I’m starting to come to the conclusion that there’s probably no real harm in easing the limitations of armor use on arcane spell casters.
It is already fairly easy to build an arcane caster who can use armor without significantly impacting their casting ability. There are a fair number of spells without somatic components, which are the only components that cause Arcane Spell Failure to come up. Most spells do have somatic components, of course, but many don’t. Between that and Still Spell, it’s fairly easy to cast almost any spell desired in armor (albeit at a higher cost, if you must use Still Spell).
What if Arcane Spell Failure applies only in nonproficient armor?
Metamagic feats are often considered undesirable because they increase spell casting levels and are inconvenient to use otherwise (prepared casters must prepare the spells with the metamagic feats applied, spontaneous casters must increase spell casting times). Still Spell is one that can sometimes literally be a live-saver, but most often not worth the expense to acquire (a precious feat slot!) or the opportunity cost.
Instead, what if proficiency in the armor being worn prevented the Arcane Spell Failure? Consider, armor proficiency is often an expensive proposition for an arcane caster — one to three feats, or dipping a class that offers armor proficiency.
In the first case, a single feat (Light Armor Proficiency) gives proficiency with armor that generally gives no more protection than casting mage armor (at least, until armor enchantments come into play). For two feats (Light and Medium Armor Proficiency) you can do a little better (up to armor bonus +6) but you’re starting to get into some serious cost for a typical arcane caster, and the armor starts to inhibit movement. For three feats (Light, Medium, and Heavy Armor Proficiency) the caster no longer suffers Arcane Spell Failure for armor use (but still may for shields). This is getting expensive for most spell casters. A human wizard can do it by third level, others by sixth.
On the other hand, dipping fighter (among other classes) may give proficiency with all armor and shields (possibly including tower shield proficiency, if a fighter). This might be a more severe cost for many casters because it reduces the character’s caster level — reducing the number of spell slots available, half the time reducing the maximum spell level castable, and often reducing the total effect of each spell cast.
Gish get a slight bump (your Ftr5/Wiz7 can cleanly cast ice storm or wall of fire without being inhibited by his armor), and I don’t see anything wrong with that.
I presume you’ll be converting this idea into Echelon. It is rather radical yet not unheard of in 3E, requiring Spellsword prestige class or base spellcasters of limited spells lists such as warmage, battle sorcerer variant, beguiler, etc.
For some people a one level dip into fighter wouldn’t be a terrible cost, at least for a wizard. Your access to higher level spells is no worse than sorcerer. A one level delayed caster level is not a tragedy. In addition to the armor, you get a bonus feat, a fortitude bump, and a few extra hit points. A sorcerer might not dip because they’re already delayed in higher level spell access; although, a few might not care about that and will salivate at an easier Sorcerer/Paladin multiclassing.
Some DMs have conniption fits about this. It was all the angst in 3.0 via one level ranger dips for two-weapon fighting. Personally, I didn’t have an issue with that nor have one with this, but hypothetically you offer this to the world, many in the world will gripe back.
You would also get griping from those who complain spellcasters Win D&D by virtue of existing, and now you want to give them more PWER!
(It’s buried in there, but again, I am ok with the idea. :) )
I got that you’re okay with it, and see your cautions.
I tried to convey in this article that arcane casters — wizards more than sorcerers, really — wearing armor isn’t really a mechanical problem. I reckon that by the time you can readily afford the proficiency, casting in armor isn’t that overpowering. There are *lots* of ways for an arcane caster to get comparable AC, and something I didn’t include in the article though I did think of it:
Arcane casters, as a rule, avoid melee if at all possible. Okay, so you’ve got a +8 armor bonus now… who *cares*, if no one gets to target you? It’s no better than if the caster had spent spells on not getting hit, and it cost him to get there, so overall I don’t see it as unbalancing. There might be flavor reasons to discourage it, but that’s frankly I figure that’s outside the scope of this article.
 which is another peeve of mine, that if you take heavier armor it’s no better in practice than other options. I want armor to be a valid choice, dammit! I’m sure you’re aware of my thoughts on the matter.