When I think of 1st Edition AD&D, I don't think of the word "streamlined". Thief skills are based on a percentile roll, finding secret doors is rolled low on a d6, initiative is rolled high on a d6, attacks and saves are done with the iconic d20. Magic resistance is a percentile. The switch to 3.x/Pathfinder changed that: now almost everything is done as a d20 roll against Armor Class or a target number, with bonuses and penalties applied based on the situation and the skills or experience of the character rolling.
Some people think the OSR is all about quirky rules, and doing things the old ways for Old Times' Sake, or maybe because we're just crusty old reactionaries. Personally, I don't like a rule because it's charming, or old, or new. I like a rule because it models a situation in a way that lets me conceptualize it and think of options for play. And different situations (arguably) need different models.
Take Call of Cthulhu's experience system, for example: if you have used a skill successfully, you roll a percentile after the adventure to see if you've improved. If you roll higher than your current skill, you improve by 1d6. The beauty of this system is that it models actual skill growth in an elegant way: when you're inexperienced, you're unlikely to succeed that often, so you don't get to roll much. If you do, though, you have a high chance of improving. The combination of success rate and improvement rate picks up speed until the 50% mark, when it becomes less and less likely that you will make the improvement roll even if you do apply the skill successfully. By the time you approach 100% in a skill, it's nearly impossible to improve, even if you succeed almost all of the time.
But that's not how the Resistance Table works. The table is based on opposing attributes (Strength vs. Strength or weight, Power vs. Power, etc.), assuming that equal ratings gives a 50% success rate. A modern designer might look at this and think "Wait! There's two systems here, where there only needs to be one." This hypothetical designer might decide to make a game based entirely on the Resistance Table, adding numbers to attributes to represent skill (Strength plus Sword Rating, or Dexterity plus Stealth); that might be a fun game to play, but it ignores the simple fact that Skills and Resistance Rolls were meant to represent different situations, and don't necessarily benefit by being modeled with the same numbers.
I mention this because my OSR-based game (cobbled from many sources) uses a modified Siege Engine mechanic (adapted from Castles & Crusades) for a lot of things. It's an attribute-based roll and offers an elegant way to customize your character without littering the sheet with feats and skills. I like the rules I've come up with, but I have no compunctions about altering or adding to them. It might be that a situation only has a few identifiable states, so it could be modeled with a d6; it might be that the situation falls in a bell curve, in which case 3d6 (or other combinations) might be called for. Or perhaps a d100 roll, if I want to break out the calculator and figure out the math to a single percent.
Whichever it is, I think it's better to allow the situation to guide the numbers. It doesn't have to be "quirky", nor does it have to hew closely to an imagined Universal Resolution Mechanic; all it needs to do is be (a) fun, (b) fair, and (c) capable of being influenced by wise player choices.