Get it done, then make it good

Legend has it that one of the ways things worked on Tomorrow's World was that a researcher, when they thought they might have a story, would immediately write a script. Only when that was done would they start making calls and trying to work out what the story was. At which point they'd update the script. Iterate until done.

While borderline barking, one of the points of working like that was that, at any given stage, there was a script that a director could pick up and go and shoot. Early in the process it was likely to be nonsense, but at least there was something workable -- and it helped focus the iterations. It's a dangerous approach, because it's easy to get trapped in a blind alley, but nevertheless it's still sometimes a good way of working.

This article about Adobe's development model for Photoshop CS3 sounds remarkably similar. Interesting.