Firstly, thanks to everyone who shared their thoughts in yesterday's post on this question. It's both good to see I'm not alone in my apprehension and that there are a variety of different solutions to the "problem" I see in the rules as written.
As I've said before, I have no issue with changing OD&D's rules. I've done it before and I'm likely to do it again. Through play, I've found that some rules don't work as well as I'd like and so I've altered them so that they do work more to my satisfaction. That's par for the course with OD&D and one of the reasons why I find the game so enjoyable these days.
Nevertheless, I try very hard not to change rules too much. That is, I prefer it when my house rules are recognizably derived from OD&D, as opposed to being completely out of left field. Furthermore, I'm of the opinion that, while OD&D is far from perfect, it's generally a good idea to avoid changing rules with abandon, because it's often not immediately obvious what repercussions might result. That's why I'm philosophically opposed to rationalizing OD&D's rules, even if I'm sorely tempted to do so quite often.
My gut tells me that, like a rug with a bit of fraying at the edges, pulling a loose strand in order to improve the tidiness of the whole could just as easily result in the whole thing's coming apart. I've seen too many attempts to "fix" Dungeons & Dragons over the years -- sometimes from official sources -- that, in the end, had adverse effects on other parts of the game. I don't want to participate in that sort of game mechanical vandalism, which is why I'm so reluctant to make big changes, even to mechanics, like ability scores, that, on first blush, seem in need of fixing.
This is why I am deeply leery of turning ability scores into the locus of a skill system or universal mechanic of any kind. Such mechanics "break the frame" for me when playing D&D. In my opinion, old school games are (generally) characterized by the use of several systems that exist side by side but nevertheless work in tandem rather than by a single system. So, perhaps it might be possible to construct a collection of related systems that use ability scores in a more substantive way, but I instinctively recoil from any attempt to introduce a single, overarching system that uses ability scores to cover a variety of in-game situations. Call it irrational, if you will -- and I recognize that, on some level, it is -- but a universal mechanic is deal breaker for me when playing OD&D. It just doesn't feel right.
So where does that leave me? Well, I've pretty well decided to eliminate the XP bonuses from high scores. I think it's a problematic rule on many levels and its loss won't make a huge difference in the final analysis. I've also decided that, much as I think a Strength 18 Fighter ought to get more of a mechanical benefit from his great strength than a Strength 10 Fighter, too much of a benefit turns the game into an ability score "arm's race." That's one of my big beefs with AD&D. This is made clear early on in the Players Handbook where Gygax notes that having 15 or more in no fewer than two ability scores is "usually essential to the character's survival." This approach places too much emphasis on ability scores in my opinion, so I'd like a more "toned down" approach, where +2 is the top modifier and it's not limited to characters with 18 in their ability score.
Related to this is another concern. My feeling is that, in any game calling itself D&D, class and level should mean more than ability scores. Whatever mechanical benefit ability scores give, they should be more along the lines of "icing on the cake" rather than the cake itself, if that makes sense. Ideally, these benefits would be broad and small enough to be universally applicable, not just to one class or a group of classes. I find myself reminded (once again) of Empire of the Petal Throne, where several abilities, not just Strength, gives bonuses to hit and damage, so that a clever but weak magic-user, for example, might nevertheless be an effective combatant. If the bonuses are kept limited, I think this would go a long way toward making ability scores relevant without making the game revolve around them.
More thinking is needed, though.