Generally speaking, if Y happens, it must have been caused by something (X). If in the mechanics, HP is lost, there must be a reason that HP is lost. If in the fiction, injury is suffered, there must be a source of that injury.
<snip>
A mechanical process in a game like D&D is there to help adjudicate the effect of the various actions the people are taking in the fiction.
The loss of hit points is represented most viscerally by a notation on a piece of paper. That is caused by a die roll mediated through someone's brain and thereby interpreted as a certain sort of "move" in a game. The die roll is in turn triggered by an action declaration, typically "I attack it".
Whether or not that notation on a piece of paper also represents something in the shared fiction of the game is a further question. Many people take the view that not all such notations do represent something. For instance, many people doubt that writing down XP totals actually represents something in the fiction. Similarly, I don't think anyone thinks that initiative scores represent anything in the fiction (they're just a metagame ranking device for making the action economy work).
Given that reducing the hit point total does move the target closer to defeat, we can confidently say that a hit point loss represents
that much, namely, that the opponent who made the attack - ie who engaged the target in combat - has pushed the target closer to defeat. Whether the hit point loss corresponds to any discrete injury is not a question we can answer just from considering the mechanical situation (unless of course the hit point loss reduces the target to 0 hp). It seems to me largely a matter of taste.
D&Dnext seems to default to the assumption that if you're not bloodied yet, then hit point loss does not correlate with injury, but if you are it does. That doesn't particularly suit my taste, and I think it can cause some other wonkiness, but it's not an obviously flawed approach. Others who want to argue that every hit point loss event at the table corresponds to an injury event in the fiction are of course free to take that approach, but I don't think the mechanics in any way mandate, or even speak in favour of, that approach rather than the D&Dnext default or (what I take to be) the 4e default that the only injuries are those which correspond to condition infliction (including "bloodied" and "dying") as opposed to simple hit point loss.
The mechanic only exists to resolve the question of "I do X. What Y happens?" The mechanics do not resolve the question of "Y happened. What X did I do?"
Says who? (Other than you, obviously.) Gygax clearly had a different view, in relation to attack rolls: you don't know exactly what you did until attack and damage are resolved; and in relation to saving throws, where again you don't know exactly what you did until the saving throw is resolved. And it's not as if Gygax didn't know a thing or two about how D&D mechanics might work!
The rolling of dice is an answer to the question, "What happens when I try to hit the goblin with my sword?"
You can treat it that way. I think it's most neutrally interpreted as answering the question, "As I engage the goblin in melee while wielding a sword, who is being worn down?"
Without that try, there's no roll of the dice. There's no outcome to resolve if there is no cause that creates the potential outcome.
You can treat it that way. Or I could say: without that
conflict, there's no roll of the dice. Nothing in the mechanics speaks in favour of your treatment over mine.
The player announces what her character *does*. The resolution tells you what happens when she does that.
The player announced what his character hopes for - say, defeating the goblin in melee - and the resolution tells you to what degree, if any, that hope was realised. Again, nothing in the mechanics speaks in favour of your approach over mine. Gygax clearly thought that at least some of his mechanics (say, attacks and action economy, and also saving throws) played better when interpreted in the way I favour. And of course games were designed in reaction to D&D (eg RQ, RM) that overtly took the approach you favour.
the action the players take in a TTRPG is "I do x." This is a core gameplay loop: Player does a thing, DM describes the results of doing the thing, player does another thing, and on and on.
Sure, the player does a thing. That may or may not correspond to the character doing a thing at the same time that correlates in any meaningful way. The player can declare "I roll an attack". In the fiction, that corresponds to "My guy engages that guy in combat and tries to defeat him", but in D&D at least it's hard to be much more specific than that until the action is resolved.
Mechanics exist to adjudicate results, not determine characters' actions.
The mechanics need an input (action) to describe an output (result).
D&D's mechanics singularly do not require an input of a detailed description of an action in the fiction. The rules of all pre-4e editions, for instance, are clear that a player can roll a saving throw to try and evade the effects of a spell without describing any evasive action being taken by the PC in the fiction. And Gygax in his DMG is quite clear that the action taken (or not taken) can be narrated after the save is resolved and the mechanical (and hence, often, fiction) consequence thereby known.
This seems to presume that the output is determined before the input, that we know the effect before we know the cause
Correct. Unless your PC is wielding a vorpal sword, until you roll the damage roll, and we thereby learn whether or not the enemy has been reduced to 0 hp, we can't know whether or not your blow struck his/her neck with enough force to sever it.
The input is broadly described ("I attack the orc!"), and the output can be broadly described ("The orc is wounded!").
As I said earlier, no one is stopping you doing it that way if you want to (although some people will, perhaps reasonably, doubt that the orc is very badly wounded if s/he goes on to win the combat and continue to carry on without signs of physical debilitation.)
But nothing in either the general idea of RPG mechanics, nor in the particular design of D&D mechanics, pushes in favour of your preferred approach. You might wish that D&D was a different sort of system (perhaps more like Runequest, say). But it's not. Gygax's way isn't the only way, but it's not like he was wrong about how his game system was meant to work.
