A given player only plays that first campaign once, and most of us don't want to then revert back to "Dwarf Fighter" as the full extent of our customisation the second and subsequent time we play. Additionally, most of us reach of point at which the level of complexity is right for us, and don't want to go beyond that point.
So, what that actually suggests is that the core game should be a nice, simple introduction to the game, with minimal complexity at the start, gradually increasing to a moderate level and then staying at that level for the rest of the level range. Then, add supplements that add complexity (but not power) at those low levels, and further supplements that allow groups to increase the level of complexity across the level range up to their preferred level (and no further).
Remember that it is trivial for a supplement to add complexity; it is nigh-impossible for one to remove it. So, keep the core to the minimum complexity you require to make the game work.
It's almost as if what's needed is some sort of, I don't know, modular system...
The trouble with trying to create a supplement that adds complexity, but not power, is that any sufficiently complex system has loopholes, or interactions that the designers did not foresee. I cite as an example True Sorcery, for 3.x (and while a single example is not a proof, it is an argument). It was built as an alternate casting system to increase versatility for arcane casters, but not raw power, at the cost of complexity. And it does a good job for the most part, but there are a goodly number of ways to break it, with terrifying results.
Even worse, when you start adding multiple supplements (or even modular sub-supplements), the number of interactions between the options in those modules grows exponentially, while playtest time stays more or less constant during the lifecycle of an edition (hell, playtest time per option probably falls over the lifetime of an edition; they test the hell out of core, and then they start slacking off). As a result, the fraction of option combinations tested by the developers falls drastically over time. Even if things aren't any more complex, systems will be broken.
However, I agree with you that something simpler that high-level 3.x, but not as simple as low-level play, is ideal. I recently declined an offer to join a 9thish level 3.x game due to a combination of RL time constraints and just not wanting to deal with the complexity (relative to Traveller, my current game of choice... which, upon further consideration, is actually really modular. Huh.).
But, to speak of actual high-level play:
I disagreed with his focus on artifacts and planes. To me, relying on external magic items for your power cheapens the awesomeness of high level. Sue me for an Iron Heroes fan. Likewise, I'd much rather high-level be about carving a kingdom out of the wilderness, waging war against neighbouring states, running guilds, things of that nature. Both an increase in scale and an increase in delegation seem in order. Planes to me are just an excuse to keep dungeoncrawling at a high level by providing a set of infinite dungeons stocked with infinite hordes of high-level baddies. I would rather see the game actually change, not just in backdrop, but in the focus and the way it's played.