If someone rolls a 17 to hit an AC of 18 the response is usually more along the lines of"So close! You just barely missed!"
No one says that if you roll a 2, and they especially don't say that if you fumble on a one.
In most versions of D&D 1 is not a fumble, just an auto-miss.
And when people say "You just barely missed", in my experience they are talking about the numerical gap between the number rolled on the die and the number required. I don't think they take the view that (for instance) a skilled fighter swings absurdly wide on (say) 1 in 4 strikes, even though 1 in 4 attacks will involve a 5 or less on the die. They certainly, in my experience, don't assume that a skilled fighter fumbles his/her weapon 1 in 20 strikes.
The d20, at least as I experience it, is about rationing outcomes, not modelling how bad a miss is. For instance, from the d20 roll how do we tell whether or not the opponent's shield made a difference?
In the old days I remember seeing it suggested that you know the shield mattered if the attack missed because of the shield but otherwise would have hit. But that makes no sense at all. If 1/2 of attacks are misses, and a shield grants +2 AC, then this means that the shield only mattered to one-fifth of those missed attacks (the other 4 in 5 would have missed even without the shield). The notion that a shield is useful in defending yourself only 6 seconds in every 30 of fighting doesn't strike me as very plausible.
Parrying and blocking are something else again - there is not any way of correlating miss results to parries or blocks unless you take the view that this what DEX is for, and do it the same way as shields - with the implication (inherent in some suggestions that we could solve the issue by making it "damage on a touch AC hit") that no one with a DEX below 12 ever parried or blocked a blow.
Sorry that's rambled and diverged a bit - I'm just trying to explain, with some examples, why I don't think that the d20 result on a miss can be correlated with the degree of finesses of the blow, anymore than is the case for the result on a hit. (Even if a crit is rolled the real determinant of the effect is the relationship between the damage roll and the target's total hp, not the die roll itself.)
Skill checks are already set up on the idea that the closer to or higher the result compared to a DC, the more successful the result, with rolls exceptionally lower entailing hazards and such. Why should an attack roll be under a totally different paradigm where everything is totally binary and a 5 is exactly the same as a 15 if both "miss"?
Skill rolls in 3E are a distinct mechanic. If attack rolls were like skill rolls, then bonus pips on the attack die would add to damage (as is the case in Rolemaster, for instance).
This inconsistency of its skill mechanics with its attack mechanics is just one reason why I'm not a big fan of 3E.
Conversely, if D&Dnext were changed so that degree of success on a hit affected damage dealt, then I agree that the "damage on a miss" mechanic would no longer be a good fit.
from a game perspective, if you're burning a limited use resource, it should usually do something. But a fighter's attack, usable all say every day, need not. No real resources are being expended so having no effect is excusable.
A resource is being expended, namely, an action in the action economy. That could be spent doing other things. Part of what the great weapon fighter's player gets for spending that resource in that way is a guarantee of auto-damage.
They shouldn't be able to just close their eyes (disadvantage) and swing wildly and still drop kobolds.
I think we can safely assume that designers have assumed the character is doing their best to hit. I'm sure they'll forgive you for ruling that, if a player declares an attack in this way (i) they don't do any damage, and (ii) the player in question is a dick.
If your worry is about the rule applying when a player is not being a dick, but is (say) fighting a medusa, that's more reasonable. In the same way that 4e imposes a "no minion death on a miss" rule, you might impose a "no auto-damage to targets you can't see" rule. (4e takes this approach to fighter AoEs, too - target's must be in the area
and visible.)
An interesting point, in that it illustrates exactly why damage on a miss is bad. If someone throws a grenade at you and you have improved evasion, it reduces or eliminates your motivation to jump for cover. Similarly, if you have damage on a miss and you're attacking an opponent who's almost gone, there's no reason to try to swing effectively.
At least when it was first introduced (the monk class in A&D), improved evasion
modelled the character's ability to jump for cover.
And as for swinging effectively, the player has no control over whether or not his/her PC swings effectively and so the issue of "reasons" doesn't come into it. Every time I roll that d20 I'm trying to roll a 20, but it doesn't seem to have effected my outcomes yet!
I have that game. Its called Toon. Very fun with the right crowd, but not what I want with Dungeons and Dragons.
And thanks for the reminder that everyone playing non-simulationist games, or with non-simuationinst preferences, is just a Toon player at heart. Without these periodic reminders that I hate consistency, sense and verisimilitude in my games I might almost forget and run a serious session by mistake!