No matter what you do you can never get the granularity right.
"Never,"
Reynard, is a deeply subjective judgement.
I looked at a bunch of systems for running
a 17th century swashbuckling game.
Riddle of Steel and
Swashbuckler were much too fiddly for my tastes.
Lace and Steel was too gimmicky.
Savage Worlds missed the strong genre archetypes out-of-the-box (and before the
SW fans get in a snit, I bought
Pirates of the Spanish Main and
Soloman Kane, though I returned the latter).
I also considered using
AD&D with the 2e historical supplement
A Mighty Fortress, but the problem of "undefined skills" and "reduced options" for me is that it also utterly failed to capture the flavor of the swashbuckling genre.
So there was a good level of granularity for me in
Flashing Blades: four archetypes, with characters choosing anywhere from three to six broadly defined skills, which makes a character on a par with an
AD&D thief or ranger (or a B/X elf or dwarf) in terms of 'class abilities.' It does what I want it to do, without doing more than I need.
Some genre archetypes are going to be good with "broad skills" and some ate not: look at Conan, noted particularly for his climbing skill.
Conan's climbing skill is part of his general athleticism; I have no problem with assuming a similar level of skill to anyone in the fighter class, subject to the character's actual attributes.
This is why essentially undefined skills - that is, reduced options- is preferable in many ways. You b/x fighter needed only a brief character background - a Cimmerians hill man - to expand because what "Cimmerians hill man" entailed and when it applied was based on judgement calla, not hard and fast rules.
While I have no problem enjoying
OD&D or
AD&D as-written, the games I most enjoy add a bit more detail than that, but certainly nowhere near as much as d20, GURPS,
Rolemaster or other more rules-laden systems.