The dispute between NASA’s Ames and Marshall centers over who gets to manage the automated exploration of Earth’s Moon begs a question. We plan to send human scientists there soon, and NASA’s budget is stretched beyond the breaking point, so why are we funding any large automated missions to the Moon? All of the major outstanding questions involving prospecting for exploitable resources and understanding lunar geology require activities in which human miners, geologists and engineers are far better and more efficient than any foreseeable robot.

Consider the questions of the moment.

While any water at the lunar south pole just might exist as a block of ice on the surface, it is far more likely to be scattered through the regolith and buried deep underground. Finding it will not be easy.

An automated lander would obtain one drill sample at one site, or at best a small number of shallow cores over a small area. More than 30 years ago the Apollo-17 crew conducted an in-depth geologic survey of an entire large alpine valley in only three days. A modern crew would have more time, the ability to traverse many tens of kilometer s and far better tools. Those should include deep drilling equipment capable of obtaining multiple cores.

Detailed strategraphic exploration of crater ejecta, with absolute dates obtained over wide areas, will provide a history of asteroidal bombardment in the inner solar system of great scientific value.

In theory, an ejecta survey could be automated, but obtaining a useful set of samples would require very large numbers of rovers executing difficult, time consuming and ultimately expensive traverses. Obtaining absolute dates requires complex chemistry laboratories on site, or returning a large set of samples to Earth. Studying the history and extent of lunar volcanism also involves the kind of field geology that is difficult and expensive to effectively automate.

Perhaps the most important scientific reason to return to the Moon is to try to find samples of Earth’s earliest continents. These would have been splashed up in the giant impacts that peppered the early Earth, and some of them would have hit the Moon. Such samples could contain unique evidence — long since destroyed on Earth by later geologic activity — of the formation of life, possibly including fossils.

Likewise, for much of the lifetime of the solar system, Earth’s Moon has been a nearby static trap, capturing a sample of the same debris that struck Earth. Samples splashed up from other planets throughout the solar system’s history, and even from interstellar space, are likely to be preserved in the lunar regolith within relatively easy reach of Earth.

Collectively, these samples from Earth and elsewhere could prove revolutionary to geology, biology and paleontology, and to several branches of astronomy.

Such samples almost certainly exist but they must be rare, widely scattered and probably deeply buried. No series of robotic missions, however elaborate, is likely to find them. Only detailed Lewis and Clark -class expeditions, featuring geologists able to travel great distances, could have the scientific efficiency to have any chance of finding these samples.

The central peaks and rims of impact craters contain materials excavated from farthest below the surface. Apollo astronauts easily and quickly climbed slopes of greater than 20 degrees, covered with loose dust and talus, to reach these locations. They identified and collected relatively unmodified samples from deep within the Moon. In a future mission, deep drilling would put the surface collection in context. Once again, complex mobile laboratories would be required to analyze and date the samples, or large numbers would need to be returned to Earth. None of these tasks can be effectively automated with reasonable efficiency.

In fact, why are we spending more than half-a-billion dollars sending an ultraviolet mapper to orbit the M oon? Trans-Earth propulsion modules will soon be parked in orbit while astronauts explore the surface. If the lunar base is at the south pole, these modules will wait in polar orbit. They could easily piggyback mapping instruments in a manner similar to the scientific instrument bays flown on the advanced Apollo Service Modules. Each Apollo mission attempted to map the areas planned for the next landing — a process that worked very well. Why have we failed to learn from that successful experience? Many argue that sending precursor missions to land on the Moon is essential to make human flights safer, but those precursor missions flew more than 30 years ago. Astronauts have landed safely on the Moon and we know how to stay alive there.

Meanwhile, many planetary scientists have agreed that their highest priority is a mission to Jupiter’s ocean moon, Europa. While we are wasting billions in a forlorn attempt to automate lunar prospecting and geology, we are not sending a reconnaissance mission to Europa — or to the many other worlds where we cannot send geologists anytime soon.

In a well-planned space program, automated landers should only be sent to locations where scientist-astronauts cannot go within the next few decades. Robots — and the severely limited ground-truth science they can accomplish — should be reserved for where there is no other choice.

Everywhere possible, send geologists on site, with their inherent abilities to quickly and efficiently do detailed and in-depth field work. Today, the locations that should wait for geologists include Earth’s Moon, the nearest of near-Earth asteroids, and possibly the martian moons — all likely to be within relatively early reach of the transportation architecture being developed by Mike Griffin’s NASA.

The only exceptions should be for very low-cost mapping missions or to test fly a lunar lander. Since we already have landed on the Moon many times with both human and automated spacecraft, and did not need to test the Lunar Excursion Module 30 years ago by landing without a crew, why do it now? The first return mission is unlikely to land directly in the unknown territory of the lunar south polar craters, but nearby, probably in terrain not unlike that experienced by the last four Apollo flights. The crew could then carefully drive into the permanently dark territory inside the polar craters.

If we do send a test lander, it should carry experiments designed to learn to lower the cost of human expeditions, or increase their ultimate scientific efficiency, not for near-term science. The most useful experiment for such a test mission would be a prototype plant to extract oxygen and other useful elements from the lunar regolith.

If humanity is to explore any part of the solar system at reasonable cost, we must take measured risks. The time to start taking those risks is now. NASA should immediately cancel all large automated lunar missions.

The agency should spend some of the savings on restoring automated missions deeper into the solar system and the aeronautics funding NASA has been forced to cut.

Most of the money should be used to speed scientists on their way to the mountains of eternal sunlight at the south pole of the Moon.

Donald F. Robertson is a freelance space industry journalist based in San Francisco.