LESS THAN TWO years after Congress sent the CETA public service job program to its grave, both the Republican Senate and the Democratic House are concocting plans for a new job creation program. The details differ, but the rationale is the same: there are millions of people unemployed and dependent on public benefits; there is much useful public work to be done; why not put the people together with the work?

The same line of reasoning has undergirded the many job programs that Congress has enacted in the past. The granddaddy of them all was, of course, the Depression-era WPA, which at its peak employed more than a third of all the unemployed and consumed 2 percent of the 1935 GNP. By the time the Depression ended, WPA's reputation had become tarnished -- partly by abuse of the program and partly by a general distaste for any reminder of the nation's hard times. But the program left a legacy of thousands of roads, bridges, stadiums, sewer and water systems, public buildings -- even LaGuardia Airport.

Since interest in direct job creation revived in the 1960s, no job program has approached either the relative size or success of WPA. Why? Too much is expected of them, for one thing -- job programs are supposed to reduce general unemployment, provide useful public services, bail out fiscally strapped states and localities, train the chronically unemployed and reduce welfare and unemployment benefit costs. But they are usually poorly thought out.

The most popular type with Congress is the public works variation now being considered. Its deficiencies were well documented during the various experiments with it in the 1970s. Because construction materials and equipment are very expensive and federally required wage scales high, such an enterprise creates relatively few jobs for the money. Because construction projects are slow to start, most of the money is spent after recovery comes. Because it's a wonderful form of congressional pork barrel, much of the money goes where it is needed least.

When these deficiencies become apparent, money tends to be redirected to "softer" kinds of public jobs where wages and overhead are lower. When these still turn out to employ many people who could have found jobs elsewhere, eligibility is tightened, training is stressed and wages are strictly limited. This is what happened to CETA in 1978. The limits worked -- in its last years, CETA employed only low-income people, and its average cost per job was very low -- but the program then lost its pork- barrel constitutency and it fell out of favor.

So here we go again. The White House, of course, may block any congressional job effort. What is most worrisome is that the very features that may tempt it to change its mind about a particular proposal -- appeal to powerful business, union and regional interests -- are precisely those that should not be part of a serious attempt to help regions and people adjust to structural changes in the economy.