Are you familiar with the Snowbluff Axiom from these boards?
I am; however, I don't think it's especially relevant here. "Exploitability", to me, means the ability of a motivated player to substantially exceed the baseline via unanticipated use of the mechanics. I would tend to agree that exploitability is an unavoidable consequence of the kind of flexibility I look for in 3.5.

But my complaint with 3.5's class balance is not really about exploitability. That is, I'm not complaining that it's possible for a motivated player of a spellcaster to substantially exceed the baseline power of spellcasters via unanticipated use of the mechanics. I'm complaining that the baseline power of spellcasters is vastly greater than the baseline power of non-spellcasters.

There's some parsing that could be done here. If you believe that the 3.5 designers intended for the classes to be well-balanced, then you could say that a spellcaster that is vastly more powerful than the baseline power of a fighter must necessarily, by that fact alone, be "exploiting" the system by using the mechanics in a way the designers never intended. This argument is tautological, which is to say circular; it classifies all imbalances automatically as "exploits" and stipulates designer error to be a logical impossibility.

In 3.5, the power differential can manifest without any ill intent and without anyone feeling like they're exploiting the system. A novice player can demonstrate this fundamental imbalance by playing a core-only single-classed wizard. I don't think that this is inevitable in a versatile system the way that obvious exploits are. It doesn't bother me that the core rules contain an infinite-wish loop; that's much easier to regulate as a DM and as a group than deep-seated power gaps between same-level characters within a party.

In fact, given this baseline imbalance, exploitability is often a blessing. By jamming combinations of the abilities that would give the designers nightmares, you can build a non-spellcaster far above the non-spellcasting baseline, which can substantially reduce the power disparity within a mixed party of spellcasters and non-spellcasters. Of course, this is hardly an elegant solution. Doing this requires fairly deep system mastery, access to many, many sourcebooks, and a tolerant DM. It's also a lot of work, and it's not like 3.5 character creation is quick and easy in the first place. Nevertheless, the exploitability of 3.5 can be used by experienced players to mitigate its fundamental imbalances. (My biggest problem with Pathfinder is that by making the system less exploitable they made it more unbalanced.)

The goal of a systematic rebalance should not be to avoid exploitability. Snowbluff was right; I'd rather have a versatile system than a bulletproof system. No, the goal of a systematic rebalance should be finding a common baseline.

The best attempt I've seen thus far (that I might actually get to play due to adoption) might very well be Starfinder. By eliminating spells above 6th level and iteratives, I think it's brining the classes more in line with one another without the mechanical homogenization of 4th or the nebulous mother-may-I style of 5th. That is however a preliminary reaction, and I'm looking forward to continuing to ingest material to see if that first impression holds up.
I haven't read Starfinder, but I'm intrigued by what I've heard so far.

@death390:

I generally agree. What I think most need an upgrade are the "bones" of the system; the stuff that a Warrior 20 could do. The lack of support for obvious combat styles is a big problem, and a pet peeve of mine is the sheer number of feat taxes required to use basic mechanics.

Lackluster class features play a big role as well, particularly for the monk and paladin. There probably are more rewrites of these two classes than any others, and for good reason. The monk's dependency on magic items is also a major problem, even more so than the other martial classes. I'm hoping to come up with a simple way of addressing this.