I disagree. I view it as an issue of economics. As it stands, the economics of building an entire orbital facility for assembling ships isn’t there if you only count it for NASA going to Mars.
Rather, I would conjecture that it would be a public/private partnership - with various companies investing money in exchange for essentially time and tonnage coming out of the dock. The initial run would be largely preset and negotiated - focusing on simpler, more proven designs at first. As the model is successful, the yard can be expanded to build larger ships.
Why would companies do this, you ask? Because it would remove major barriers to achieving their long term goals - principally ones regarding infrastructure and complexity. Rather than have 8 or 9 companies scrabbling for 30 years to get to the point where they can afford to haphazardly duplicate each others efforts (Which they won’t - they’ll form partnerships on their own - but it’ll still time and cost consuming), we have one initial government led effort that helps build and launchpad the economic underpinnings - I.E. getting out there to mine, tour, and live - for those companies to begin generating ROI. Thus not only accelerating the commercialization of space travel, but giving NASA a long term role and influence that will last well past 2050, particularly in the areas of design evolution, operation, and best practices.
But to do this, it has to also be economical. Currently, NASA is congressionally mandated to re-use STS components to build the SLS - the Space Launch System. In my view, this is not an optimal or economically sustainable plan, but it works for NASA based on the premise that it will let them launch larger modules for a ship into space. Part of the problem is achieving economies of scale, particularly relating to the production of rockets.
Contractors tend to view these types of large, high budget, limited run items as risky. After completion they can well anticipate a drought or downturn in revenue as the program winds down. The incentive structure there leads to cost maximization, rather than reduction, in an attempt to hedge against the gap between Program A winding down and the next contract reaching Full Rate.
One of the ways around this is achieving a reliable volume of launches - giving the company enough certainty about the future of the contract. This is a problem that accompanies many high budget items, but it particularly affects heavy and ultra-heavy launch platforms - due to their generally high costs, achieving reductions from scale is challenging. Which in turn places limitations on the number of launches, due to the cost, which can further increase costs.
NASA recognized this problem, and attempted to correct it with the Space Shuttle. But due to the higher safety needed for man-rated launches, and the thus the attendant downtime for inspections, rebuilds, and maintenance, the Space Shuttle program largely failed in that aspect of its programmatic objectives.
My proposal is that instead, the currently invested ultra-heavy lift assets (SLS) be used to construct an orbital assembly yard, which will then be supplied by medium lift, re-usable rockets, carrying unmanned payloads with automated docking and navigation.
This would allow a higher volume of launches, over a longer period of time, allowing for better scaling and sustainability of launches, on top of those achieved from an RLV. The payload value would also be much lower on average, dramatically reducing the risk and insurance factors involved, as well as attendant safety concerns.
Naturally, all designs have limitations. However, certain shapes, sizes, and dimensions cannot be launched from earth, or are incredibly inconvenient to do so using currently infrastructure. By breaking those components down and assembly them in orbit, they can be moved to orbit more economically, opening up additional design methods that differ from those conventionally in use.
The station would be only staffed by 2-4 persons, the majority of the work being done by automated robotic arms on gantries, with supplies being stored and retrieved by a system of monorails on the perimeter of the station. The programming and blueprints would be vetted and test assembled by NASA on Earth (Using existing facilities).
The construction process would thus be largely automated, allowing the expensive, dangerous, and time consuming EVA missions to be used for selected, delicate or inspection and maintenance related tasks.
As to on-orbit refueling: It has in fact been done, and is routinely done to refuel station keeping thrusters on the ISS. It was tested during the Apollo program, but ultimately the LOR approach meant it was unnecessary. The way around this that I would posit is that the ship can be kept in an unfueled state until it is complete, at which time the manuevering thrusters can be charged off station reserves - a fuel depot can then be launched into an orbital window which would allow the ship to move to a safe distance before the automated depot conducts an intercept and fuels the ship for the first time. If problems emerge, it can then be taken back to the dock and serviced.
Additionally, on orbit refueling is currently a commercial service being explored by a few companies to extend the service life of satellites.