Given the recent mini-controversy over the AC change to leather armour in the revised edition of Labyrinth Lord (from 7 to 8), I'd like to talk about a related issue: ascending versus descending armour classes in D&D retro-clones. Some clones have 3.x-style ascending AC (and to-hits), others have traditional descending AC with descending THAC0 and/or a to-hit chart.
This issue is a bit of a special case in the old-vs-new D&D debate, because new-schoolers seem prone to claim that ascending AC is objectively superior. Jonathan Tweet (co-creator of 3rd Edition) said that it's “just clearly better” in his blog:
So... is it? Well, with descending AC, you need to either work from a chart or use THAC0. Dependence on a chart for a fundamental combat function can be inconvenient when juggling all the other things involved, while THAC0 and subtraction makes the mental arithmetic just that bit harder. The descending AC doesn't bring many advantages as such in return: I don't count the fact that descending AC is a D&D tradition as an advantage in itself, because I don't think that tradition for it's own sake has any value.
However, whether it's objectively better or worse isn't the right question to ask, I think. A better question is: “Is it actually a problem?”. This is where you really need to look at the AC system in the context of the combat system as a whole. 3.x's ascending AC may be a tiny bit easier to work with, but this is in a system that desperately needs any little bit of streamlining because the overall system is overly complex and bursting at the seams with modifiers. In OD&D/BD&D, the system is already simple and easy to work with. So you might need to look up a table for every attack roll – but there's only a couple of tables that you need in total, so it's easy to have them all on a single page right in front of you. If you use THAC0 instead, the mental arithmetic added still doesn't make running combat hard.
Also, a positive advantage of descending AC is seamless compatibility with old D&D material (plus, descending AC tends to be a de facto standard in the OSR (old-school renaissance) community, and there's plenty of free content available online). What little speed in-play that might be lost through working with descending AC is gained right back by not needing conversion.
All in all, I think that Tweet's claim that ascending AC is “just clearly better” than descending is (pardon the pun) fantasy, and irrelevant besides. It's really no big deal either way in the games it was used in; it would be awkward if (for some reason) descending AC were ported into 3.x, but that points more to a problem with 3.x than with descending AC.