How do you learn modern front-end development when the tooling feels overwhelming and over-engineered?

Learning front-end when it feels over-engineered

Pick one thing and take the default configuration. The thread's strongest agreement was to stop sampling frameworks, start from whatever the chosen one blesses as its standard setup, add nothing to it, and build something small — then add tools only after feeling the specific problem each one solves.

The method the thread agreed on

One technology at a time. Several replies read the frustration as the cost of learning five things at once.
Chase the conceptual model before the details: what the key abstractions are, and how they interact.
Take the standard starting configuration and change nothing, build the smallest real thing on it, and note where it hurts. Add the next tool only against a pain you have actually felt.
Build small projects rather than following tutorials to completion. More than one reply drew the line between understanding a tutorial and being able to start from an empty file.
Read something longer than a tutorial. One reply argued a book gives the deeper model a tutorial never does.
Shore up plain JavaScript first if it is shaky, because the framework will not cover for it.
Read the source of small projects you find interesting, until the shape of an application becomes familiar.

The reframe that did the most work

One reply answered the phrase over-engineered directly, by pointing at how plain a well-known search page looked in the late nineties and suggesting the asker start from the interface problem and iterate, rather than from the stack. Another made the connected point that none of it is genuinely new: front-end frameworks inherited their ideas from desktop interface frameworks and from server-side web frameworks, so an experienced developer is usually meeting old concepts under new names rather than learning from nothing.

Why the tool names are left out here

This thread is from 2016. The specific libraries, starter commands and state managers argued over in it have since been deprecated, renamed, or replaced, and repeating those recommendations now would be worse than useless. What survived a decade is the method above, which is also the part that was never about any particular framework. Several replies were also written by the authors of the tools they recommended, and those are left out for the same reason.

What it does not answer

Which framework to pick. The thread argued about it and settled nothing, and the answer would have expired anyway.
How long any of this should take, which nobody claimed to know.