Perhaps, but in which case this concept actually adds value? Imagine you have a working environment (development, scientific research, etc) carefully set up to do your job. You wish to upgrade, say, your web browser, but now have to upgrade the whole environment. Changing versions can be very disruptive in many situations.

> Changing versions can be very disruptive in many situations.

You assert that, but is that really true? Updating "stable" distro should be non-event, the very opposite of disruptive.

Yep.

Imagine you're rolling out *nix 5000 endpoints; change is going to make the problem hard, so the idea is stables ABI/APIs should be the same, so it can like a platform you can build against.

The problem is, browsers pretty much have to be the latest versions, or websites won't work, and most hardware improvements are going to need a new kernel (which to be fair, given linus don't break user space position, is a probably safer than updating userspace), so it doesn't necessarily work for all use cases.

A common approach in webdev has been to keep current, but run a test suite so you know what breaks, and deal with it when it happens, but I don't think the average infrastructure person has kept up with web dev here, and unless you're running containers, you're probably still on a stable OS.