To determine really what people want you need to be the proto USER, you need to understand the needs in solving a problem and what the outcome should be. Great

To determine really what people want you need to be the proto USER, you need to understand the needs in solving a problem and what the outcome should be. Great software ideas that stood the test of time were mostly developed by someone that wanted to solve his own problem first. I would start with Knuth and TeX. He wanted to solve his own typesetting problems. I will leave it for others to add more examples (Lotus? AutoCad? HTML?)