Idea Generation

Idea Generation. An essay by Sam Altman, from May 2020.

Ideas are a skill

This began as Altman's notes for a 2018 talk in China. Founders' top question: how do I get startup ideas? The runner-up: do you have one for me?

But handing a founder an idea almost never works. Running a startup takes a steady stream of new ideas, not just one.

Y Combinator once tried funding promising founders who had no idea yet. Altman thinks every one of them failed.

Strong founders are full of ideas, about everything. So if you can't find one, treat idea generation as a skill, and practice it first.

Find the right people

First, your surroundings, above all the people. Seek optimists with a feel for the future: creative, open to long-shot plans, and overflowing with ideas.

Great ideas start out fragile. They sound silly at first, so most people never even say them out loud.

So you want people with few filters, around whom a bad idea is safe to say, yours or theirs.

Avoid the world-weary, who belittle ambition. Sadly, that's most of the world. They're anchored in the past; your job is to live in the future.

Live in the future

One method: picture the world 20 years from now, then reason backwards to today. That's a long way off, so radical guesses are fine.

Another: find the tectonic shifts under way. What's changing at the root, and what does it newly make possible? The last big one: mobile, 2008 to 2012.

Altman thinks another one is overdue. And when the world moves that fast, nimble, focused startups tend to beat the big incumbents.

But tell real trends from fake ones. Real: a few people use it a lot. Fake: lots of people use it a little.

Newly possible

Notice anything that's possible this year but wasn't last year. It may be the start of a great startup, especially if next year is too late.

Another good sign: you're certain something is coming, and your only doubt is whether your company will be the one to pull it off.

Altman felt this about Uber. After his first ride, calling cabs was clearly on its way out. Whether Uber would win was far less certain.

Could it be huge?

Early on, ask yourself the question Altman suggests: “Could this be huge if it worked?”

Plenty of ideas are good. But few can get enormous, because most businesses don't build an advantage that keeps piling up as they grow.

For Facebook or Airbnb, that advantage is obvious. Often it's subtler, so look for it early.

Fit, and a final test

Also ask what you're unusually good at. Looking inward rarely answers that; a mentor, or people you've worked with, can.

In fact, Altman now believes founder/company fit matters as much as product/market fit.

One final test: can you explain why most people would call your idea a bad one, and also see what they're missing?

In short

Ideas are a skill: practice generating them instead of waiting to be handed one.

Stay near optimists who make long-shot ideas safe to say.

Live in the future: watch for big shifts and what just became possible.

Ask if it could be huge, if it suits you, and what the doubters miss.

The original is short, and better than this. Read it: blog.samaltman.com/idea-generation