Every commercial mail service is enshittifying.
Most FOSS MTAs are, too. And self-hosting is extremely difficult because the requirements placed on senders in order to interoperate with that hellish jungle of providers without getting flagged as spammers by scam blocklists become more stringent every year.
It looks like the OS ecosystem is going to enter a phase of diversification as we desperately try to rebuild from the bits and scraps that are surviving the AIpocalypse. That means that in the near future, it's going to be important to have software that focuses on interoperability and protocols, and less on details of a system.
The path Linux is taking is considerably diminishing my interest in the Linux-specific plans I had: s6 extensions to handle cgroups, dbus replacement, PAM replacement (which I'm not entirely giving up on yet because other OSes need a better authentication framework too). I find myself looking at fundamentals and RFCs again.
Even what I have been working on this past month - sdcl, which I may talk about more in the future - does not feel as important anymore, because it's about interoperability of service managers, which is a good idea, but not necessarily what we need the most right now.
For the longest time I have asserted that I was not getting into the MTA world, but given the speed at which the ecosystem is deteriorating, I find myself questioning my assumptions and wondering about things that did not make any sense even six months ago.
Is it time for qmail 2 ?