let's say 5 people start watching a 20 minute youtube video. they do this with a gap of three minutes. after 15 minutes there are 5 people watching a 20 minute youtube video. now the network can stream in in real time, or at whatever bitrate, or just send the whole darn thing in the first 20 seconds. this means every three minutes and 20 seconds, the users gets the file, and the network is done communicating with them. in the first scenario, at the end of fiteen minutes, the network is tied to five devices, and sending them the video in teh diff bandwidth of each device. the benefit of the last scenario is the amount of devices the network is tied to, and the amount of time it is doing this for. this option is not considered because of speed limits, without these, if just the content is billed, then the network can respond when and where there is demand, instead of responding slowly over time to all the demand, at all those specified rates. Podcasts are at times preferred to streaming for this reason.
In any case why not option 3? When have we ever not voted with our wallets? This might make net cheaper for us