Now that we have decoded the human genome, why don't we improve it?

The question is at present theoretical but could well emerge as the hardest of all bioethical issues. Biologists routinely alter the genes of mice, with methods that are not yet acceptable for making inheritable changes in people, but one day genetic engineers may figure out how to apply safe patches to the human biological software.

Everyone would like to have children who are healthy, beautiful and gifted. But people vary widely in all these qualities, depending on their parents' genes, and the pure luck of the draw at conception when each child gets allotted a random selection of half the parental gene pool.

Most human genes exist in several different versions in the population: some of them are great to have, some so-so and some downright deleterious.

This month the Icelandic company Decode Genetics found three quite common versions of a gene called BMP-2, each of which considerably increases its owner's risk of osteoporosis and bone fracture.

Suppose it were possible to delete any bad version of BMP-2, and of all other human genes, in a human embryo, and to replace them with good versions, without any risk to health. Would that be the right thing to do?

Parents who made such a choice would know they had given their child the best possible start in life. However expensive the procedure, it would be cheap in the long run if it saved a lifetime of medical bills, and therefore could be made available to all. Life's most serious unfairness, the difference in genetic endowment, would be erased from birth.

"One day, people may view sex as essentially recreational, and conception as something best done in the laboratory," Dr. Gregory Stock wrote recently in "Redesigning Humans. " Parents may start to believe it is "reckless and primitive to conceive a child without prior genetic testing."

Yet there are weighty arguments for not making inheritable changes to the human genome.

On the practical side, many genes have more than one effect, and swapping out the bad version of a gene can have unpredictable complications. The new gene, for example, may interact badly with the person's other genes.

But if the elimination of disease-causing variants of genes should prove successful, there might be no holding the line against parents who wanted to enhance strength or intelligence as well.

Upgrading the imperfect human material is all very well, but handling the transition between the superpeople and the ordinary variety promises to be awkward. Social stresses may emerge, especially if the technology does not trickle down quickly and smoothly.

Soup up those genes for I.Q.? Altering the genes that shape human behavior is not to be undertaken lightly. Human nature is a subtle blend of contrary qualities, the only survivor of evolution's many disastrous experiments. What could justify the risk of messing with such a delicate brew? Can't we be happy as we are, just as nature has shaped us?

"The human body and mind, highly complex and delicately balanced as a result of eons of gradual and exacting evolution, are almost certainly at risk from any ill-considered attempt at `improvement,' " the President's Council on Bioethics wrote in a report last month on the dangers of enhancing the body's natural abilities.

As the products of evolution, people may seem churlish if they challenge evolution's wisdom. But of course, evolution has none. It is a blind process that depends on constant error to create occasional lucky accidents.

By culling the unfortunate owners of bad genes, evolution keeps animals healthy and vigorous until the age of reproduction, and a bit beyond for species that provide parental care.

But evolution's rigor at favoring good genes that act early in life is mirrored by a weakness in screening out bad genes that act after the age of reproduction. Because of this weakness, evolution has failed to eliminate the bone-fracturing variants of BMP-2, and the bad, late-acting versions of many other genes in the human genome. This is the very reason that we age and die.

If evolution cannot help us after a certain age, why should we not help ourselves? Should not everyone have a right to the best versions of the genes in our collective genetic heritage, or at least to be born free of the worse ones?

And yet, if we reduce genetic differences, we risk turning the human population into one giant clone, tedious to meet with and bereft of the variation needed to respond to changing environments. The pursuit of perfection, if carried to extremes, is a sure recipe for extinction.