I just tried to transcribe the Gates interview and discovered it would take me AT LEAST 2 hours and would have a lot of mistakes in it. Plus it doesn't make a lot of sense on paper (e.g. "your you understand")
When the secretaries in my office do transcription they use special hardware. They have a foot pedal that will advance the sound foreward and backwards, without having to take hands off the keyboard. From watching them, it seems like this would be pretty
essential to doing any non-trivial amount of transcription. I have no idea how much this hardware costs, but it might be an obstacle to a cooperative subtitling scheme.