That first presentation skips over the most fascinating piece of the universe's structure seen to date, the Sloan Great Wall
http://apod.nasa.gov/apod/ap071107.html It seems that at the very largest scales the universe shows a "foamlike" distribution of matter, with galaxies and galaxy-clusters concentrated in the walls between open bubble "voids". A cross-section through it looks rather like the scum left on the rim of a cup of latte.
Try comparing the NASA images with this:
http://alt-fractals.blogspot.com/2010/10/fractal-coffee.html It's fascinating because this structure wasn't predicted by current theory, and although people are now retrospectively invoking all sorts of hypothetical hand-crafted dark-matter-related explanations, these don't tend to work terribly well.
What
appears to be happenning is that the "voids" are expanding faster than the rest of the universe, so that matter ends up being concentrated in the walls between adjacent faster-expanding voids.
This interpretation is interesting because
(a) ... it does away with the "flat-floor background" assumption of general relativity, which was put in to make the universe appear a more tidy and less chaotic place. That may well be wrong. Cosmology may be chaotic.
(b) ... without the "flat-floor" assumption, some of the basic structural architecture of general relativity may be wrong. A foamlike large-scale structure is suggestive of acoustic metrics, where local physics doesn't conspire to produce a smooth uniform background except as a first approximation. A background that's foamy rather than homogenous suggests that a number of basic theoretical assumptions that are used in general relativity for simplicity's sake may be unsafe.
(c) A homogenous background leads to other assumptions that tend to confirm the idea of a homogenous background ... but a foamy background tends to lead to a more fractal cosmology in which faster-expanding voids can in turn spawn smaller even-faster-expanding voids. In that scenario, the entire observable universe might be a single, mature, almost spherical bubble that's attached to a larger spacetime structure. The fundamental physical constants in our bubble might not be the same as those in other bubbles, and from there it's a small step to make the constants a feature of location on the larger structure, with the interplay between the continuum of variable "constants" spawning lobe-like structures at given locations, which each inherit the constants associated with their point of origin.
With that sort of amended physics, and with the associated more chaotic spacetime structure, the regions between galaxies are also more tenuous and have a lower degree of geometrical connectivity that you'd expect from Newtonian mechanics projected onto a flat background, and galaxies get a greater say in the local physics of their region, leading to spiral arms that are more easily able to rotate with their galaxy without being torn apart and flung off. That difference between the Newtonian prediction for the shape of galactic spiral arms has already been noticed - it was the reason why we invented dark matter.
The combination of galactic rotation curves and large-scale "foam" suggests that instead of trying to explain these structures by inventing new forms of matter with arbitrary properties (like dark matter and galactic string filaments) to increase the degree of interaction in higher-density regions, perhaps we should be considering the effect of a diluted interaction strength with the outside world for signals passed through lower-density regions. The reason why we didn;t take this approach initially was because it leads to an unstable and chaotic universe, and when general relativity was originally designed, the mathematical ideal was considered to be a smooth, infinite, unchanging universe, not the dynamic expandng one that we now appear to live in.
But if we're not scared of chaos, or fractals, or large-scale variable-density geoemtry, we end up with a more fractal cosmology that's a little reminiscent of Lee Smolin's "fecund universe" idea
http://en.wikipedia.org/wiki/Fecund_universe, "black holes begetting universes containing more black holes" ... except that in
this model, the spawning happens as a consequence of lobing expansion rather than gravitational collapse.
Lobes spawning lobes, spawning yet more lobes. I wonder where I've seen that sort of pattern before ...