Part of the challenge in Curse of Strahd is, I think (or perhaps this was just how I ran it) that you cannot effectively remove all of the spheres of support he has available to him.
<snip>
I do think that there was skilled play in what my players did. And they did set about removing certain strengths that Strahd has (he has a right hand man, there is a witch who is like a mother to him, and he also had Ireena as a hostage, and they ultimately rescued her).
But all of these were crafted more to present an interesting scenario and to see what the players would have the characters do. Yes, there were challenges they had to face and overcome, but that wasn't the focus of play the way it may be in some other scenario.
As you have probably noticed, I've really appreciated your Strahd example for providing a concrete focus for discussion in this thread.
Everything you say here makes sense. And resonates with my own play experiences (not of Strahd, but of other campaigns).
For instance, I'm going to conjecture that your players used skill in (say) rescuing Ireena. (I don't know the details: maybe there was a clever stratagem, or they used their spells well, or even just chose sensible targetting in defeating the guards who were holding her.) But as you say, the rescue of Ireena didn't really bear upon the subsequent encounter with Strahd from the "skilled play" point of view: it changed the range of story outcomes (eg maybe even if the PCs lose to Strahd and have to run away, at least they saved Ireena), but it didn't actually make it easier, when the rubber hit the road, for them to beat up on that key NPC.
There are some systems that allow a bit more "bridging" from scene to scene, but when I mention the systems you'll see why I still think this falls short of supporting a "skilled play" imperative in any robust fashion:
* In Burning Wheel, a successful check can allow a bonus die to be carried forward into a new check (this is called a "linked test") - in this sort of context that might be a Barovian Rumours-wise check to gain the bonus die for a Folklore check made at the moment of encounter with Strahd to deploy his weakness (eg five cloves of garlic tied in a bundle with a lock of Ireena's hair) against him;
* In Prince Valiant, emotions and morale can adjust the size of the dice pool, so maybe Strahd suffers a -1D penalty because he knows Ireena has been rescued;
* In Cortex+ Heroic/MHRP it is possible (based on GM adjudication) for success in a scene to generate a "persistent" Asset that is therefore able to be carried forward into a new scene - there are a range of ways that might happen in a Strahdish scenario that could mimic either of the above dot points or be something different again (eg Knowledge of Strahd's Parapet-Brooding Habits, which could easily figure as a bonus die in a range of pools for actions that benefit from knowing where Strahd is, or from reliably avoiding being seen by him).
There may well be local moments of skilled play in generating these sort of augments/debuffs; but in all of these systems the GM still has a
lot of authority in actually framing the ensuing confrontation with Strahd, including the benefits Strahd gets from any regrouping and/or marshalling of resources.
I think story curation is kinda baked into TRPGS. Even where it's not the GM, someone (or someones, or plausibly the whole table) will be making choices based on what the story is, and where they think the story should go.
I just don't think this is true. A perfectly feasible D&D campaign could start with B2 KotB, played in classic "skilled play style"; then move to (let's say) The Isle of Dread and Castle Amber played the same way; then White Plume Mountain; and finishing, just for laughs, with Tomb of Horrors.
This isn't a campaign I would want to GM or play in - see my most recent posts in the
D&D General - Is this a fair trap? thread - but I think it's feasible, and is the sort of campaign that inspired traps like the one under discussion in that thread.
Even when we move away from that style of D&D (or D&D-adjacent eg Tunnels & Trolls) play, it still needn't be the case that we have story curation. The ideal of a whole host of RPG design is that the game works if the players make decisions based on
what they think their characters would do and the GM makes decisions based on
what the mechanical outcomes permit and
what they think the fiction demands. I've approached my Classic Traveller game like that. The only time it really didn't work was when it came to onworld exploration via land-vehicle: it quickly became clear that I just had to make a decision, independent of any action resolution process, that the PCs arrive at the place they are heading to.
(The designer seems to envisage that this will be avoided because the GM will be able to do map-and-key resolution of the players action declarations about where they drive to. Besides the risk of terminal boredom, this is obviously utterly hopeless for a game that takes it as given that the PCs will travel from world to world to world.)
Of course in making decisions in my Traveller game about (for instance)
who is in a just-encountered starship I make choices that (I hope) are interesting and provocative rather than dull and stultifying. But I think
@Campbell already gave a good explanation of how this is different from making decisions based on story outcomes.