Nearly a decade after its creation, the Americans With Disabilities Act of 1990 is a case study in the limits of social engineering. Last week the Supreme Court either gutted the ADA or imbued it with common sense -- depending on your point of view. The court ruled that people with common problems (eyesight that requires glasses, for example) aren't "disabled" under the law. But even if the court had gone the other way, it couldn't have rewritten the ADA's record of underwhelming achievement.

The ADA aimed to bring the most disabled Americans into the mainstream by getting them jobs. Work (the logic went) would increase income, limit dependence and raise self-respect. So the ADA outlawed job discrimination against the disabled and required employers to provide "reasonable accommodations" for otherwise qualified workers. Despite these protections, the ADA has raised the employment of the most disabled only slightly.

The Census Bureau says that about 9 percent of the working-age population (those 20 to 64) are "severely disabled" -- meaning, generally, that they use a wheelchair or cane, have serious hearing or sight problems, or are mentally retarded. Among them, employment rose from 23 percent to 26 percent between 1991 and 1994. A private poll by the Harris Survey in 1998 suggests a similar job level. Conceivably, the increase (800,000 individuals between 1991 and 1994) could stem from the ADA. But it could also reflect the economic recovery.

This does not mean the ADA has utterly failed. John Lancaster, executive director of the President's Committee on Employment of People with Disabilities, puts it this way: "The ADA [has altered] attitudes about people with disabilities: what they can and can't do. . . . They're able to get around better. You see more [disabled] people out there. The attitudinal things break down when you see people as people and not as wheelchairs."

Point taken. The ADA also has made the public landscape friendlier for the disabled. Businesses and governments had to make themselves more accessible. Doors were widened, curb sides lowered, toilet stalls enlarged, ramps installed and parking spaces reserved. A lot of this was uneconomical. (The commuter buses I ride were fitted with wheelchair elevators; I have seen them used only once.) But the changes not only helped the disabled; they also signaled a larger social acceptance.

Still, the employment effect is weak. Why?

Laws can do only so much. Even before the ADA, many severely disabled people worked. Character and circumstances count. My best friend, a newspaper editor, developed Parkinson's disease a decade ago. He never thought of quitting, and even without the ADA, I doubt his paper would have fired him. He's too competent and conscientious. But the ADA rendered the issue moot by requiring employers to make "reasonable accommodations."

The costs of doing so are probably not high. In about 70 percent of cases, companies' expenses are less than $500, says the President's Committee. What's false is the accompanying assumption that the disabled lack jobs mainly because employers are indifferent, ignorant or hostile. The messier truth is that some disabled can't work -- and some don't want to. What to do?

Well, provide a safety net. In 1998, Social Security's Disability program had 6.3 million beneficiaries and cost $48 billion; the Supplemental Security Income program had 5.3 million disability beneficiaries and cost $24 billion. Most also got government health insurance, Medicare or Medicaid.

Here lies the great contradiction. The ADA wants the disabled to work. But if you pay people for being disabled, more will say they're disabled and won't work. This is especially true of people with low skills, whose job prospects and potential wages are poorest. Among "severely disabled" college graduates, half worked in 1994. Among high school graduates, less than a third did. The easiest way to increase work among the disabled would be to reduce disability benefits or raise eligibility requirements.

This isn't likely. It would seem heartless. Indeed, the term "disabled" has grown looser over time. Many advocates who want the disabled in the mainstream also want more people considered disabled. They see little conflict between demanding that the disabled be treated like everyone else, while insisting that more physical and mental problems be labeled disabilities, entitling people to special treatment.

In the 1980s, Social Security tried to hold down disability costs by tightening eligibility. The effort collapsed after a political outcry and adverse court rulings. As a result, Social Security's disability rolls include people who can't work (a fourth die within five years) -- and many who could. The White House would cope with the dilemma by allowing many disabled to keep government benefits while working. The Senate passed a bill to permit up to six years of Medicare coverage for beneficiaries who work. This skirts the hard political, administrative and moral problems of critically inspecting who's disabled.

A similar dilemma afflicts the ADA. The broader the definition of "disabled," the more the law becomes a tool for the already-employed to raise their pay. In one case before the Supreme Court, two pilots at regional airlines applied for better jobs at United, which rejected them because their uncorrected vision fell below the company's standard. In another case, an auto mechanic was fired because his job involved driving trucks and his blood pressure exceeded the company's requirement. He got another job. The court ruled these workers weren't disabled.

That's common sense -- but perhaps not congressional intent. When it passed the ADA, Congress didn't strictly define "disability." The law exudes grand ambitions and vague language. To be precise would have highlighted the central dilemma: The problems of the seriously disabled aren't easily solved; and the problems of the mildly disabled aren't very serious.