All of the checks you mention are opposed and have consequences of failure. "Picking the pocket| of a bag sitting unattended and unobserved requires no check. Someone proficient with a weapon attacking a training dummy won't miss.
But why is this not true for casting a spell? As I posted already upthread, the somatic component of a spell helps define the target and/or AOE. This seems like it should have consequences for failure? So why is no check required?
Sure, a M-U can cast Spider Climb and climb automatically for a while. A M-U can cast Knock to open a lock without issue, but how often can he do it? A limited number of times before he needs to get slots back, and only if he happens to have the right spell prepared at the right time.
Meanwhile, in comparison, the thief can do his "thing" as often as needed and in what he can do, he can do whenever he needs to--it is always "prepared."
It is a balance between always good but limited in use (spells) vs. always use but limited in good (skills).
Yea, it's complicated. Balancing auto-success but limited use abilities versus increased probability all the time abilities is very much ability and situation dependent.
Like, if you have a certain task (like picking locks), and the normal chance of success is 50%, which would you prefer to have? A class ability that pushes the chance of success to 70%, or a once per day ability that lets you auto-succeed on picking a lock? That seems a pretty tough choice, based on my expectations for how prevalent locks are.
Now, do I think that there's room for non-magic classes to get resource-limited auto-success abilities? Absolutely. Heck, they're a main class feature in an OSR game I recently Kickstarted. But I can also understand limiting access to various combination of abilities in the name of class design heterogeneity.
My OP was not about class balance. It was about verisimilitude and consistency in the mapping of mechanics onto the fiction.
Here is the same point made in the context of the 5e Basic PDF (pp 60, 79):
Dexterity measures agility, reflexes, and balance. . . . A Dexterity check can model any attempt to move nimbly, quickly, or quietly, or to keep from falling on tricky footing. The Acrobatics, Sleight of Hand, and Stealth skills reflect aptitude in certain kinds of Dexterity checks. . . .
Whenever you attempt an act of legerdemain or manual trickery, such as planting something on someone else or concealing an object on your person, make a Dexterity (Sleight of Hand) check. The DM might also call for a Dexterity (Sleight of Hand) check to determine whether you can lift a coin purse off another person or slip something out of another person’s pocket. . . .
Spellcasting gestures might include a forceful gesticulation or an intricate set of gestures. If a spell requires a somatic component, the caster must have free use of at least one hand to perform these gestures.
So why does performing spellcasting gestures not require a Dexterity check of some sort? Why can a character with Dexterity as low as 8 (possible under standard PC build rues) of 3 (possible under rolled stat rules) perform these "intricate set(s) of gestures" with no trouble?
I appreciate that, as per the 5e rules, a check is only called for "when a character or monster attempts an action (other than an attack) that
has a chance of failure" and/or "[w]hen the outcome is uncertain" (Basic PDF p 58) but it seems to me like the ability of a clumsy person to perform those gestures would be at least as uncertain as the chance of a trained person to perform a bit of stage magic.
Many of the posts in this thread seem to amount to
it's OK to ignore verisimilitude and give MU/wizard characters a free success in the interest of class balance. Eg
@billd91,
@(Psi)SeveredHead,
@Paul Farquhar,
@Mannahnin,
@Ruin Explorer,
@Gadget,
@6ENow!
But if that was the right design principle than 4e would not have been as controversial as it was.
Others (eg
@Blue quoted above in this post,
@Cap'n Kobold, also
@6ENow!) have used the notion of "unopposed" checks. But the notion of opposition here isn't very well defined - why does a trap "oppose" being disarmed, or a lock "oppose" having its pins or whatever moved by a wire rather than a key, or a coin "oppose" being plucked from someone's ear, but the somatic gestures not "oppose" being performed?
AD&D doesn't have any general notion of a check being required when an action is "opposed", and in 5e as per the passages above the key notion is
uncertainty or '
chance of failure. Which brings me back to the OP question - why, of all the feats of manual dexterity performed by characters in the AD&D game, is there one category, namely somatic components, that can be performed flawlessly every time regardless of Dexterity?