But characters going for their dramatic needs with every fibre of their being and us seeing what happens when they do IS the story! The players decide what their characters want, the GM decides what "the world" wants, they both go after it with gusto and the story arises when we see what happens as a result. How do you do meaningful RPG story generation in any other way? Even authors I've read talk about generating stories this way in their own minds; the protagonist has a dramatic need and tries to get it, the antagonist (be it a character, an organisation or the world) tries its damnedest to stop them and the story is what happens as a result. That's how stories are made.
The story arises when we see what happens, but there is more than one way to go about that process. Namely, both the means and the effects contribute. Determining which of these is "logically prior" to the other when resolving actions with respect to the fiction makes a key difference, I think. The relative freedom or constraints on the means and effects may also be different. I'll try to outline what I mean, hopefully giving a fair shake to both of what I think are the two major approaches.
With a stronger "simulation" (means have primacy over effects) bent, I think the process is rather as follows.
1) DM frames a situation, hopefully with enough detail that players have a reasonable idea of what "scene resources" (e.g. chandelier, burning fireplace) are or may be present.
2) The player/PC has high-level goals they would like to achieve. For example, moving an enemy into a vulnerable position.
3) The player/PC examines what in-world means are available for achieving these goals. Even mechanical abilities are considered in this light. Physical pushing is likely to result in movement, so pushing is a sensible thing to pursue. If none of the available means seem likely to succeed, go back to step 2. At this point the fictional means are essentially fixed.
4) The action is resolved, but the DM considers whether the in-world circumstances merit a bonus/penalty/other to the process. The DM tries very hard to be impartial about how the in-world circumstances are applied to the process. The DM also tries very hard to be impartial about any changes to effects (e.g. only half normal distance moved).
With a stronger 4e-like approach (effects have primacy over means) I think it looks more like:
1) DM frames a situation, hopefully with enough detail that players have a reasonable idea of what "scene resources" (e.g. chandelier, burning fireplace) are or may be present.
2) The player/PC has high-level goals they would like to achieve. For example, moving an enemy into a vulnerable position.
3) The player/PC examines what game-mechanics means are available for achieving these goals. If a suitable power fits, it is selected. If not, return to step 2 or examine whether a scene resource can justify a stand-in. At this point the (probabilistic) effects are generally fixed.
4) The action is resolved, and the player and/or DM optionally indicates how the effects correspond to the in-world actions the PC took. Both try very hard for this result to make sense within the fiction.
In the first approach the player has considerable freedom choosing means bounded by what they believe makes sense within the fiction, although this is, of course, informed by their mechanical abilities. The DM has considerable freedom choosing effects, bounded by the means the player has indicated. Proponents like that these means clearly precede effects from the perspective of both the player and the character, and that the mechanical effects will generally have an effortless correspondence to the fiction. There is a risk that the desired effect is foiled by a poor DM decision, but that is tolerable on occasion. In any case, it is worth it because through this process the player is empowered to see the fictional world through the eyes of their character.
In the second approach the player has considerable freedom choosing effects within the bounds of the character's mechanical abilities, but this is, of course, informed by the fictional scene. The player/DM then has considerable freedom choosing the means employed, bounded by the effects the player has chosen. Proponents like that the fictional means employed can be almost anything that reasonably leads to the result, which really induces creativity and variety. There is a risk of failing to create means that reasonably lead to the observed effect, but that is tolerable (or even convenient) on occasion. In any case, it is worth it because through this process the player is empowered to see the fictional world through the eyes of their character.
I've tried to present a neutral picture of how both these styles can enhance the game for those approaching with a suitable mindset. (And, of course, it is possible to prefer a mix. For example, preferring method 2 for situations where the mechanical results are already quite abstract, but method 1 where the results are intended to have a clear correspondence with the fiction. Or some players might play mostly with 2, but reject any hypothetical action for which they can't come up with an explicit justification.) I fall decidedly into the first camp in most cases, but I'm trying to keep an open mind here.
I'd be interested in seeing whether we could create a design pattern that lets the same basic rules work "natively" with either style. For example, a maneuver system that separates the means and the effects but lets a fighter focus on building up greater strength in one set or the other. The means-based mechanics might allow a broader total range of effects, for example, but relatively less control over what effects actually occur. Perhaps a fighter could train in "forced movement" (effect-primacy) or "man-handling" (means-primacy). The former would indicate relatively broad training to achieve specific goals, the latter would indicate focused training to achieve relatively broad goals. At high levels they might not actually be so different in overall capability, but because the process of arriving at an action is different it might feel distinct.