I think that the biggest reason why P2P desktop applications have fallen by the wayside is mentioned. P2P applications are at least an order of magnitude more complicated than client-server apps (which are another order of magnitude more complex than single-user apps).
For that much complexity, P2P doesn’t offer a whole lot in return. Sure you can leverage your users’ bandwidth and hard drive space, but if you have any sort of profit mechanism you can probably afford S3. Bandwidth is a lot cheaper than it was in 1999. Plus the bandwidth you get probably sucks, due to the asynchronous nature of consumer broadband.
The biggest benefit of P2P I see is that you can avoid directly breaking laws that curb what you can publish. While there are noble and ignoble reasons to do so, I think it has a limited value that the existing players have mostly exploited.
There’s also the question of whether the value of P2P lies in the network protocol or the user experience. What made Napster great from a user perspective was that you put your music into the pot and mix it with everyone else’s and benefit from the aggregated pieces. Isn’t the same thing happening with “web 2.0” sites like Wikipedia? Wikipedia’s architecture is only slightly less P2P than Napster’s central-server-indexing scheme.
Oh, and it must say something about the target audience that a proposed P2P SIP is offered as evidence that P2P is still healthy but BitTorrent isn’t mentioned.