I understand zlevels sounds odd but my theory is based off minecrafts astronomical success. A TON of minecraft development was spent on almost invisible or just light at first glance changes, which also opened Pandoras box for modders / mapmakers.
And face it, that was a huge part of its success. Yes zlevels is almost endlessly daunting, but if done properly?... The possibilities would not end. High cost yes, but VERY... VERY high reward.
a very good point in some ways, if terraria had a 3D world we'd surely see a much different popularity list, it's undeniable that 2D is better for platforming, but at the same time it's also pretty undeniable that for creativity/survival sandboxes 3D provides a much more long term interesting environment, just try thinking about it like this for the programming minded:
you have 100 tiles in which to hide something the player can find in some way or another, that's less permutations to worry about than there are brute forcing a combination lock (3 digit) in a 2D plane, that is to say that many places (not considering anything else) it could be placed.
now i'm going to just preface this right here and now, i'm amazingly bad at math so my apologies in advance but the general idea should still be true to the best of my knowledge
same deal, 100 tiles to use to hide something that must be found, except this time you have not only x and y (horizontal and vertical respectively) as well as mister z level, this means that at the very least when you had 100 total places something could be squirreled away in the previous example in this one you should have at least 300 permutations at the very least given that extra plane of perspective we call z-level (depth).