Socketsupply.co has been around as a project since at least 2014 following a similar path to Safe Network in some respects and with similar ideals.
The founder is approachable, active on Mastodon as well as other SM and I include a thread below in which he introduces the project with some history, or you can visit socketsupply.co
In the thread he mentions having tried libp2p, found it wanting, notably in NAT traversal, and how they went on to build their own socket runtime which is FOSS and allows building of cross platform apps.
In summary, there’s no server infrastructure and every socket app is a node. So the infrastructure is all the apps, and it supports secure, local-first solutions.
Here’s the thread which also describes a demo app they use in place of Discord for their internal chat. Maybe some here will try it out and report back.
Uses websockets to establish client<->client comms
It has some fallbacks, but could also use long http requests (fairly inefficient) to get message passing as a form of comms.
Great to try from every angle for sure. My own feeling is p2p comms has been an absolute nightmare from day 1 on the web. Had it not then google/ms/facebook etc. would all have got nowhere near market share and data centers would not be required.
My feeling is libp2p for all it’s flaws is head and shoulders beyond anything we have right now and contributing there is a good idea to get that tech as the network layer for all projects. Then we all win, but of course the speed of improvement can be slow, the direction possibly set by a small group, bus as the tech is used in more places the improvements should come more democratically too.
I know if we had a ton of cash like some projects we would be sponsoring many grants for a ton of improvements there.
As I say though, love seeing as many attempts as possible at p2p networking