Why Did Talk-to-code Fail for 50 Years?

David Heinemeier Hansson says the last 40 or 50 years of programming kept promising that ordinary people could write software just by talking. Fourth-generation languages were supposed to do it. Lisp was presented that way. Smalltalk was presented as an environment where regular people would build their own applications. None of it arrived in the sense that matters now. What people actually got was Microsoft Access databases and Excel spreadsheets, useful end-user tools and nothing like the output he is shipping.

He has been programming for 25 years, and the change he trusts is the one he can see: the output, the acceleration, the quality, and then the fact that it ships. Someone who has not seen the current quality will still hear the claims and think the speaker sounds a little nutty, because everything up to that point said the same promise had already failed.