And to piggyback on this question, what is the advantage of creating new protocols to replace IP, TCP and UDP? They're well known, well understood, and have thousands of implementations out there. Sure, it may be possible to make a more efficient protocol for low-bandwidth environments, but most of the time there you'd either use a very minimal higher level protocol, or translate to TCP/IP at the other end anyway.
I know I've done a deep dive on this in the past, and found some interesting things, but nothing jumped out at me as a clear win over IP.
And now I have to go read the RFC's for IPv4, IPv6, TCP and UDP (and anything else I discover while I'm at it). Thanks. I had better things to do with my day off.
I don’t see a lot of value in replacing TCP/UDP and further up the stack. But the value in replacing IP is having a permissionless way to join the network and not have a centralized address issuing authority. I also envision some sort of mesh wireless network as a permissionless physical layer to complement the permissionless addressing layer.
Depending on scale, doesn't Zeroconf address this? I'd also have to look at a bunch of the mesh networks that are popular in the ham radio space to see what other protocols are in use.
In any case, I'd guess this is mostly a solved problem, but it's solved ABOVE the IP layer, not as a replacement.
Oddbean new post about login | logout 
