Let’s be honest about it. The past year of soul searching in the U.S. civil space program has produced at least one widespread realization, crystallized in the final report of the Columbia Accident Investigation Board: The goals of the nation’s human spaceflight program must be set high enough to justify the cost and the risk associated with the endeavor — and a few people circling Earth in the international space station does not meet this standard.

Few now argue with the essential thrust of President George W. Bush’s proposed Exploration Initiative — return humans to the Moon and proceed onward to Mars. But there is legitimate concern about whether, ultimately, there will be enough money in the projected NASA budget –assuming neither big increases nor dramatic cuts — to accomplish the task.

The Gordian knot of space exploration and its cost is the problem of launching all of the mass — much of it fuel — that needs to be put into orbit to accomplish any human exploration mission. Even a minimal Mars expedition will require a mass of some 500 metric tons. That is just about the size the international space station will be when it is finally completed after more than a decade of assembly requiring dozens of launches of today’s shuttle and Proton systems.

Returning to the Moon is a little easier; an Apollo-like capability such as sending a couple of people to the Moon for a few days can be achieved with about 100 metric tons in low Earth orbit (LEO). But clearly, any sort of extended lunar presence will require substantially more. Our biggest current launch systems are the two Evolved Expendable Launch Vehicle (EELV) families of rockets, each offering up to roughly 20 metric tons to low orbit at a minimum $300 million price tag.

Are we prepared to accept a system architecture that requires a minimum of five EELV launches, and extensive LEO assembly activity, merely to affect the most basic lunar return capability? I don’t think so, and it is beyond outlandish to consider assembling a Mars mission in LEO if more than two dozen EELV launches — at a cost of about $8 billion — would be required to do so.

And yet, many people are talking about using EELVs as the backbone of the space exploration launch infrastructure.

But if such scenarios are daunting, it is almost equally daunting to consider developing a new, clean-sheet-of-paper heavy lift launch vehicle. It is the right thing to do technically, of course, but from a budget perspective it seems likely that any exploration architecture requiring such a development will be dead on arrival.

If we still had the capability to produce the Saturn 5, we wouldn’t be having this discussion. The old Apollo launcher’s price/performance niche, 140 metric tons to LEO for about $500 million of today’s dollars, looks pretty good right now. But we don’t have it, and we won’t, so another path is needed.

The lesson learned from the Saturn 5 experience ought to be this: Don’t throw away large flight systems for which the development costs are behind us when no replacement appears on the horizon. But in our zeal to complete assembly of the space station, retire the shuttle and get on with true space exploration, I fear that we may be once again heading down exactly this path.

What is needed is to retire the Shuttle Orbiter and its expensive support infrastructure. It simply does not serve the needs of exploration and is too expensive, too logistically fragile and insufficiently safe for continued use as an LEO transport vehicle.

But the other components of the shuttle stack — the external tank, the solid rocket boosters and even the shuttle main engines — can form the nucleus of a very serviceable heavy lifter.

Such a choice would bypass at a single stroke the key obstacle to the new exploration initiative: how to get the necessary mass off Earth without posing completely untenable budget requirements.

Nearly every reader of this newspaper will of course be aware that studies of such vehicles have been ongoing for well over a decade. Many variants are possible, in two basic categories: the Shuttle-C design, which replaces the orbiter with a cargo carrier; and the Magnum design, an in-line configuration with a more conventional payload fairing.

Many payload options in the 70-120 metric ton range exist. Broadly speaking, it may be said that the Magnum design offers somewhat better performance and growth capability at lower recurring cost, while the Shuttle-C design makes considerably better use of the existing shuttle ground processing infrastructure.

But it is not important at this point to engage in a detailed discussion of the tradeoffs between one shuttle-derived launch vehicle configuration and another. What is important is to recognize that design studies have consistently shown that we can have such a vehicle for $3 billion to $4 billion in non-recurring engineering costs, with recurring costs in the $750 million to $1.2 billion range.

This latter figure is uncomfortably high, but at least it is for largely known and understood hardware and systems, not viewgraphs. The task before us, if we really want to bring about an exploration program, should be to tackle the challenge of reducing the recurring cost of utilizing shuttle-derived systems.

It will be far easier to do that than to initiate a return to the Moon, or an expedition to Mars, using EELVs.