I actually think the real problem with those is the exact opposite. Those projects were initially hacked together for their own specific workflow, and are now going through the "growing pains" of becoming popular and somewhat ubiquitous.

On GIMP: There have been various forks throughout the years led by various groups of artists, but they have typically been very hacky. Currently the Glimpse fork is planning to do a redesign of the UI.

On Libreoffice: This is the result of the split from Oracle -- the project is now driven by consultants and ISVs. It works if you're willing to pay one of those or become one yourself. If you pay $0 and expect to get a perfect clone of MS Office then you will be disappointed.

On systemd: The design seems to be heavily inspired by the "giant ball of C" design that is already used by the Linux kernel. I would argue that GNU/Linux in general is too decentralized to have an overarching philosophy.

From my perspective as someone who does UI on open source projects, ultimately a lot of the high-level problems come down to lack of resources. It's hard to find good designers who are willing to contribute to open source, and it's expensive to put together a focus group to gather feedback. The typical corporate methods of gathering telemetry and doing A/B testing are not really useful to any of those projects you mentioned.

First of all, they won't get more popular than they are now. Second, I doubt any person who uses said software for professional or amateur reasons has the skills or time to cobble it together. Because they either do one thing or the other.

GIMP: "Groups of artists." See, there's your problem. As much as I despise Steve Jobs, I finally came to understand why he was so appreciated.

Libreoffice: What you said doesn't explain why Apache OpenOffice is still better. I do not expect an MS Office clone (although an MS Word clone would be welcome) but I do expect those writing the software to check out MS Word and point out why it is better... and then slowly but surely move in that direction, not away from it.

systemd: You can't compare an init system, which has the role of starting daemons and whatnot, to a kernel which is basically hardware drivers.

As for your perspective... if you are working for 0$ do not complain about lack of resources. You said it yourself, didn't you?

You have asserted that systemd is "written and designed by people completely disconnected from the user". This is not true at all. The systemd people most definitely use systemd. In fact, I am hard pressed to think of any system or service management toolset where its authors were not also some of its users.

Moreover, the headlined article is talking about graphical user interface design, something which is largely inapplicable to systemd, and to system/service management toolsets in general.