I have this impression now, playing for 5 days and don't be matched vs shaman once, the class just disappears.
Maybe I was the first one to complain about the insanely overpowered deck Galakrond shaman was in the release of DoD, only 3 days after and the ladder was more than 60% shamans, I can say is the worst possible meta, when one class is Tier S and you have to choose lose or play with the broken class, happen something similar in days of midrange shaman.
So, they nerfed the class twice, the first wave are a must but I don't sure about the second one but maybe it is just impossible to do this right.
Shaman always is Tier S invencible or trash tier unplayable, no mid ground?
I think this just highlights how difficult it is to balance games. Alot of players (with zero experience in game design) think it is trivial, but it isn't... not by a long shot. So, practically speaking, it's impossible. Especially given that you can't please everyone in any game, and in this one, when you change one card, you affect 3 modes (standard, wild, and arena).
I agree that the first post-DoD release nerfs to shaman were necessary. Shaman was put more in-line with other decks and classes. Galakrond shaman then started using the frog package, and while the dragon's pack nerf definitely hurt, I think Invocation of Frost was the real kicker. It threw a wrench into the flow of spells for the frog package, along with a eliminating a good turn 1 play with the card.
Then you have the known fact that shaman (like priest) has one of the weaker basic / classic set of cards of all classes. So once you hit current rotation good cards for shaman (or priest) with a nerf, they can't "fall back" on their basic / classic set like other classes (*cough* rogue *cough*).
On top of all this, card games are far less granular for adjustment than say an RTS or FPS. In those games, having fractions of numbers is fine, or having large numbers to give greater granularity (for the minimum change of 1 integer) works too, but using such large number / factions / decimal points is far less practical or enjoyable to play in a digital card game. I mean, just consider Fiery War Axe: it went from one of the best weapons in the game at 2 mana, and in pretty much every warrior deck regardless of archetype, to being not a card in any relevant meta deck for years at 3 mana. And that was one of the smallest changes they could make to the card...
There is no such thing as a "balanced" meta in my opinion. Only metas that have some imbalance versus those with alot of imbalance.
It's because they have a broken core card set (basic + classic). There are essentially no core cards in that set, there are no build up cards and you could argue there is no identity to the class. However, there are loads of finishers and super efficient burst. That means that in order for Shaman to work, you need to introduce in large numbers very strong core and build up cards in expansion sets. Once there's enough of those, Shaman suddenly becomes broken because of all the available burst. It's all or nothing with this class.
It's only impossible to balance a class, insofar as it's impossible to match your idea of "balance". What is "balanced" supposed to mean? What is your "middle ground"? Good, but not too good? Common, but not too common? High variety, but not viable in all ways? And more importantly, is it really about classes or decks or the metagame as a whole?
Some would say, "balanced" can't coincide with a class having two or even three strong decks (like Hunter currently), because that would be a "clear sign" that the class is "too strong", regardless of how strong excactly the decks are. However, some might actually see multiple strong decks of a class as proof of balance, because it would mean that different themes in a class are equally well developed, giving you more options to choose from instead of relying on one specific trick that may or may not continue to work out when the meta changes. So, already it is debatable whether "balanced" needs to mean, that a class has only one good deck, or many. And then it's another question how good the decks are allowed to be.
To some people, "balanced" means that a class has only "tier 2" decks, because that means the class isn't at the top but still viable. To others, "tier 2" might not be good enough, perhaps even too weak, if you have a decent chance to beat other decks, but are likely to lose against the best (and usually most popular) ones. But if a class can only be balanced if it isn't at the top, it would mean that a class can never be balanced AND strong, which is a bit silly. It would mean that a class is only balanced if it's significantly weaker than others; instead of being as strong as others. Not exactly what I would consider as balance.
Part of the problem here is that everything at the top is almost automatically considered as "too strong" and "cancer" and "needs to get nerfed". However, for as frequently as you hear people complain about something being "broken", it's not easily identifiable what is really unfair. Even the best decks usually have poor matchups, and their advantage is that their counters are generally not as viable or have more variance in their performance. Is the deck "too good", or is their counter simply not good enough against everything else? In that case, you can't really talk about whether a class or even a deck is balanced, but whether the metagame is. In other words, there would be no unfair decks, only insufficient counters.
To the developers, "balance" actually depends on the prevalence more than on the powerlevel. To them, a deck with a 40% playrate and a winrate of 54% is more problematic than a deck with a 90% winrate and a 0.2% playrate, because people complain about the decks they see the most, and hear the most about, not about decks they almost always lose against but only see once in a blue moon and that very few have even heard of. 54% isn't a huge winrate, but when you'd see the same deck several games in a row, you will lose eventually, and if nothing else be annoyed by always seeing the same cards. And the developers care less about classes being "balanced". While it can be considered problematic if 50% of all players play one out of 5 different Rogue decks (and each 10%), they don't think it's necessary that there is always one good deck for each class, as long as different playstyles (and thus counters) are represented.
Bottomline, it is indeed pretty much impossible to have a class balanced. It's difficult enough to define what would make a class "balanced" in the first place, and what you end up with would likely go against sentiments of most players: A deck, or class, is either decent but unremarkable, thus inferior to others, or about to become the next deck that everyone plays, turning it into a "problem". You really only have it balanced when every class is exactly on the same level, and that is very unlikely to achieve, especially with a design team, that sees no reason to force equality.
The other way out is to just give up on the whole "balance" idea and only point out decks that really are sporting a very high winrate against almost all other decks over an extended period of time (that is more than a week or two), and in some rare cases, there are decks that warp the metagame so much that they are actually problematic on a deeper level. These are the exceptions. For all the other cases, just deal with it that there are decks and classes that will periodically be better or worse than others.