But despite the clear efficacy of this method, it never occurred to me to do it on my own, just to learn something. It's a lot of work and somewhat painful, but I suppose a more complete mastery always is.
To sum up this subthread: we all understand "it" better when we have experience of active involvement with it, not just passive studying.
I think (as someone else mentioned too) teaching is generally good because one never knows where and when a challenge may present itself, but I'd suspect a paper gives deeper results, whereas teaching gives broader results, and both give more than study with neither.
Most specifications can be turned into a database for some form of Prolog, play with it. At the very least, you might get yourself a type or model checker, or you might get lucky and actually get an expert system out of it.
Of course, not all ideas can be so neatly expressed in code. Constructing essays and explanations for people is similar, though it's easier to finesse gaps with rhetorical tricks.
However, if the way you learn is much like mine, when you try to start writing you will realise that you haven't grasped it with anything like the fullness that you had thought. Start writing anyway: you have a bibliography, and hopefully you have a rough idea of the thesis you want to defend. Write the literature review sections. Sketch outlines of arguments, and then try to take them apart. Find counterexamples. If your position has any merit, you will eventually figure out some convincing arguments. Flesh them out: write a first draft, as rough as you like. Improve it by smoothing out poor phrasing and removing extravagant, unsupported claims (it may only be obvious once you've written them that this is what they are).
After some time you will have a reasonably presentable paper. More than that, though, you will actually understand the issues: how the various ideas in the field fit together, why you really disagree with some of the papers you've read, and perhaps even what your own take on it is. It's hard work! But writing is what makes the difference. If you just read a lot you can come away with the misapprehension that you understand the field, when in reality you just know a lot about what people have written.