Before Cataclysm came out, Ghostcrawler wrote a great post on what he wants from healing. Here's an excerpt:
Mana efficiency is ideally part of this calculus. If mana doesn't matter, then either your highest healing-per-second spell or your fastest spell (depending on the situation) is always the best choice. Increasingly in LK, it's just the fastest spell that wins because of the nature of incoming damage. If mana doesn't matter, then you aren't trying to hit the bullseye I described above because missing the bullseye (overhealing) has no consequence. If you use the proverbial bazooka to kill cockroaches, then who cares?
I would tease out three overall goals from this and other posts by the Blizzard developers:
- Mana efficiency should matter. Healers should choose spells not just on their healing per execution time (HPET), but their healing per mana (HPM). Healers should gear not just for bigger, faster heals, but for mana regen.
- Damage should be slow enough that healers feel free to use slower spells. In Lich King, a slow spell like Healing Touch could easily mean people die before the heal lands.
- Healers should not cast the same spell over and over. They should intelligently make choices about what to do.
Based on the normal-mode 10-man raids I have been in, I believe the game has accomplished this quite well in the current tier. For the current tier:
- Mana efficiency is huge. Regrowth is only used in emergencies, for example, due to its high mana cost. Resto druids are highly prioritizing mana regen on their gear and in their specs.
- Damage intake is slow enough that you don't need super-fast healing. Slow and steady healing is just fine. The main exception is Chimaeron, where Caustic Slime comes very rapidly in the phase 1 and is devastating in the phase 2.
- Healers have a lot of choice in spells. This is weaker than ideal for nourish versus healing touch, because their HPM are so close. However, there are lots of interesting choices in where to use swiftmend, when to use tranquility, when to cast a regrowth after all, and when to use tree of life.
The second two items look fine for future tiers. Damage will go up, but so will stamina, especially now that stamina gives 14 hitpoints per point. Furthermore, much of what is interesting in spell choice nowadays is about other factors than HPET versus HPM, so even if we switch over to only good HPET spells, we'll still have more to think about than in Lich King.
Mana regen is a different story. Since spells take a fixed amonut of mana, it seems inevitable that mana regen will eventually get to the point that druids can spam cast healing touch and rejuvenate indefinitely. At that point, which is pretty much top-end tier-11 gear, players have no more benefit from regen. Thus, I expect regen in Firelands and beyond to be more like the hit cap: it will be very important up to a certan amount, and after that you won't care.
Fixing the mana regen problem is pretty hard. To make healers keep caring about mana, it seems like we need a way to heal even more than they currently can, but at the expense of using more mana. One way would be if higher spell power meant you also went through mana faster. That would be a rather dramatic change in how healing works, though. A milder change would be if players could take a talent to get more healing but much worse mana efficiency. For example, 50% more healing but 100% more mana per spell. Players in the current tier wouldn't use it, because they'd end up out-of-mana very quickly. Players in later tiers would have to take it, though, just so that their max throughput can keep up. Once they take it, they care quite a lot about regen again.