September 2017

The most valuable insights are both general and surprising.

F = ma for example.

But general and surprising is a hard combination to achieve.

That territory tends to be picked clean, precisely because those insights are so valuable.

Ordinarily, the best that people can do is one without the other: either surprising without being general (e.g. gossip), or general without being surprising (e.g. platitudes).

The most valuable insights are both general and surprising — F = ma, for example. But that combination is hard: the territory gets picked clean.

The best we usually manage is one without the other: surprising but not general (gossip), or general but not surprising (platitudes).

The most valuable insights are both general and surprising, like F = ma, but that combination is hard — the territory gets picked clean.

Where things get interesting is the moderately valuable insights.

You get those from small additions of whichever quality was missing.

The more common case is a small addition of generality: a piece of gossip that's more than just gossip, because it teaches something interesting about the world.

But another less common approach is to focus on the most general ideas and see if you can find something new to say about them.

Because these start out so general, you only need a small delta of novelty to produce a useful insight.

A small delta of novelty is all you'll be able to get most of the time.

Which means if you take this route, your ideas will seem a lot like ones that already exist. Sometimes you'll find you've merely rediscovered an idea that did already exist. But don't be discouraged.

Remember the huge multiplier that kicks in when you do manage to think of something even a little new.

Moderately valuable insights come from small additions of the missing quality — usually generality, like gossip that also teaches you something.

The rarer route: take the most general ideas and find something new to say. Because they start so general, a small delta of novelty is enough — and small is all you'll usually get. Your ideas will resemble existing ones, but the multiplier on even a little novelty is huge.

Work the most general ideas, where a small delta of novelty suffices; that delta is usually all you get, but the multiplier on it is huge.

Corollary: the more general the ideas you're talking about, the less you should worry about repeating yourself.

If you write enough, it's inevitable you will.

Your brain is much the same from year to year and so are the stimuli that hit it.

I feel slightly bad when I find I've said something close to what I've said before, as if I were plagiarizing myself.

But rationally one shouldn't.

You won't say something exactly the same way the second time, and that variation increases the chance you'll get that tiny but critical delta of novelty.

And of course, ideas beget ideas. (That sounds familiar [blocked].)

An idea with a small amount of novelty could lead to one with more.

But only if you keep going.

So it's doubly important not to let yourself be discouraged by people who say there's not much new about something you've discovered.

"Not much new" is a real achievement when you're talking about the most general ideas.

Corollary: the more general your ideas, the less you should fear repeating yourself. You never say it the same way twice, and that variation is your chance at the critical delta.

Ideas beget ideas, but only if you keep going. "Not much new" is a real achievement with the most general ideas.

Working general ideas, don't fear repeating yourself; the variation each time raises your odds of novelty, and ideas beget ideas if you keep going.

It's not true that there's nothing new under the sun.

There are some domains where there's almost nothing new.

But there's a big difference between nothing and almost nothing, when it's multiplied by the area under the sun.

It's not true that there's nothing new under the sun. Some domains have almost nothing new — but the difference between nothing and almost nothing is huge, multiplied by the area under the sun.

There's almost nothing new in some domains — but the difference between nothing and almost nothing is huge, multiplied by the area under the sun.

Thanks to Sam Altman, Patrick Collison, and Jessica Livingston for reading drafts of this.

Thanks to Sam Altman, Patrick Collison, and Jessica Livingston for reading drafts.

Thanks to the readers of drafts.