possibilities. The most concise descriptions seem misleadingly narrow. So for example a group that has built an easy web-based database might resist calling their applicaton that, because it could be so much more. In fact, it could be anything...

The problem is, as you approach (in the calculus sense) a description of something that could be anything, the content of your description approaches zero. If you describe your web-based database as "a system to allow people to collaboratively leverage the value of information," it will go in one investor ear and out the other. They'll just discard that sentence as meaningless boilerplate, and hope, with increasing impatience, that in the next sentence you'll actually explain what you've made.