Lately I've been using GNU Emacs more than usual, in a situation that's driven me to actively customize it (specifically, I've been pushed to work in MH-E because my remote X experience from home got worse). As one often does in this situation, I've been reading articles about how people set up and customize their Emacs environments with third party packages from ELPA and MELPA, and have been tempted into trying some of them out. Over the course of this experience, I've come to realize that the packages I'm most likely to like and stick with aren't the big impressive packages, but instead the modest ones like which-key, backward-forward, and most recently vundo (git-timemachine also qualifies here, although I don't use it often).

These packages don't attempt to make any sweeping changes in your Emacs experience, the way heavyweight packages do. Instead they sit there quietly, only surfacing when you invoke them or in their special situation (in which-key's case). They give me a little hand, not a big renovation. I understand that the heavyweight packages might really improve my Emacs experience, but at this point I have a long history with Emacs (although a somewhat off and on one), and I mostly want my Emacs experience to just work that way I'm used to (much as I am with vim and fancy 'improvements' to its basic behavior).

My impression is that it's hard to find out about such packages because articles about them are uncommon (partly for understandable reasons; there just isn't much to write about something like which-key). This is unfortunate, because I'd certainly like to find and look at other such modest packages that might quietly improve my Emacs life a little bit. I suspect that there are plenty more of them out there.

(I asked a bit on the Fediverse and got some suggestions.)

Probably this means I should stop even trying to look at the big third party packages that keep coming up. That makes me a bit sad, because they're often very shiny and Emacs is the kind of editor where that can really work. But my experiments have been time consuming and not terribly successful so far.

(Even some smaller packages have eaten up a surprising amount of time to look into, understand, and customize.)

PS: Looking back at my experience with some of the popular big packages, I think that two reasons they mostly haven't clicked is that many such packages are opinionated about how they expect you to use them and take a significant amount of work to adopt, with a fair amount of learning, customization, getting used to the different experience, and so on. Modest packages necessarily have a confined scope with less to learn and less that can be customized.