But yeah I'm going to walk away from this because it's tangential at best to my original point.
Probably for the best, at this point.
the DM could frame a scene wherein the advisor uses powers, skills, etc. against the PCs, but then we're moving into a combat encounter or some other framework.
Thanks to the wonder of 'exception-based design' (in context, 4e's much more constrained nod to DM Empowerment) the DM could give an NPC powers or traits that directly interact with a Skill Challenge. There aren't precedents for that (that I recall atm, anyway), but you don't need precedents ('exception' based, afterall).
Or the DM could model such within the framework of the SC system by, say, describing how the advisor augments his slick words with a subtle charm, thus setting that particular skill check by the PC as a hard DC (Arcana) or something similar.
One thing that's easy to lose sight of when discussing 4e, especially for those who have little experience and/or appreciation of the system (but even for those of us who have more), is that 4e got further away from the rubric of 'realism' than did other editions, and was a good deal more abstract. Like 1e & 2e, 4e used completely different types of stat blocks for monsters, and used them for NPCs, as well, even more often than those earlier editions. The resolution mechanics in 4e are more player-facing, too. They play well 'above board,' rather than behind a screen, not requiring secrecy to work or maintain illusions.
That does mean the modeling of PCs is pro-active and action-oriented, PCs do things, move the story, and are the focus of that story. Everything else provides the back-drop and challenges of the PCs' story. An NPC isn't ever a protagonist, even if it's an ally of the PCs, and helping it accomplish something is a focus of their story for a time. An antagonistic
NPC, like the one in pemerton's example, is, by definition an antagonist to the PC protagonists, it would be absurd, not just in the sense of silly or counter-productive, but logically nonsensical, to try to give him 'protagonism' or agency.
I think the point is, rather, that the abilities of the NPC should figure into the skill challenge, as that paints the challenge more vividly and makes it more interesting. The basic SC mechanism - n successes before 3 failures, at DCs determined by the
level of the Skill Challenge (not the party) - does not have a lot of space for an opposing NPC. The opposition (or just involvement) of an NPC might determine the level of the challenge and influence the difficulty of checks & number of successes required.
For instance, the level of the challenge in pemerton's example might have been based on the level of the Vizier, as the main antagonist, or on that of the Baron, as the object of both the PC's and Vizier's maneuvering. The difficulty of individual checks might also go either way. The active opposition of the Vizier could mean more successes required.
That's fine as far as it goes, but I've taken it further with the above option of giving an NPCs powers that directly affect a Skill Challenge. It'd've been nice if Skill Challenges had had more than a couple of years to develop and evolve, though.
In these--any many other, I'm sure--examples, the DM uses the framework of the system to represent NPC actions and adjudicate outcomes of these actions in response to what the PCs do rather than relying solely upon her judgment, which, without such mechanics and to return to the original point of this post, sets up the possibility of railroading.
I'm not too bothered by the idea of a DM favoring linear storytelling over sandboxing, but there are advantages to having a solid resolution system that the GM needn't constantly apply his judgement to (though he could still overrule it if he saw fit, and, of course, uses his judgement in deciding when/how to apply it). One of the more apparent is that it's a common resolution system, it's the same for everyone at the table, and it can be reasonably fair (hopefully balanced, too), allowing players to share the process more readily, and to make decisions with some idea of what's at stake & their chances of success.