+10000. This is a trap that I've seen way too many hackers and entrepreneurs fall into.<p>Get out of the room. Talk to <i>actual</i> people. Find out about their processes. Understand their pain points.<p>Then, build, iterate, and build some more.
This is especially true in the ed-tech startup world. Many entrepreneurs are creating applications for teachers without ever speaking to a teacher or school administrator. They have experience with receiving education and self-education, but they don't know how teachers use technology.<p>Teachers and administrators have to manage multiple learning styles in a classroom of ~20 kids, and what works for the kid who will grow up to be an entrepreneur doesn't always work for the kid who will grow up to be a historian or accountant. As such, the initial abstraction of projecting our own educational experiences onto entire classrooms generally ends up flopping.
I don't think you need to be an expert glassmaker before you go and find out if there's a market for "glass services"(if that's what you are interested in). It's enough to go out and talk to glass owners and find out if they have a need for an expert to maintain or restore their glass and if they do, what they will pay. If they do, find an expert glass maker and hire him or her. Also your interviews might help with what kind of "expert glassmaker" you need.
I disagree. There is not a one-size-fits all way to start a business. Some businesses need big customers such as the cathedral example. But other businesses -- the ones Paul Graham likes to write about -- productize their service, and the product is an investment. If no one wants the product, the investment was lost, but if you did it right, the cost was minimal. And usually you can re-use what you built and pivot into something else.<p>The great thing about this industry is that so much is virtual. You can now work from anywhere, at different times, and still get things done, if your tools are good: Skype, Google Docs, Mercurial, Redmine, what have you.<p>Things can get prototyped quickly on a small budget. You can launch your MVP and iterate until you achieve decent growth through viral channels. You can get advertising to offset your hosting costs, or do something more interesting like freemium. You may license your tech or sell it to bigger players. This is the essence of the free model.<p>Nothing wrong with this. It's just a bigger risk and bigger reward.<p>Not to mention that you can now sell on app stores, and make $$$ for your company.