I think at this point it's pretty clear that the word "complexity" is more subjective than objective.
I think it is vague, poorly defined, and we don't agree on what the definition is... but I don't think that it is subjective. I just think the idea represented in 'complexity' is complex enough that it is hard to offer up an accurate description. Competing definitions don't necessarily mean that there isn't something real underneath the confusion. It just means no sees clearly (yet).
(A related problem is that 'complexity' might actually be a superset word, and that there are many types of complexity which themselves need to be defined. Certainly, this is how the problem is being attacked for now.)
This problem isn't limited to RPGs. Defining complexity is an important topic in several fields. We tend to "know it when we see it" but we can't yet offer rigorous definitions of it.
One thing I've noticed is that "crunch heavy" games tend to only be as complicated as the GM and players make them.
It's certainly true that tables tend to ignore any rules that they feel are too complicated. 1e AD&D might be one of the definitive examples of this, but I'm sure it happens with many rules sets. Some systems are deliberately modular, where they offer up different levels of complexity you might want to play with.
I'm a rules tinkerer myself. There is hardly ever a game I play that I don't end up extensively house ruling to try to make the rules clearer, more flexible, easier to apply, more balanced, or whatever. My 3.0e house rules tend to make the game slightly more complex than RAW 3.0, while at the same time reduce a ton of the rules bloat in spells, prestige classes, feats, options and so forth. 3.X in my opinion suffered from the problem of there being more than one way to do things, which is almost always bad design. Some things got more complicated (my flanking rules for example), while the overall system has vastly less 'rules'. I've recently played and been thinking a lot about Mouse Guard, a rules set I very much have a love/hate relationship with, and I'm almost certainly going to rewrite those rules into a rules light set of rules in the long run as there is a ton of needless complexity in the rules, bad math, and poor rules flexibility. So, in a very literal sense, when I play a set of rules I end up making them as complicated as I want them to be.