paul graham how to start startup.jpg

Last Thursday, I attended a talk by Paul Graham (YCombinator), Matt DeBergalis (Meteor), Sam Altman (Loopt), and Jeff Byun (Orderahead) at MIT.

Quick Takeaways:

  • Your first goal should be to build something that makes a few people really happy. Then go from there.

  • Do not work on made up ideas - work on ideas for problems you have that scale to others.

  • It is not possible for a new product to be both big and good. A big company can make something big and bad and then have the financial depth to make it better. A startup has to make something good for a few people and then make it big.

Paul Graham (YCombinator):

“Socially up to this point, you've been part of an institution. As you enter the working world, you transition from becoming smarter to becoming a net creator of wealth."

“What's most interesting is not most valuable”

"if you are creating wealth, you can do this for yourself instead of for someone else"

"Your parents are going to be fighting the last war." (And therefore will encourage you to take the brand name approach, eg Google, or in its day, Microsoft.)

“Don't start a startup unless you are suited for it. Why?

  • It’s shockingly hard in ways you haven't experienced.
  • More like being in prison--moral weight grinds you down, and there’s no one to help you.
  • There’s the shocking weight of being responsible for everyone.
  • Practically impossible to know if it will be successful.

You should be frightened of starting a startup. That's why you need a cofounder so you can be out of sync when demoralized so you can cheer each other up. Ideally friends. (NB: Noam Wasserman disagrees, as there is a lower rate of failure when the founders are not friends)

There's no pre-startup class in school, unlike premed. Don’t  plan a startup in college; do what college is good for now, like networking and learning. You can't do a startup while doing something else. It is all-consuming.

Dropbox sucked smart people out of MIT. Facebook sucked smart people out of Harvard. Airbnb sucked all the smart people out of RISD. ;0

You don't have to start a startup right out of school either. The cutoff comes. Cutoff comes by co-founders being tied down, having unvested options in another company, getting engaged to someone in med school.

Startup ideas do not come from a light bulb.

Get together with friends and work on solving a new problem.

A PhD in Machine Learning might work on chat for teenagers, because they don't know the space. This is a huge mistake.

Do not work on made up ideas - work on ideas on that scale for problems that you have. Mark Zuckerberg started Facebook so he can stalk people. Woz started Apple so he could have a computer and then Jobs saw he could sell it.

Jeff Byun (Orderahead)

Don't chase the shiny new ball. Work on something you are passionate about and something you know.

10 smart people will give you conflicting advice. Find your own style.

A founder needs to be able to sell and hustle, more than he/she needs technical skills. Jeff started selling Orderahead without having written a line of code. He hustled and got paying merchants before the app was written.

Matt Debergalis (Meteor)

Be open to pivoting if the need is there. Meteor started in YC as an iPad travel app. With four days to go before Demo Day, the team switched to what they knew - a JavaScript framework, because they saw that all other YC companies had problems prototyping their apps.  

Get your butt where it matters.

Sam Altman (loopt)

It's ok to work at a startup before starting a startup, just don't work at Google. Choosing Google out of college is actually taking the riskier path. You’ve used up three of your most productive years when you had the chance to start something.

Build projects in school that people will actually use (so you get real feedback).

The degree you are willing to succeed in a startup depends on being able to do what you don't like doing.

Show you have nothing to lose.

Advice is useless. All you should care about is building a product that people love.

All startups do things by exception.

You'll learn more in three months at YC than 3 months at school.