If you don't do that and connect directly to a central node, you've done what Skype has done: become a client-server application with centralised servers.
As a result, it's much easier to maintain a single architecture than two different architectures with rapidly diminishing number of desktop nodes taking all the traffic.
> rapidly diminishing number of desktop nodes
Do you have any data on that? Or are you counting laptops here as mobile for some reason?
What a nice techy answer. Users want an audii-video application that didn't suck their data plans dry by sharing their network bandwidth with other participants in the peer-to-peer network.
The moment you stop sharing your bandwidth and CPU you become a client-server application.
> Do you have any data on that
Any source you pick. And no, it doesn't include laptops. Just an example: https://searchengineland.com/report-57-percent-traffic-now-s...
Even if there's no fixed server you're talking to? You seem to have this weird idea that you can't have a heterogenous peer to peer application.
2. The fact that the server is not fixed doesn't make it a peer-to-peer system. Example: when you watch a video on Youtube, the video isn't streamed from a fixed server. This doesn't make Youtube or whatever client you use to watch it peer-to-peer.
3. Where do these "no fixed servers" come from? Do they magically appear out of nowhere? No. They are some servers that people set up as servers, so this means that:
3.1. If Skype were to retain its peer-to-peer architecture, it would rely on Skype users to be knowledgeable enough to set up their servers (and to turn off peer-to-peer capability in the event of poor machines and unreliable networks).
3.2. The servers by necessity would have to be on desktop computers. Even in 2011 it was apparent that the desktop traffic was rapidly declining, as more and more people were switching to/using mobile. It means Skype would have to route more and more traffic through fewer and fewer nodes. Oh, and yes, those nodes would also see their data plans drain (as not everyone in the world is on unlimited plans). Or — surprise! — set up its own servers, which it did.
3.3. On top of that Skype was en route to ditching peer-to-peer even before MS acquisition, as peer-to-peer architecture had been the primary cause of at least two major (unpredictable and uncontrollable) outages that could easily be avoided with a centralised setup.
Original Skype network existed in a world where mobile wasn't a thing.
Original Skype network went down twice exactly because of its architecture.
Original Skype network was already on its way out for the reasons I described several times above.
But sure, keep on believing that peer-to-peer can work in a mobile-first world.