This is an old and painful question. Achievement-oriented players typically object (loudly) to being "forced" to spend more than a few seconds going from Point A to Point B no matter how far in space these two points may be separated. For a lot of these players, the "world" aspect (travelling should take time) is far less important than the "game" aspect (any time spent travelling is time wasted because it can't be spent playing the game).
On the other hand, if you let most of your players avoid terrain, some undesirable things can happen:
- less congregating, which is good (less lag), but also bad (less socializing)
- less need for traveling merchants (which takes away a valuable economic activity)
- less need for guides who know an area (again, loss of an economic activity)
- less clustering of homes by roads (hurts immersion)
Either choice of travel options has consequences for everyone who plays. That's why you have players "telling each other how to play the game." Transportation has consequences for everyone.
1. Consider the no-instatravel option. The good thing about expecting all your players to deal with geography is that it gives the game world texture. By "texture" I don't just mean physical topography -- I also mean economic and storytelling depth.
When you define some areas as being harder to travel through than others, and you carefully regulate the transportation mechanisms available, you immediately generate a large number and variety of interesting choices for your players. For example, maybe dangerous places need guides. A player who knows an area could serve a useful (and possibly lucrative) role guiding other players through the area.
Or perhaps it's possible to learn specific transportation skills, such as Ferrying or Short-Range Teleport. Allowing players to learn and use such specialized skills lets your players help each other. That contributes to an interconnected and diverse economy, which is usually a Good Thing. (Though it's also true that such skills need to be of limited scope, and payment models need to be reasoanably grief-proof. But that's a whole 'nother essay.)
In short, letting your world have geographical texture means more economic opportunities for your players. It's not just a RP thing; it's about making sure your players can do a lot more things than just beat up on the local wildlife. That's not just a worldy feature -- it's about having gameplay options. As such, it benefits everybody who plays.
2. At the same time, plenty of players just aren't interested in such second-order effects. If it's too hard to get to the exciting content, they won't play... and (up to a point) that's bad for everybody who plays, too, because a living, breathing, dynamic virtual world needs people in it.
The trick is to let these folks have the specific thing they want, but cut back on the negative effects by imposing a cost to this benefit. (This follows what I believe is an important game design principle: Let players do what they want, but every action should have a proportionate consequence.)
I never played Shadowbane, but I've read that it had a feature called Summoning that could teleport other players to the summoner's location. I like that... but with a twist: allow some players to be able to summon other players, but give this ability a cost that inversely proportional to how many people are grouped with the summoner. If you're by yourself, it's going to cost you a bundle to Summon one other person. But the more people you're grouped with, the lower the cost per Summons.
The value of this approach is that it enables groups of people who like playing together to get together with relative ease, but it retains the beneficial effects of geography in other situations.
There are probably a lot of other (and better) ways to resolve this particular tension, but this one seems like it might work... I think!