In F/OSS circles pre-Github a fork was when there was enough dissatisfaction with a F/OSS project (for many reasons) that people went through the effort of taking the source of a project at a given point and making an entirely new project based on it. Some famous examples of this kind of fork would be the GCC/EGCS fork, the Xemacs/Emacs fork, the DragonflyBSD/FreeBSD fork, the X.org/XFree86/Freedesktop multiway fork, the OpenOffice/LibreOffice fork, etc.
In this sense of the term "fork" it's a major watershed event in F/OSS that sometimes shapes the way future projects run. (And sometimes, like the GCC/EGCS thing, one of the branches becomes the "new normal".)
Post-Github, a fork is just what Github calls cloning a repository on their platform within their platform. Any time you look at a project on Github, if you have an account on Github you can "fork" it (in their sense of the term) which basically means you have a cloned snapshot of that project in your account. It's functionally identical to typing "git clone " on your own machine only it's all kept in Github's own ecosystem.
What I find funny about the people protesting the second use as some kind of Github conspiracy is that the alternatives they themselves recommend instead ... do exactly the same thing (but aren't subject to the same conspiracy theorist tripe)! Cognitive dissonance is a HELL of a drug...