Paul Graham: The Hardest Lessons for Startups to Learn

This is part of my set of notes from the Startup School 2006 sessions at Stanford.

Paul Graham of Y Combinator shares the counterintuitive things startups learn over time (including his experience with ViaWeb) in the vain hope that he can avoid having to keep repeating himself.

#1: Release early

  • Get a version one out fast, improve based on user feedback. Don’t release something full of bugs, but something minimal. Pays to get version 1 done fast. It’s the right way to write software – seen a lot of software dies because it was released too slow.
  • If you don’t release early, you won’t know who your users are – as you don’t know your users, you don’t know what they will like.
  • WuFoo release their form-builder early – they haven’t even release the engine underneath it. Many Linux users complained early about too much Flash, so they removed it
  • Initial release shakes out the bugs earlier
  • Makes you work harder – when you’re working on something that’s released, it makes problems be critical to fix
  • Getting into the English channel gradually is going to take a long time

#2: Keep pumping out features

  • Don’t make application more complex; one quantum of hacking that makes the user’s life better
  • Improvements beget improvements – the more ideas you implement, the more ideas you’ll have. Not just a good way to get development done – also a form of marketing: users love sites that constantly get better. Will like it even better if you’re improving in response to their input, because they’ll recognize that you’re listening – thus generating fanatical devotion to the application.
  • If your product seems finished, there are two possibilities: it is finished, or you lack imagination.

#3: Make users happy

  • You can’t force anyone to do anything, to do deals with them. A startup sings for its supper, otherwise they die.
  • If users pick you up, no competitor can keep you down
  • Vast majority of users will be casual users, and it’s for them that you need to design
  • Think of all the links you’ve visited on the web; a vast majority of them have led to something lame.
  • Do things to make people pause (from hitting the back button):
    • Concisely explain what your site does: You have to be able to explain in one or two sentences what you do. Not just for users, but also reporters, partners, press. You probably shouldn’t start a startup if it can’t be explained in one or two sentences.
    • Give users something immediately up front: The front page is probably the only page most users will see. Show, don’t tell – just like as in writing fiction.
  • The job the site is conversion – convert visitors to users; if you have decent growth, you’ll win

#4: Fear the right things

  • Startups right to be paranoid, but fear the wrong things. Disasters are normal in a startup. They won’t kill you unless you left them.
  • “What if Google builds the same thing” – most startups worry about large companies. Don’t fear the established players, fear instead the other startups. Like you, they’re cornered animals. Don’t fear what people could be doing, fear what people are actually doing. No matter what you’re doing, there is someone else doing it.
  • Competitors are not the biggest threat – internal disputes, ignoring users. Almost everyone’s initial plans are broken.

#5: Commitment is a self-fulfilling prophecy

  • Would like to believe that Viaweb succeeded because they were smart, not their commitment
  • What students lack in experience, they make up in determination
  • You can lose quite a lot in the brains department, and it won’t kill you.
  • There’s always a disaster – if you’re inclined to quit, there’s always a reason at the ready
  • If you lack commitment, it will seem that you’re unlucky. If you determined to stick around, people will have to pay attention to you. Y Combinator mistakenly has funded groups who have decided to “try this startup thing for three months and see if something great happens”.
  • You have to be the right kind of determined; determined, but flexible, not stubborn.

#6: There’s always room for new stuff

  • Example: Suggested to founder to add social networking component – founder responded that the social networking area was pretty much played up.
  • The reason we don’t see the opportunity is that we adjust to the way things are, not the way they should be
  • There’s no limit in the number of startups. People don’t argue there’s no limit on the small number of large, slow moving companies, why should it be different for the large number of small, fast moving startups?

#7: Don’t get your hopes up

  • Founders are naturally optimistic
  • It’s OK to be optimistic about what you can do, but you should assume the worst about machines and other people.
  • Things change suddenly, and often for the worst.
  • If you’re doing a deal, just assume that it won’t happen. Then, when things work out, you can be pleasantly surprised.
  • Not said to prevent people from being disappointed, but to avoid leaning company towards something that’s going to fail. Example: Don’t stop looking for VC deals just because you’ve got one on the table. Deals are dynamic, there’s not a single point where you shake hands and it’s over.

Tim O’Reilly: How to Think About the Future

This is part of my set of notes from the Startup School 2006 sessions at Stanford.

Unfortunately, I returned from lunch a bit late and only caught the tail end of Tim O’Reilly’s presentation; hence, my notes are a bit sparse.


  • “Find a parade and get in front of it”
  • For example, when launching the Whole Internet User’s Guide, O’Reilly used it to tell the world “the Internet is coming” as opposed to trying to simply hawk the book. Sent the book to Congress members to inform, not to peddle.
  • Amazon One-Click Patent: Know who your friends are – used Scripting News, Slashdot, others to push the agenda against Amazon 1-click

Asymmetric Competition

  • Figure out where there’s an industry that right for disruption – look at how Craig Newmark is hurting the newspaper industry. Craigslist is at number 7 on the list of top sites on the net in 2005 – 18 employees versus 9,000 of Yahoo (#1 on the same list)!
  • The best way to be ahead of the curve is to be working on something that you’re passionate about