Great post. I’m curious to see how long it will take the labels to reach your conclusion #2. Given Pittman’s backing, I’d argue for #1 also, but unless the labels are willing to reduce the rumored rate of ~$0.005 ($5 CPM per song) to something quite a bit lower, it’s hard to believe there’s a viable model here. Pandora can barely make it work at 28% of that rate ($0.0014), and it’s supposedly getting solid CPMs through sponsorship deals (presumably much better than a social network like imeem can garner).

OTOH, if I were a label exec, I would be unwilling to go much lower for on-demand streaming rates, particularly in view of the phenomenal take-up of Pandora on the iPhone. After all, a well-designed Project Playlist app for the iPhone means downloads are increasingly unnecessary for a meaningful segment of the market, so streaming royalties would need to offset potential cannibalization.

A starting point in the math might be: “How many times does the average buyer of a download listen to it over time?” Ignoring proper DCF analysis, if the answer is 120, then one could argue the labels have already priced on-demand streaming correctly (i.e. ~60 cent wholesale margin on downloads / 120 = $0.005).

There’s more to it than this, of course, as many people who would never have purchased the track will in fact stream it. But it’s difficult to estimate this incremental benefit, and also how often people simply use PP for discovery, and then go download it (and how those downloads in turn break out between iTMS/Amzn vs BitTorrent or similar).