Just recently, a friend directed me to a link discussing how "fun" and interactive playing a healer is in an MMORPG like World of Warcraft. The healer role in a raid is often stereotyped as, essentially, a whack-a-mole game with the various health bars of people in the group, and, perhaps, the description is not too far off.
The problem needs to be defined, first though. In World of Warcraft, most of the healing is reactive. The target of your healing takes a hit, and as such, he is at a health deficit. You target him and cast a healing spell to attempt to minimize that deficit. There are two mechanics present in the game for doing this -- direct heals and healing-over-time effects. There are a few proactive damage mitigation/healing skills, the most notable of them being the priest's Power Word: Shield and Prayer of Mending. There are a few other skills with similar effects, but tend to be on long cooldowns (Pain Suppression, Hand of Sacrifice), or are merely time-savers (Earth Shield).
As one might guess, there is a disproportionate bias in favor of reactive healing mechanics in the game as it stands. This, perhaps, is the first part of the problem with healer design in this game. Reactive healing does take a certain degree of the skill, especially in timing when there's a significant burst of incoming damage, however, it generally is always effective. It's hard to heal the wrong person, since the heal only happens when there is an actual health deficit.
What advantage do proactive defensive mechanics have over reactive ones from a design standpoint? First off, proactive measures, to be used effectively, require knowledge of how the opponent works. If you throw up a shield on the wrong person, the shield is wasted. Mana and time are expended putting it up, and the effect is not consumed. Secondly, they must be timed appropriately. Against an enemy that actually pays attention to the effects on his targets, shielding someone can act as a deterrent, enticing him to switch targets since he may know that any attacks against that particular character will be blunted; therefore, the shield must be placed on the target as close to the actual attack as possible. Additionally, putting a shield on after the fact is of minimal use -- the attack has happened already, it can no longer be mitigated. As a result, a player must actually pay significantly more attention to a battle and the pacing of the combat to make his skill usage effective.
Costing of such spells is also tricky. Proactive effects, in general, should always be more efficient than their reactive counterparts, as effective usage of a proactive skill requires a higher level of skill on part of the player than a reactive one. Most of the reactive spells as it stands right now are so much more efficient than shields or such that if it the proactive effects are almost unnecessary.
Now, having at least defined what we consided to be a more engaging design for healing effects, we come to the main problem, which, perhaps surprisingly, is not actually related to the skill design in and of itself. It stems from the actual encounters involved. PVE encounters in WoW are basically a matter of adjusting the rates if incoming damage vs. the rates of incoming healing, modified by mitigation available. "Only tanks should be getting hit!", or "just get a druid to pick up the raid damage" are symptoms of a type of design that makes players assume incoming damage to anyone besides a tank stems directly from a mistake. As a result, proactive healing takes a backseat to simply using standard healing mechanisms to keep tanks alive while the rest of the encounter happens. There is no large difference in proactive mechanisms and reactive mechanisms on a tank, since bosses attack extremely predictably, and any shields or similar effects are more or less guarunteed to be consumed in an optimal manner. (tl;dr: the tanking mechanic as a whole forces healing into a certain optimal setup.)
Let us conjure up a hypothetical combat scenario. Instead of a boss that smashes your tank in the face, you have a group of enemies that need to be dealt with. They have a better method of picking targets. Instead of smashing on a tank, arguably one of the least enticing targets, they attempt to head for your healers or DPS. To stop them, DPS can snare them, or whatever along those lines (before someone mentions it, the issue of crowd-control effect design in WoW also needs to be looked at in-depth at some point. For now, please imagine that they are in some reasonable state). Taking a hit on a DPS or healer wouldn't be a one-shot, but it should be strong enough to provide a severe incentive not to get hit. As a result, when enemies switch targets or when people are positioned in dangerous places, proactive healing once again becomes useful, as both a way to blunt incoming damage that couldn't be avoided, and as a way to dissuade attacks on specific characters.
So, while healing characters in WoW could be designed much better (and, in fact, should be designed much better -- one hopes Blizzard is at least working towards such a goal!), the bulk of the issue with healing in PVE scenarios stems mainly from the lack of incentives to use specific healing effects in efficient manners due to encounter design.