As stated previously, it seems that anything promoting player proactivity makes the campaign more exciting: as it is precisely this allowance for multiple people to participate in the construction of the drama which separates RPG's from novels or movies (and even such advances as hypertext nonlinear narrative -- because there is always a real present moment, in which decisions could be made differently, or outcomes to occur differently).
In this respect, the design of dungeons (static locations) is important in order to (1) make PCs proactive, making the decision to move and adventure wholly theirs and (2) at the same time make the world manageable for DMs (as it is unreasonable to expect DMs to plan out continuing movements, migrations, and gained experience all the time over a whole world -- setting up static areas at least reduces the DMs work necessary to a finite, if large, amount). To a certain extent, one must be willing to "waste" effort in the sense of designing sufficient areas that PCs will probably not explore them all.
Consider the difference between modules X1 and X4. X1 features a fairly large tropical island, with many set location encounters scattered all over the place: because of this, it is most unlikely that any party will confront all of them. In seeming reaction to this, the author of X4 (David Cook) writes on page 2,
"With the exception of the first few encounters and the last one, none of these encounters are set. Instead, they are organized by the type of terrain in which they occur... The DM controls the timing of all the encounters. He does not have to worry about the characters missing an important encounter by not going in the right direction."
In other words, for this adventure, it doesn't matter where the players go. On the one hand, the designer's job is easier, because every encounter is guaranteed to take place: no excess scenes are written. On the other hand, it makes no difference where the players decide to proceed unless they (possibly sensing this and feeling ornery, as has happened to this DM) decide precisely to turn as far away from the set adventure as possible. Again this designer's motivating idea was to plot out a particular story and needed and expected to be able to manipulate PCs and events so as to facilitate it. (Something like module X10 fares far better, as numerous encounters are included, the timing of such are based on PC decision, and not all are expected to occur; although the scheduled climax, mandating one special weapon to defeat the one enemy's defense is rather weak, in my opinion. EGG's AD&D modules fare particularly well in this regard. Often, the writer must have the strength to forego any preplanned climax between particular people at a particular place and time: which gives the game that much more reality, as such climactic moments are rare and almost never conclusive. Again, my postmodern sensibilities will be detected by the astute reader.) Modules X4, X5, X6, and X9 all suffer from the author designing areas much too large to be adequately stocked.
(If some writers display a resistance to the design utility of static locations, one can draw an analogy with proponents of "method acting" in film and theatre. While the DM may see the façade being presented, the players will not, because it is precisely their presence and activity which create movement and dynamism in the environment.)
A worst case scenario is given here. I have refereed a game using (West End Games') Paranoia adventure "The Yellow Clearance Black Box Blues", in which all the PCs are poisoned and end up dancing spastically and uncontrollably at the end of the scene. While completely hilarious to read (note that the author is a Nebula prize winning novelist), it was absolutely frustrating and boring to players who could only listen to my (unmasterful) attempts to regale them with their own humorous antics. The players were ultimately unnecessary to the scene, and they knew it. They had come, in contradistinction, to play a game.