Paul Graham Keynote


  • Paul Graham
  • YCombinator

Silicon Valley

  • The center of SV moves around the peopl who make the next generation of stuff * So, this room is right now
  • The frightening-ness of big startup ideas
  • List of 7 gigantic startup ideas
  • Scary, maybe I should do that recipe site instead

Next Google

  • Start next Google
  • Microsoft lost their way when they got into the search business
  • Google has been getting into the social network business
  • Nostalgic for the right answer from google * Seems based on Scientologist: “What’s true is what’s true for you”
  • Find tiny idea that turns big idea * Dinosaur egg
  • Search engine for top 10k hackers
  • Make the search engine the one you want
  • Don’t worry about something that constrains you in the long term

Replace Email

  • Any big idea has a bunch of people nibbling around it
  • Not designed to be used the way it is now * Bell labs “Want to go to lunch?”
  • Now a shitty todo list
  • Tweaking the inbox is not enough
  • Todo list protocol insteayd of messaging protocol
  • Sending emails to yourself
  • Want to know what they want you to do
  • When does it need to be done?
  • Whenever powerful people are in pain, that is the way to make lots of money
  • Gmail has gotten painfully slow
  • People will pay for faster email

Replace Universities

  • claps
  • Last couple of decades, universities seem to have gone down the wrong path
  • Expensive country clubs

Kill Hollywood

  • Hollywood was slow to embrace the internet
  • Internet beat cable
  • Bolted an iMac to the wall, found it better than a TV
  • TV seemed like it was designed by the same people who designed the thermostat
  • How do you deliver drama via the internet?
  • You kind of want to know what you’re going to get with a show

A New Apple

  • If Apple won’t make the next iPad, who will? * Empirically, it’s none of the incumbents
  • It will be a startup * Not crazy, Apple did it
  • Steve Jobs showed us what one person can do
  • “Steve Jobs unrolled the future like a carpet”
  • The next CEO might not live up to Steve Jobs, but doesn’t need to * Just needs to be better than HP, Samsung, Motorola

Bring Back Moore’s Law

  • Circuits are going to get twice as dense, not twice as fast
  • Hardware would just solve software’s problems
  • Need to rewrite it to be parallel
  • It would be really great by making a lot of CPUs look like one
  • The most ambitious is to do it automatically via a compiler * “Sufficiently smart compiler”
  • If not impossible, expected value is really high
  • Less ambitious is to start from the bottom * Build programs out of more parallizable lego blocks * Programmer still does a lot of the work
  • Middle ground is a semi automatic weapon * Looks like a sufficiently smart compiles, but there are humans in there
  • Make a market place, let people do it * Maybe make bots that will do it

Ongoing Diagnosis

  • Imagine the ways we will seem backwards to people in the future
  • Seem barbaric to wait for symptoms to be diagnosed
  • Bill Clinton had to wait for arteries to be 90% blocked to find out
  • Launch fast and iterate may not work for medical. * Work on pigs first * Sausage company on the side
  • The medical profession will be an obstacle to this
  • Doctors are alarmed to look for problems that aren’t there
  • If you start testing people all the time, you may get a lot of terrifying false alarms
  • Think this is an artifact of current limitations
  • Going against medical tradition

Tactical Advice

  • For big problems, don’t make a frontal attack
  • “Are we there yet?”, Haters
  • Notice that you replaced email when it’s done
  • Start with small things, let them get big * Facebook
  • Maybe big ambitions are a bad thing * The bigger they are, more likely to be wrong * Don’t identify, just think there is something out there * When the opportunity comes to move, move there
  • Blurry vision may be better