OK - *If* we have to have a space station, it should be designed to be extensible and to support human habitation. The current one does neither.
Many people envision some monstrous rotating ring when they think of a space station - but we can't start that way. It'd be too big and too expensive.
It could start much more simply: with a hab and a (equal) counterweight separated by at least 3 cables 680m long (any 1 of which is sufficient to hold the hab or counterweight under Mars g). If we spin these about their center of mass at 1 RPM, we'll generate an acceleration in the hab equal to about 38% of the gravity we feel on Earth. This just happens to be equal to Mars-G, so we get some experience in living in that.
The above crude drawing is an example (close to scale) - a hab and a counterweight/hab. The system rotates around the space in the middle - where we have a nonrotating dock.
Now - this doesn't look like much of a space station - and it isn't. But it has a huge advantage: It is extensible. We can add another hab+ counterweight. We can add a zero-gee chamber at the intersection of the cables. Additional habs can be linked around the perimeter, with airlocks connecting them.
After a few years or a few decades, we might have something like this
We'd have to spin up and spin down the station to add new pieces - which we could do once a year. One requirement of new pieces would be that they are capable of accelerating themselves. The outer surface of the station spins at 35 m/s, so the external rockets have to be capable of a delta-V of at least 4X that for the mass of the hab - not really a huge constraint.
The above drawing shows 7 habs on a side, arranged in a linear fashion. As cool as it would be to eventually close the circle, it wouldn't happen for a *long* time because we'll first start expanding in the Z direction, out of the page, probably starting with hab #4. So, you wouldn't get 7 habs in a arc as shown above; they'd be in some sort of 3x3 (with a couple of holes) clump.
With an extensible, modular architecture (think tinkertoys), various habs could exist. Hilton could finance construction of a luxery space hotel (there'd be enough of a customer base at $1 million per person per week... we still need to work on that launch-cost issue). EU/England/Russia/China/India/Australia/Japan could put up their own habs. A greenhouse / farm hab could try to generate food. We'd have the Nike hab (featuring .38g basketball), the Disney hab, the Bill Gates hab, and the Hugh Hefner hab (located off the zero-gee chamber in the middle). If we have defined the requirements correctly, any country/person/company/group is welcome to build a module, finance the launch, & join the Borg.
If a hab is (say) 10 meters long, we could have pi*680/10 = 212 habs per level, and the number of levels is theoretically unlimited..
Airlocks + elevators riding on the cables could transport people from the habs to the nonrotating zero-g chambers in the middle. The dock would be there too.
As cool as it would be to put this in L5, or in orbit around the Moon, or even some sort of 2-year Earth/Mars cycling orbit - we shouldn't, for a number of reasons:
So - high LEO would be best. In fact - the current ISS could even be hooked to the 0g part in the middle (which would be the best fate for the ISS I could envision)
This is a great example of what I think NASA should focus on: part space engineering, part infrastructure.