A far better vision would be much more expansive than Space X's -- which in my opinion consists of nothing more than building well engineered reusable reliable rockets at affordable prices.
Some guidelines:
1. Never use a rocket for material you can hurl or lift into space (i.e. non-G sensitive "mass").
2. Never use humans when robots can do much of the work (i.e. systems assembly, parts replacement, etc.).
3. Minimize the risks that humans face (keep them out of space as much as possible or well sheltered from the hazards there).
4. Invest only once. Build the factories to use materials from space in space.
You would start with (1) by throwing out the idea of rockets that can lift increasingly larger payloads. Instead you would invest one or more times in building ocean-equatorial based rail/mass guns [7] (to launch fuel, H2O, O2, food, "station"/"factory" subunits using solar power. This would lead to the construction of orbiting sky hooks which could augment the mass guns and/or pick up astronauts from SpaceShip Two type "ferries". Then SpaceTugs pick the astronauts up from the hooks and relocate them to ships under construction in "Dry Dock" (@ L1|L2).
But before one wants to engage in a vision like this one needs to *seriously* have a discussion regarding when molecular nanotechnology, i.e. when can nanofactories build nanorobots, when can nanorobots build nanofactories (allowing exponential expansion either on the Earth or in space). Nanorobots and nanofactories significantly lower the costs of access to space as well as the development of space (because they eliminate the need for biological "human" environments, safety systems, resource supplies, etc.). So one has to face up to the question of whether we want "human" or "nanorobot" development of space (when one path is clearly less expensive and likely to be more efficient), though perhaps less emotionally fulfilling.
Many engineers 'dis molecular nanotechnology, but for people who understand genome biology, that genomes are "software", that enzymes, esp. DNA polymerase, RNA polymerase and the ribosome are "assemblers", and who may have read Drexler's 1981 PNAS paper in which biological systems were cited as existence proofs for molecular nanotechnology, and perhaps who have read Nanosystems as well, the only questions that remain are how and when we could engineer systems of such complexity.
Then the question becomes whether we spend billions of $ on 40-50 y.o. visions (rockets to the moon or Mars) or equivalent or even greater amounts on say a 11-29 y.o vision... [1]. It is clear, at least to me, that the 40-50 y.o. vision provides some great stories, improves our technologies and lets us go where we have never gone before. In contrast the 11-29 y.o. vision frees most individuals on the planet from having to ever work again to survive, may indefinitely extend their lifespans and enables the evolution of humanity from a pre-Kardashev Type I level civilization to a Kardashev Type II level civilization [6].
I know which vision I'd be inclined to vote for.
1. Drexler's PNAS paper was published in 1981 [2]. Engines of Creation (Vsn. 1 was published in 1986) and (Vsn 2.0 published in 2007) [3]. Nanosystems (Eric's MIT PhD thesis) was published in 1992 [4]. Nanomedicine Vol. 1 by Robert Freitas was published in 1999 [5]. Almost all other nanotechnology "literature" tends to be long on either speculation or technical details and short on "vision" and facts. Those are the references for "science "visifact"ion.
2. http://www.pnas.org/content/78/9/5275.abstract
3. http://search.barnesandnoble.com/Engines-of-Creation/Eric-Drexler/e/9780385199735
http://en.wikipedia.org/wiki/Engines_of_Creation
http://www.wowio.com/users/product.asp?BookId=503
4. http://search.barnesandnoble.com/Nanosystems-P/Drexler/e/9780471575184
5. http://www.nanomedicine.com/NMI.htm
6. http://en.wikipedia.org/wiki/Kardashev_civilization
7. http://www.popsci.com/technology/article/2010-01/cannon-shooting-supplies-space