I agree it is very confusing to figure out the exact intent from the text. In the model i suggested you would look up and input a code that referred to a song,

I agree it is very confusing to figure out the exact intent from the text. In the model i suggested you would look up and input a code that referred to a song, and it would start the recorder when that song was played (working like traffic report alerts meets TiVo). No spamming. By broadcasting real-time streams via TV you would get a bandwidth ~400 songs/day/channel (a big selection for the time). To me 'subscribing to interest categories' would be exactly like subscribing to a cable channel; and it would give you the legal right to that group of songs.[0] It seems 'TV cable' was definitely a transmission method (very unclear if the phone was an alternative data vehicle or just for control), and AFAIK there was no way of doing anything on-demand with TV cable. To get similar quality of audio via the phone would have taken ages[1], tying up the phone for a long time (+ associated costs). The only problem with the streaming model is that you have to wait for a song to come around. But you would be able to get a newly released song right away on it's first play (and nothing was on demand at the time) so it wouldn't be a big issue IMO. Occam's razor suggests the stream concept is most likely; it fits the existing models and technology much closer. It would have been technically simple to implement and vastly cheaper. -- 0. I think this is the real 'innovation' in this, rather than delivery method, which i'm not certain Zappa was 100% on. 1. i remember 128KB songs took ~15 minutes on 56K, so we are talking over an hour, even if we ignore the lack of MP3, processing/ temp storage requirements. Remember he talks about quality. Don't forget your landline was the only real-time contact you had with people who weren't in your road.