newsfilter.io
Conference Presentation, Fireside Chat

A Conversation with Paul Graham - Moderated by Geoff Ralston

  • Paul Graham founded Y Combinator in 2005 after his previous company, ViaWeb, was acquired by Yahoo in 1998.
  • ViaWeb became the first "web app" (software running on the server, accessed via browser) out of a desire to avoid learning how to write Windows client software.
  • Graham initially thought updates could be sent via email, but pivoted to using HTTP to allow browser-based control after realizing he could deploy code to the server.
  • The core motivation for ViaWeb was not the benefits of web apps, but specifically the ability to avoid the "pain" of learning Windows development.
  • Graham advises that promising startup ideas often feel "implausible" or "naughty" (taking advantage of something that seems like it shouldn't work) at the outset.
  • Startup outcomes are highly indeterminate; Graham notes that even with perfect selection, a batch of 150 startups might only yield five that become "giants" due to the power law distribution of success.
  • Graham defines a specific type of "laziness" as a productive force: a refusal to perform "gratuitous schleps" (like installing software on client machines) leads to elegant, scalable solutions.
  • Graham's co-founders at ViaWeb were Robert Morris and Trevor Blackwell, both of whom were exceptional programmers but difficult to work with.
  • Robert Morris was a "truculent" founder who was eventually convicted under the Computer Fraud and Abuse Act for creating the 1988 Morris Internet Worm, which was motivated by curiosity rather than malice.
  • Morris was expelled from Harvard for reconnecting the university to the ARPANET; Y Combinator later used this "expelled for working on a project" criterion as a recruiting tactic.
  • Trevor Blackwell was recruited because he was "super smart" and capable of rewriting the entire codebase (Smalltalk) in two weeks, proving he was a "super productive hacking monster."
  • Graham suggests a founder selection heuristic: hire people you trust, and then hire people they trust, as trusted individuals can accurately judge intelligence even if they are poor at judging trustworthiness.
  • Graham argues that startups are "counterintuitive" and that founders must ignore their gut instincts, which often push them toward easy but wrong choices.
  • Y Combinator's mantra "Do things that don't scale" advises founders to perform manual, artisanal work for early customers to learn directly, rather than automating immediately.
  • Graham states that for early startups, the primary value of manual effort is that it forces the founder to learn what the customers actually need.
  • Most startups fail due to poor execution by founders, not competition; Graham notes that being killed by a competitor is extremely rare (approx. 1 in 1,900 cases).
  • Founders should view competitors as runners in a different lane in a 100-meter race: the best solution is to run as fast as possible regardless of what others are doing.
  • Determination is significantly more important than intelligence for startup success; a moderately smart but determined founder will likely succeed over a brilliant but undetermined one.
  • A founding team is critical for morale; sole founders face high risks of giving up because there is no one to "cheer them up" when the venture encounters inevitable struggles.
  • Graham advises founders to launch as soon as they have a "quantum of utility"—meaning at least one person is glad the product exists.
  • Founders often delay launches because they are embarrassed by shipping unfinished products, but "if you are not embarrassed by what you launch, you have launched too late."
  • When choosing between building what customers "want" and what they "need," founders should prioritize what customers will actually pay for (e.g., unhealthy food tastes better than healthy food).
  • Graham warns that high school students should not start companies unless they are 100% committed, as this period is crucial for human growth and exploring options.
  • Regarding pricing, founders should set a price that attracts customers to learn from them, as prices can be lowered later without complaint, or raised by grandfathering existing users.
  • Raising excessive capital too early is dangerous; it creates a "gravitational effect" that forces spending and leads to hiring more people than necessary.
  • Graham notes that he generally avoids funding "Steve Jobs" types (arrogant founders) early on, as being "nice" and benevolent is a better predictor of long-term success in the YC portfolio.
A Conversation with Paul Graham - Moderated by Geoff Ralston — Summary