"In an age when space flight has come to seem almost routine, it is easy to overlook the dangers of travel by rocket and the difficulties of navigating the fierce outer atmosphere of the earth," President George Bush remarked in his announcement of the shuttle Columbias re-entry disaster. "These astronauts knew the dangers, and they faced them willingly, knowing they had a high and noble purpose in life." In the aftermath of the accident, one of the sentiments most commonly expressed is that such tragedies remind us that the cost of exploring new frontiers is measured in human lives.

The idea is true, but it would be wrong to let it hang unexamined. One hundred and thirteen times we have sent a shuttle into space. Twice weve seen catastrophic failures and the loss of all hands. How do those odds sound?

They sound just about like what you would expect, based on the history of rocketry. As described by Theodore Postol, professor of science, technology and national-defense policy at the Massachusetts Institute of Technology, even the most reliable space vehicle systems (booster rockets) have roughly a 2 percent failure rate. At that rate, if the shuttles continue to fly five missions annually, then we should expect to see a Columbia or Challenger disaster every decade.

NASAs stated estimates of the risk of failure for any given mission are one in 145, or 0.7 percent, and its engineers and scientists strive constantly and heroically to make it zero. Still, the sheer numbers of shuttle flights work against them. (Columbia alone saw space 28 times.) Assuming that NASAs 0.7-percent-per-mission risk estimate is correct, then over 113 missions the likelihood that one shuttle will be destroyed reaches about 55 percent.

No one may know with certainty how best to calculate the odds against the shuttles because risk determination can be such a complex and uncertain science. One finds little reason for confidence that the number could have been much lower when reviewing the shuttle programs history.

For all that the shuttle can be saluted as an engineering marvel, it is also the spacegoing equivalent of a camel--a committees attempt to build a horse. When conceived in the late 1960s, the shuttle was to be a modest reusable craft for inexpensively delivering small crews to a space station and returning them. Shuttle and station were supposed to be stepping-stones to Mars. That original space station project was scuttled, but the shuttle survived, with a less defined purpose. New missions were soon found for it, however. Researchers wanted the capability to perform more experiments in orbit, so room for those was added. Using the shuttle to launch commercial satellites looked like a source of revenue. The Pentagon wanted the craft to carry military satellites, and so the cargo bay was enlarged further.

"Mission creep" gradually transfigured the shuttle, enlarging its cargo bay and morphing the coupe into a pickup truck. The engineering consequences were not minor. Building a reusable craft also proved more difficult than anticipated.

None of this history or the estimates of calamity are obscure; if anything, they littered the media back during the early years of the shuttle program and immediately after the Challenger explosion. The Challenger, of course, was felled by faulty O-rings, and physicist Richard Feynman and the Rogers Commission that investigated the crash argued successfully that this flaw was the result of bad planning and inadequate oversight built into the program. Problem identified, fix applied, and the shuttle program was far safer for it. Still, an unintended consequence of the defective O-ring discovery may have been to distract the public from the fact that, beyond problems of error or incompetence, the shuttles jack-of-all-trades design and operational profile pose fundamental safety problems that virtually guarantee eventual disasters.

Of course, even though the Columbias wreck has refocused the issue, perhaps awareness of the 2 percent loss rate doesnt change anything. Such losses of life (and equipment) might be acceptable because space exploration is such a spiritually rewarding pursuit, one befitting a great civilization. A human presence in space is symbolic, if nothing else, of our species curiosity and deep-seated drive to explore, some of the noblest motivations in our nature. In past centuries, mariners dared to sail the oceans in search of new lands with far worse odds of survival. The astronauts themselves certainly knew the dangers. Perhaps as a society we should not quail from the risk now.

But if we do value astronauts lives, it would be immoral not to reexamine what we are gambling them on and whether the odds might not be improved. One dead crew per decade is a high price for a manned space program with uncertain aims and uneven scientific value.

One strategy is to continue with the current shuttles while whittling away at the foreseeable dangers every way we can. Investigate the causes of shuttle accidents and prudently increase the margin of safety. A number of ideas are worth considering: enabling astronauts to diagnose and repair shuttles in orbit more easily; installing escape pods for rapid evacuations; making sure all shuttles can dock with the space station, and many more. Minimizing flight crews and replacing shuttle flights with unmanned missions would also reduce the threat. In short, look at what the benefits and costs would be, and then decide just how much we are willing to spend to prevent another Columbia.

The more drastic alternative is to address the NASA manned space programs current lack of specific, highly targeted goals, even if it ultimately means scrapping todays shuttles and half-built space station. In the absence of such goals (and funding appropriate for them), it is almost impossible to evaluate the current program and its trade-offs. New goals for NASA might be as ambitious as landing someone on Mars or as relatively simple as encouraging space tourism. Whatever the case, let the agency concentrate on building the right vehicles for the job.

Even a new space program will carry a risk of dangerous failure. One could hope that it would be far lower than that of the shuttles, but there are no guarantees. Even so, the scale of the catastrophes might be smaller; the numbers of flights involving people might be lower; we would have the cold comfort of knowing that when vehicles failed, it was not because they had been burdened with design requirements that raised the inherent risk.

Such a plan would be expensive and probably yield little scientific knowledge at first (the Apollo program did not provide much good science in its early phases, either). But it would be a program that more intelligently balanced respect for human life against the value, both scientific and spiritual, of venturing into space. Lets salute the bravery of the astronauts, not insult it by asking them to risk their lives in a poorly considered cause.