Ideas come from people, obviously. But under what conditions are groups, corporations, and even nations most likely to foster new ideas? It's not an easy question. Many of the essentials of a fertile, creative environment are anathema to an orderly, well-run organization. In fact, the concept of "managing research" is an oxymoron. Setting short-term goals, then quickly testing to see if they will bear fruit is similarly absurd. Jerome Wiesner, former president of MIT and science advisor to President Kennedy, was fond of saying, "That's like planting a seedling and, a short while later, yanking it out to see if the roots are healthy."

Ideas may come like thunderbolts, but it can take a long time to see them clearly - too long. And ideas are often born unexpectedly - from complexity, contradiction, and, more than anything else, perspective. Alan Kay, father of the personal computer (among other things), likes to say that perspective is worth 50 points of IQ (it may be worth more, Alan). Marvin Minsky, father of artificial intelligence, says that you don't know something until you know it in more than three ways. They're both quite right.

Incrementalism: creativity's enemy
As large, high-tech corporations around the world reengineer themselves by downsizing and rightsizing, the first casualty is basic research. And with good reason. The uncertainty, the risk-reward ratio, and the sheer expense come at too high a price for a cost-conscious society, which includes belt-tightening managers and nearsighted shareholders. Japanese corporations have long pooled their funds with government subsidies to achieve what they call "precompetitive research," which spreads the cost but doesn't help innovation. And, as senior executives in Japan are the first to admit, no self-respecting Japanese company sends its best people to these projects. Lackluster novelty only costs less in this scheme. Homogeneous, disciplined Japanese society has all the ingredients to refine concepts better that anyone else, but none of the juices to invent new ones.

Some computer science problems lend themselves perfectly to gradual improvement, or incrementalism - a step-by-step process of making something a little better each time. Very-large-scale integration is an example: scientists have succeeded in placing finer and finer lines on silicon. CPUs get consistently faster for the same price, in more or less the same-sized package. Incrementalism works in this case, but as a function of local refinements, not big new ideas.

On the other hand, being digital (sorry) is more global and cuts across most of life. Joel Birnbaum, the luminous head of research at Hewlett-Packard, calls future computing "pervasive": "something you do not notice until it is missing." Such research must look outward, because it's not just about the next-generation PC, it's about life. IBM and Intel, among others, have sometimes suffered from looking inward too much and growing only their own company's kind of people. None of these businesses would want Albert Einstein or Bertrand Russell in its labs - let alone running them - even though the presence of such minds would surely bring perspective and help dampen incrementalism. Companies just don't work that way.

Universities do
Research universities are a good example of a source of new ideas, but they are suffering from federal cutbacks and hence, looking for corporate support. Some faculty members and administrators complain that turning to industry for funding compromises their research, shackles researchers, and makes scholarship shortsighted - "prostitution" is a word I have heard mumbled. Boy, are they wrong. We are precisely at a time when universities can do exactly what corporations cannot do and the government should not do: foster and nurture new ideas. Let me qualify that. Government is not needed as a patron (the National Science Foundation could go away). But government may be a creative client, like a corporation, which in some ways is how the departments of energy, transportation, defense, and others work. Economic recession may be the best thing that has ever happened to university (as well as government) research, because companies have realized that they cannot afford to do basic research. What better place to outsource that research than to a qualified university and its mix of different people?

This is a wake-up call to companies that have ignored universities - sometimes in their own backyards - as assets. Don't just look for "well-managed" programs. Look for those populated with young people, preferably from different backgrounds, who love to spin off crazy ideas - of which only one or two out of a hundred may be winners. A university can afford such a ridiculous ratio of failure to success, since it has another, more important product: its graduates.

Maximize the differences
The best way to guarantee a steady stream of new ideas is to make sure that each person in your organization is as different as possible from the others. Under these conditions, and only these conditions, will people maintain varied perspectives and demonstrate their knowledge in different ways. There will be a lot of misunderstanding - which is frequently not misunderstanding at all, but the root of a new idea.

My advice to graduates is to do anything except what you are trained for. Take that training to a place where it is out of place and stimulate ideas, shake up establishments, and don't take no for an answer.

This poses an interesting challenge to any research organization: be even more nimble and supportive of the unconventional, tolerate more way-out and expensive ideas, and encourage the seemingly disheveled behavior of hacker life. In the pool of knowledge at a university, professors are not the fish, but the pond. The water is not chlorinated, clear, precisely circumscribed, and inhabited by one kind of perfect goldfish.

It is a muddied habitat with fuzzy edges and home to all sorts of people, including those who do not fit traditional scholarship. That is where new ideas come from.