Sometimes I wonder, though, if methods like Lean Startup aren't solidifications of this very illusion: that just building something awesome and slightly gonzo is too much fun to be trusted. Lean Startups tend to miss things people don't know they want, or that yield negative results when market-tested with crappy prototypes. It's like "Design by Focus Group". It's probably safer. The success rate is probably higher. But I think you pay for it in loss of craziness. At any rate, startups either don't need to hear this, or shouldn't listen to it.