It seems to me that the focus on colonizing other planets [edit: using closed/airtight structures] is misguided, and that building space stations with permanent populations would be easier, and would make sense to do before colonizing say mars. [edit: I am not considering terra-forming as an option in this scenario].
I know some builders prefer to build a house from scratch than work on renovations, as working around what already exists introduces another layer of complexity.
The same principle seems to apply here. Mars has a certain level of gravity, certain surface and subsurface characteristics, and chemical compisitions, which would react in various ways with the materials used in construction. There would be certain weather effecting sunlight availability etc, and you're stuck at a certain distance from the sun, and all this stuff would have to be factored into designing a livable habitat. Then you'd try it for real and find other things we haven't thought of, surely.
The void of space is the ultimate blank slate allowing a top-to-tail system to be designed for maximum work-ability and nothing else.
There's also the gravity disadvantage. Everytime you needed something from earth, or you would have to use resources to manage a landing, retrieve the craft, and possibly return it to earth. And that's assuming one way traffic... and why is earth supporting this large scale colony that doesn't send anything back? If we mirror the history of exploration and colonization on earth we would expect space colonization to be focused on mining and so on (I imagine space based manufacturing would also make sense, since it would be cheaper to land a finished bicycle on earth than all the ores and other inputs).
A space station could receive and dispatch craft using a fraction of the energy cost that a mars based colony could. And it would have easier access to asteroids and so on (which seem to be the most easily exploited source of mineral wealth in the solar system.
People have said previously that the energy cost of maintaining an orbit would be the problem, but it seems to me that's probably a more manageable and predictable form of entropy than the ones that would be encountered on a planetary surface.
EDIT: I agree this question has already been asked, or at least answered, here: What off-Earth colony would be easiest to build?
I do no think, though, that it is a duplicate of this: Why terraform at all? as I am considering a period before terraforming has been successfully achieved.