Should be easily solvable with something like a co-op owned SMTP relay. That way, every member has a patent investment in the reputation of the relay.

I think the best option for self-hosters who don't want too much headache from their email is to host the email yourself, but send your emails through a relay other providers actually trust. It sucks to get so close to full control and give it up, but if you're like me, you send almost no email, so there's almost nothing at all to see for the relay, and all your received mail still comes straight to your server.

then you end up on a VPS with a bad spam/RBL reputation because they have other customers being spammy dickheads, and their entire IP allocation is listed.

Sure, if you don’t do your own research and legwork, first.

In fact, there are a shitton more bad corners you can paint yourself into if you are too fucking lazy to do some trivially basic research ahead of time, from hosts to software chosen to how you configure things.

Hell, a lot of these bad corners also involve turnkey solutions hosted by others. So simply having someone else do all the work can hose you just as badly. Sure, Azure/O365 or AWS Workmail or Google Workspace solutions are usually not that kind of corner, but unless you are willing to pay out the nose for minimal needs, you’re usually going to go for something a lot smaller and possibly more sketchy.

Good point. Though you are of course right, and it will cost money in perpetuity, I think it is important to distinguish between locations and which top level domain one buys under. At least where I live, I have government guaranteed rights regarding switching registrars and keeping my domain, and a court is needed (or lack of payment) to force me to abandon my domains. That way I avoid the problem of vendor lockinn.

The one thing that rubbed me the wrong way is that it treats the underlying host as “disposable”, in that it intentionally does not support in-place OS upgrades. As in, if you want to jump from an older LTS to the most recent, you need to explicitly spin up a blank, new LTS host, run the installation script from scratch, and then restore the data from an overnight backup.

I can understand an immutable container/binary with external data storage and config settings, but flipping the script and locking down the underlying OS? Yeah, that seems excessively unnecessary.