I must admit I have to disagree here.
If there is 1 thing I completely look forward to star citizen for it is that it will be a game to actually test out PCs and to help people with high end rigs get some value out of them.
For too long now people with top top end rigs (who I am not one of at the moment) have had nothing to really show the value of their investment.
Sorry, but I don't think this is a valid argument. SC uses a modified engine that was created for a first person shooter. So basically everything that will be implemented in that game are "maps", essentially not different from the maps of Doom, for example. It's very easy to overload even high end systems by just building overly complex maps. This does not prove anything, and in my opinion it does not show people the value of their investment.
So when Mr Roberts states that he will bring even high end PCs to their knees I read it as "we are trying to archieve our goals with inadequate means, and you will have to pay".
The Cobra Engine is vastly different, because it was build with the idea to use the calculating power of modern CPUs for procedural generation of content. Modern CPUs are incredibly powerful calculating tools, even "outdated" ones are, when compared to what was possible 20 years ago.
Just look at the galaxy map we already have. Are you aware that you are looking at a structure that has the same level of complexity as the human brain? You can look at any part of that in real time, zooming in and out at will.
I'm convinced that this is just the beginning. You often hear that content that was generated procedurally is bland, boring and artificial. This is certainly true if the rule set from which the content is generated is primitive. But we are beginning to see what content could look like if the rule set imitates laws of nature, and this is only the beginning. We will see landscapes that are created on the fly by an imitation of continental drift, volcanic activity and erosion. We will see persons that represent their genetic outfit and the effects of aging.
The calculating power for that is already there. Progress will come by tapping into that resource, not by demanding more gigapixels of fillrate.