So several points:
1) If JKA had made JKR consistent enough to dominate the meta again, that would be a bad thing - agree
2) BUT ensuring metas do not overdominate is not mutually exclusive with releasing content that doesn't need immediate rework, that doesn't cause players to invest resources and then have that investment be nullified. (I'm assuming the response to this is your final point, "tweaking things when design goals collapse... is also game design" will get to that in number 4)
3) "Putting Anakin on a Kevan team was not a triumph of clever theorycrafting". Okay, agree, but what about palps?
I honestly don't know the answer to this question. I can see what you are getting at with the JKA vs Magmathrawn comparison, but I can't draw clear lines between theorycrafting and not theorycrafting. If the solution is easy (one read of palps zeta can tell you he has potential) is it still theorycrafting? Also if the palps discovery was easy, why didn't the devs see it? This seems to be a long, separate discussion.
4) Just because something is game design, doesn't mean it isn't also punishing theorycrafting. To say "tweaking things when design goals collapse is not punishing theory crafting. It's also game design." seems like a false dichotomy. Something can be both punishing to theory crafters and valid game design. That being said, because the term theorycrafting is a bit subjective, we can still draw a clearer line using player loss instead. To illustrate:
Devs release content > Player invests resources in counter devs didn't see > Devs use valid game design tactic and nerf counter in very quick time frame BUT do not refund recent investments in said counter > result is player loss of resources without time to reap benefits of investment (along with preserving the desired meta)
The meta is designed yes. Not a lot of the game is organic, sure. Tweaking is game design, certainly. But is tweaking at the cost of player loss good game design? Even the developers seem to think otherwise as evidenced by their desire for a better, more rigorous, testing process before the release of content.
