William C. Norris grew up on a Nebraska cattle and hog farm in the drought and depression 1930s. Twenty-four years ago he started Control Data Corp. from scratch; today that giant computer firm and its subsidiaries boast $3.8 billion in yearly sales. In the past decade Norris has plunged Control Data into an array of job-generating enterprises on turf most big corporation avoid like a plague -- troubled inner cities.

But now, with an audacity that equals his own past record, Norris is ready to "return" to the farm. But in this incarnation, Norris and Control Data will be proclaiming what many people consider a srangely antiquated gospel: that there is a role, indeed a strong and promising future, for small, family-run farms in America.

For a grass-roots glimpse of Norris' plans in action, one has to visit this rural town of Princeton, Minn. (population 2,500), where a prototype Control Data "agricultural business-center" -- a room full of computer terminals, keyboards and video screens -- is getting under way. And then to visit some of the farming familie around Princeton that Control Data is using to test the idea that computer-based technology can make start-up farming profitable, even in these days of sky-high land and equipment costs.

Flying -- as it does -- against every precept of modern-day agricultural economics, the Norris-Control Data scheme has to be considered a risky venture that won't conclusively prove its validity (or invalidity) for years.

But it has its true believers, and they make a case hard to dismiss out of hand. Conventional big-scale agriculture, says Norris, is far too dependent on capital-equipment and fossil fuels and highly vulnerable to inexorably rising costs of petroleum, pumped irrigation water and energy-based fertilizer. With massive farm machines it's necessary to plant in rows, where runoff from fertilizers and chemicals is greatest. Soil erosion and overreliance on pesticides and herbicides make big farming an environmental menaced. And the death of small farms threatens the life of small towns.

Where's the answer? In smaller units and more labor-intensive methods, says Norris -- "not a mule and 40 acres or back-breaking labor," but a return to more of a "closed loop" model in which each farm integrates its crop, livestock and other activities.

Hasn't history proved such small, self-sufficient farms uneconomic? Not so, says Bryan Roth, Control Data's agricultural services manager. Big farms only look more economic "Large farms more often succeed because they're well-managed and have sound information systems." And they're naturally favored by feed suppliers, equipment and implement dealers, agriculture extension services and university-based research. "We find small farmers just as capable of making decisions, but lacking the same information flow," says Roth.

Enter, then, Control Data, with a variety of computer-based services designed to balance the scales. They are all on display at Princeton. Either through computer information-question-and-answer courses or videotape, or both, the small farmer can educate himself one very topic from feeder pig production to raising honeybees, from nursery management to farm recordkeeping. I watched a videotape showing how to improve the lamb survival rate when ewes give birth -- cutting the umbilical cord properly, twisting a piece of straw in a newborn lamb's nose to make it sneeze and so start breathing.

Second, a farmer can use the Control Data computers for information as a kind of electronic library on the latest farm technology. Entering key words will bring up answers to questions such as, "What do I do when my lambs start shivering?" or "What bugs attack potatoes and how do I combat them?" The farmer can also utilize the computer as a kind of electronic mail, posing hard-to-answer questions that will be answered within a day or two by university agriculture departments or other experts linked to the system around the country or world.

Finally, some 10 computer-based management systems are designed to make it easy for a farmer to keep financial or production records to aid him on every front from lanting decisions to keeping tax records, from marketing his products to getting a bank loan.

Within a few years, Norris suggests, these services may be so easy to use that farmers will want computer terminals in their own farmhouses. Roth, a bit more pragmatically, suggest that Control Data can market is information-and-query services through "agricultural business centers" run by co-ops, banks, extension services or private entrepreneurs in farming communities across the country. Many farmers may find it easier to let these centers operate the computers for them, especially in recordkeeping.

But the objective is the same, and crystal clear: "We want," says Roth, "to provide the information base to let people start, manage and profit from small farm operations." Norris adds: "We're not talking about 'hobby' farms that only provide subsistence for the family. We mean a family income in the $25,000-to-$30,000 range. We mean intensive, diversified farming that contributes to the country's food chain."

To make all that happen, Control Data is also plotting possible ways to get small farmers into markets that often freee them out -- possibly by "networking" several hundred small producers together for coordinated selling. The networking by computer would presumably be more efficient than farm co-op methods. Such services, Roth maintains, would be economically priced -- though he acknowledges that the entire package of farm programs "just drips with Control Data income streams."

One needn't own Control Data stock to welcome that concept. Small-scale farming with computer back-up is an intriguing concept. Among other things it might bring a hint of fresh life to small rural towns that were left behind in a rush to agricultural bigness that didn't always prove to be better. And if the idea is to have a fair try, the best possible force to have behind it is a profit-motivated corporation.