I'm very much in favor of all low-powered magic enhancing existing abilities instead of serving as replacements. Knock gives a hefty bonus to open that lock (perhaps for several tries) instead of opening it direct. Charm Person gives a +10 to appropriate social checks for the duration. And so forth. Arcana Unearthed/Evolved use this technique to great effect.
I definitely agree that magic shouldn't serve as an all-purpose replacement for any skill, possibly excepting the caster expending some pretty hefty character resources. A short term replacement that avoids becoming the obvious choice is also fine to me. Adding big numerical bonuses on checks usually gives me pause because, as I alluded to earlier, an already skilled person could use magic to become unstoppable in a particular area. (Hence why I favor dice tricks of various sorts that keep the final results of checks in the same range as the base bonus, but make it more likely to achieve near the higher end of that range). Magic that enhances vs. magic that bypasses existing rules usually have independent balance issues, and I think keeping them separate is usually a good idea unless one can smoothly scale the former to reach the latter in a very controlled way. I think the d20's uniform distribution generally makes this difficult.
For something like Knock, therefore, I'm OK with letting the caster simply replace the Thievery check with an Arcana check that does the same thing (basic functional continuity) as long as it is sufficiently different in the good, fast, cheap sense from normal lock-picking to not cheat the pure rogue. However, this version of Knock isn't that useful to a person already good at mundane lock-picking, and I think it could be beyond just being a second chance. For that reason I could see a letting Knock let a person make an Arcana *and* Thievery check against the lock and using the better of the two. Perhaps the Thievery check can only be used for this purpose if the caster in trained in the skill, to reflect the synergy between magic and mundane in this case. Either way a pure caster is pretty much are counting on Arcana to get them by, but an arcane trickster with both skills is actually getting some value added for their efforts without making them an auto-opener.
Other spells that traditionally obviated what we'd consider skills can find a middle ground in this way, in many cases building on from what 4e did with them with rituals. I would very much like it, for example, if something like "disguise self" was designed to work in a pinch, but simply could not replace a disguise skill without some serious effort. A good start, for example, might only allow the spell to disguise the caster as something they can currently see or with which they are intimately familiar. (Rather like teleport in earlier editions.) A person who already has the disguise skill, however, might be able to use it to craft more devious or flexible disguises on the fly (since they can direct the spell more firmly), or use both mundane and magical disguises for complementary features. Higher level versions of Disguise Self might loosen some of these limitations, but mundane disguise rules should be written in such a way that the two are complementary rather than in competition.
But you are right that this is a great place to use alternate dice instead of a straight plus. A decent sneaker in an elven cloak with 2d10 has a nice boost, while an already excellent sneaker in such a cloak is really hard to detect. You get the effect you want (skill still matters), with some definite boosts. Perhaps in the mundane to magical distinctions, it could be roughed out like this:
- Fantastical but Mundane: Elven cloak gives you 2d10 with sneak, doesn't help clumsy, untrained guy in plate all that much.
- Magical Aid to Mundane: Cloaking spell gives you +10 to sneak, helps everyone in obviously magical way, as now that plate guy sneaks like a trained, low-level rogue in leather, and the rogue can do things that simply aren't normally possible.
- Straight Magical: Invisibility - you are or you aren't. Skill still matters for sound here, but you are undetectable to sight.
I think that's a pretty good list of categories. I was going to go on a mathematical excursion here, but I think I'll save it for a separate thread that I hope to get around to writing. That is because it is just as much about the sneaking/spotting skills as it is about whether 2d10 or +10 are good implementations of these categories, and because this thread is more about the concept than the details of execution.
I'd like for most of the low-level magic to be of the first two kinds. As you move to more magical at a given level, the costs go up. (Obvious cost in a Vancian system--the straight magical is limited to the 1/day or slots or whatever. You'd have different costs with other systems.) As the caster gains in power, they start doing some of the first two more or less whenever they feel like it, either because it is now "at will" or similarly frequent, a ritual that costs little at this stage of the game, or other such reasons. The straight magical gets more frequent, but definitely retains limits.
That leaves room for mundane equipment to operate, still growing gradually less important by character level, but parts of it remaining good backup options well into the higher levels.
I'm totally with you here, although I think there is no need to describe mundane equipment as more or less important by character level, since whether equipment is magical or not should be decoupled from level (and I know you feel similarly). Even in a world with high magic mundane options shouldn't be frowned upon, which I think is what you were getting at. By extending the reach of the mundane throughout the game it also helps keep back the feeling that magic can become humdrum through overuse, or is simply necessary for every little thing. If we can make the big things that break the (mundane) rules feel set apart, I think that will support the fight against some of the demystification of magic of the past couple editions.
