There's a group in my neighborhood that adopts public sector projects and runs them all from a small cluster that they operate.

I keep thinking they would do better if they got ahead of things and suggested a toolchain for future projects, that would increase the odds that they get adopted.

Getting a few groups of volunteers together to learn a handful of LTS technology stacks instead of a cartesian product of all of them that grabbed two people's fancy three years ago and now they're bored/out of money. It would make it a lot easier to get to a more PBS-adjacent model of internet for the public good.

In some respects this is a different sort of sustainable than what they mention in the article, but amortizing a bunch of relatively low-pop services across a single cluster and admin team still counts as an efficiency, versus having them scattered on disparate hardware, disappear from neglect, to be recreated again in a few years from scratch, after someone squats the old URL and refuses to give it back.