EVERY FEW YEARS Congress, with the help of various experts and advisory commissions, enacts legislation to put the Social Security system on a sound footing. Not many months later the first cracks generally appear in the newly laid foundation, and debate begins again. Last week, the House subcommittee on Social Security took the first steps toward yet another overhaul.

Why can't we get the thing right? Some of the best economic, legislative and administrative experts in the country give the system their continuing attention.The trouble is they face a forbidding, next to hopeless job. Forecasting the short-run business cycle is hard. Predicting the 75 years for which Social Security forecasts are now made is complicated to the point of impossibility.

The surest set of facts on which to base a projection relate to the likely numbers of future workers paying taxes and retires drawing benefits. Statisticians know who's alive and in the country now and have a pretty good notion of current trends in births and deaths. But they don't really know whether current low birth rates will continue. Higher birth rates would ultimately produce more taxable workers, on the one hand, but also demand more child-related public services, such as education and welfare, on the other. It is known that more women are working -- which beefs up payroll taxes with relatively small increases in future benefit claims -- but unknown if this trend will continue and how it will, in turn, affect likely birth rates.

Life expectancies have increased markedly in recent years, a fact that contributes heavily to the projected long-run deficit. But again it's unknown whether the current projections of continuing gains are realistic or if it makes sense to assume also -- as the projections do -- that claims arising from premature disability will continue to grow too at a brisk pace. Whether longer life also means more years of active health is another important unknown. The acceptability of proposals to postpone the retirement age depends on whether people think they will be trading a few years of enjoyable retirement for a few years on a life-support system. When you talk about changing the terms of what amounts to a long-term contract between generations, you have to start guessing about future attitudes as well.

Moving into the realm of the economic, you are really off into the wild blue yonder. What will productivity and inflation look like 50 or 75 years hence? No one even has a good fix on next year.

With a huge population and a volatile economy, Americans can't know what the country will look like 50 or 75 years from now. And yet we do need to make our best guess about it so that we don't build in expectations for the future that we may not then be willing to meet. This periodic process of course correction shouldn't upset us. In a country as prosperous as ours, there is no real question of future default on our basic commitments to old-age protection -- a system that, when all its many parts are included, already accounts for about a third of the federal budget. What is really being debated is whether, given all the uncertainties involved, we would rather preserve some additional flexibility for future generations.