I 100% agree with you that the wild magic sorcerer surge was completely flubbed. But just because there are (more than one!) instances where "rulings not rules" clearly failed, it doesn't mean that the philosophy as a whole is wrong. It is very difficult to have a "complete" ruleset, and attempts at it are often complex and not user friendly.
Which is why you should take the alternative that is
neither (a) "rulings not rules," where no game exists at all unless the DM permits it and literally every line of text can be line-item veto'd (or un-veto'd) at a moment's notice,
nor (b) "absolutely perfect coverage," where the game has a single, discrete rule for literally every possible interaction and factor and complication, no matter how baroque.
Instead, you go for extensible framework rules. Rules that treat
classes of things (in the philosophical sense--"types" or "varieties," if you want a term without the D&D baggage) collectively, so that you can cover nigh-infinite variety with finite rules. Rules that range over all members of a not-necessarily-finite set. The necessary sacrifice, then, is that these extensible frameworks must be, to one extent or another, abstract. But that "sacrifice" is already true of all games ever, full stop, so you really don't pay
that much for it.
4e's Skill Challenges (coupled with Page 42 and the DM's Best Friend) are an excellent example of extensible framework rules, albeit ones that needed some more refinement time before first publication (and
desperately, horribly needed better explanations and MUCH,
MUCH better examples in published adventures.) Likewise, most of Dungeon World is built on extensible framework rules and specific pre-built applications thereof (e.g. the "Undertake a Perilous Journey" move.) You
can do this, and it can be an extremely functional, sometimes shockingly light structure that still manages to cover almost everything one might want to do.