What if the DM is applying realistic obstacles to player goals with the objective of both adding to the shared story and accurately arbitrating reality?
This is roughly how Burning Wheel works. It depends fairly heavily on a fairly tight constraint (by D&D standards) on PC power levels. And it also has other mechanical devices (both with respect to player resources and action resolution) to ensure that players can achieve finality in resolution, without everything being dependent on GM adjuciation of when, in "reality", a task would have been successfully completed.
It's an interesting system, which in my view resolves many of the problems facing gritty games like Runequest and Rolemaster.
Because I've never tried, I'm not sure how amenable 3E is to being run this way. I suspect you could probably play E6 or P6 along these sorts of lines.
Is DM force a dirty word?
No. In my experience, at least, good CoC play (i) is a lot of fun, and (ii) depends almost entirely on GM force.
But GM force is not the only technique going, and I personally prefer, as a GM, to use other techniques and to use a system that will support those other techniques without the need for force. (Adding for completeness: I don't regard myself as good CoC GM material.)
To me, it's far worse if the plan the DM has creates a bad game experience and he does nothing to change it, however arbitrary that might be.
<sip>
if you're going to take the good, you have to take the bad.
In my case, "DM plans" = framed scenes. If they're boring, you just bring them to an end (eg NPCs surrender; the PCs succeed; conflict otherwise is resolved in the PCs favour).
If you have a GM who, on a regular basis, can't frame interesting scenes, that's a basic GMing problem which system, or force, won't fix.
If you have PC builds which, on a regular basis, cause scenes which should be interesting to become boring in resolution, that's a system issue. Scry-buff-teleport is frequently an instance of this; so, I believe, is 3E Diplomacy in at least some uses. Save-and-die more generally is subject to criticism on these grounds.
What about circumventing them when there is something at stake? And who decides what the stakes are?
In the indie playstyle, the stakes are generally set by the players in the abstract - "Here are the things we, via our PCs, care about" - and by the GM in the particular - "Here's a scene in which one of those things is under threat". If the GM says yes, and therefore gives the players want they want and the players were hoping to struggle for it, that is bad GMing. (Luke Crane has a nice discussion of this in the BW Character Burner.) If the game doesn't create engaging play, for a table, out of the PCs' struggle, then as I just noted that is a sign of a system problem.
In the wargame playstyle, the stakes are generally set by the GM. Classic D&D tournament play is a limiting case of this.
And what about a more realistic example, wherein the player says that and the DM says "no" through some concrete rationale? The king is not receiving visitors right now. There is a standing policy not to receive anyone of the PC's race regardless of circumstance. The king is out of town. Etc., etc.
Different systems have different ways of resolving this. It also depends on what is action resolution and what is scene framing. [MENTION=16586]Campbell[/MENTION] said some sensible things about this 20 to 30-odd posts upthread.
If there are any mechanics that might cause the chamberlain to act that way, then you are arguing a strawman argument.
Campbell said some sensible things about this too, to do with the transparency to the players of what is at stake and what is possible within a given situation.
Character creation and actual play (AP) are completely different processes and have different requirements.
Agreed.
Spell acquisition. As ready mentioned mid tread, I am on the opinion that the DM controls the access to [uncommon] spells.
My own preference is for a game system which doesn't require the GM to oversee PC build in this sort of way.