Particularly in the wake of Third Edition's conflation of the concept of Armor Class with a target number (or Difficulty Class, to use 3e terminology) needed "to hit" an opponent in combat, it's fashionable in many quarters to claim that the OD&D/AD&D approach is somehow inexplicable, or at least anti-intuitive. While I'll grant that the 3e approach is simple -- no charts are necessary to derive the "to hit" number -- I think its simplicity comes at the expense of coherence. What many people forget is that, prior to 3e, Armor Class really was (at least in intention) a class. That is, the AC numbers were designators for different types of armor, each of which had different strengths and weaknesses against different types of weapons. The infamous "weapon vs. AC" tables of Greyhawk and AD&D were intended to make this more plain for the "alternative combat system" of the three little brown books. In Chainmail, whose man-to-man combat system OD&D uses, the connection is quite explicit, with a character's chance "to hit" dependent on the type of weapon he is wielding and the type of armor his opponent is wearing.
You'll note that, in the preceding paragraph, I put "to hit" in quotation marks throughout. There's a reason for that. Remember that, in both OD&D and AD&D, a combat round lasts 1 minute. Even in the case of combatants with multiple attacks per round (like monsters or high-level fighters), it strains credibility to imagine that each attack roll represents a melee action taking up to 60 seconds to execute. In the real world, that's ridiculously slow, even in the case of men encased in plate armor and wielding "slow" weapons. But both games use a very abstract combat system and each attack roll doesn't represent a single attack at all, but a series of attacks over the course of up to 60 seconds, including parries, feints, and other defensive actions intended to make it harder for one's opponent to score a hit.
But what is a "hit?" It has long seemed to me, given the abstract nature of combat, that a "hit" can only be a blow that lands either hard enough or in a vulnerable enough spot that it deals damage to one's opponent. In the course of a melee round, an attacker undoubtedly lands many blows against his opponent, but only those represented by a high enough "to hit" roll deal damage sufficient to subtract hit points from his total.
For this to be workable, even given the abstraction inherent in the system, you need to take into account the peculiarities of weapons against certain types of armor. If you don't, then it becomes harder (to my mind anyway) to figure out just what is happening and what Armor Class is supposed to mean. The Dexterity bonus to AC throws a wrench in the works. I can accept the idea of someone who is nimble being harder "to hit" in combat; that makes sense. However, his nimbleness does not affect his Armor Class, which is a constant. Instead, his Dexterity bonus should serve as a negative modifier on his opponent's chance "to hit." This may seem a small thing, but I think it's important, because otherwise Armor Class acquires some incoherence, particularly if continue to use the weapons vs AC table.
Of course, many people don't -- and never did -- use the weapons vs AC table, seeing it as an unnecessary complication. On some levels, I agree; I often ignored it for large combats back in the day. At the same time, I very much like the feel of fighters choosing their weapons based on the type of defense their opponents possessed. This gives a fighter a reason to carry multiple weapons and it rewards players who consider the benefits of doing so. I approve of that and think the game should encourage it. The only trick, I suspect, is finding a simple way to present the "to hit" numbers based on AC that can then be used in conjunction with the class-based attack matrices. Conquer that "pedagogical" issue and I think most people wouldn't find the weapon vs AC concept quite so off-putting.
But don't get me started on negative AC ...