In my experience there's two ways to do it: (1) use the house style (and standard auxiliary packages) of whatever publication you are writing for and/or (2) develop your own style files that use the subset of the 'thousands' of available packages that suit your needs and uses. I've got a small set of .sty files that I've been developing for decades to suit my own needs. When I needed a pseudocode notation for some papers on algorithms, I played with a half-dozen or so of the top contenders before settling on one, integrating it into my other macros, and writing a few of my own to make it easier to work with. No-one should learn LaTeX for a single paper, or only a few. That's like learning C for a single program:- sure, you can do it, but it seems a lot of work for small result. I treat LaTeX as the environment in which I develop my own ideosyncratic publishing style (that can now be shared with others), not as a giant garden that everyone is supposed to be simultaneously using all of.