Here's the paradox of how Cornell (and most other universities I know) is coping with economic recession has interacted with our government's measures to deal with the same: Departments at Cornell were forced to cut their budgets 5 percent last year, and are likely to face another cut, albeit smaller, in the coming year. In meeting after meeting we agonize over the cuts, and by and large take them from the education part of our departmental budgets. We cut the number of teaching assistants, do not replace faculty or give up faculty positions.

At the same time, around $98 million of federal stimulus funds have already come to Cornell, largely in the form of new research grants or add-ons. We have at Cornell right now about $350 million in research funding from federal and state sources. It does not take a Ph.D. to realize something is weird here -- research funding is going significantly up in the next few years, while the resources for teaching will have fallen by 5 percent at least, and more so at other institutions.

A university is not just a research institute. There is a limit to how many teaching positions we can cut. The education of our undergraduates will suffer, by our own standards. It is only a matter of time before what we do will be perceived as simply selfish if we just grab the extra research money -- in the form the stimulus incentive has been taking -- and use it to, say, increase the number of our postdoctoral associates.

I believe Congress has to channel more stimulus money to undergraduate education. And something has to change in how our federal funding system treats the interaction of graduate education and research.

In the United States, science and engineering Ph.D.s rarely pay for any of their education (quite a difference from college, right?). Neither does their future employer. In chemistry, in a typical five-year Ph.D. program, students are initially supported as teaching assistants for two years, then as a research assistant by a professor's research grants for three years. Plus, the student typically gets a $25,000-$30,000 stipend, enough to live on.

How did we get here? I would say as a result of a strategy by U.S. industry to get its most highly trained employees, Ph.D.s, free of charge. Free, because industry contributions to universities are small (4 percent of research funds at Cornell). As for taxes paid by industry -- well, they are, let us say…minimized.

We give the young people who work with us the joy of taking part in exploring the universe, a work ethic, approaches to wisdom and sound professional training. But we must write proposal after proposal to find the means to support them. The university acquiesces enthusiastically, institutionally addicted to the flow of research funds and the reputation that accrues from our research.

And Congress does its part, moved by lobbying more than by a national plan. When there are too few American Ph.D.s for industrial demand, rather than have market forces work to increase their salaries, industry lobbies effectively to lower immigration barriers for the most highly trained.

But have we ever had a national referendum in which people have agreed that Ph.D.s in physics or chemistry should pay nothing for their education, while future physicians should pay, and some doctoral students in the humanities teach much of their Ph.D. careers? No question that we have in American graduate science education a remarkably successful social apparatus for innovation. Yet other countries (take Japan or Germany) support graduate study and research differently, and they have been successful -- look at patent production. I think we have just fallen into a comfortable system, steered by short-sighted industrial interests, our addiction to research and the universities' economic and prestige concerns.

I would propose that:

1. Government-granting agencies eliminate research assistant salaries from budgets. Instead, use the funds for a system of competitive fellowships for graduate students to use at the school of their choice. These fellowships should have a tuition component, which might be less than the tuition cost.

2. Universities charge science graduate students tuition and not pay them stipends, except for teaching services. That would generate extra income for universities (to be used to alleviate teaching cuts).

Such radical policy changes would need to be implemented in stages. The net outcome, if the student is not a government fellowship holder, would be that Ph.D. students may have to take out loans -- as law, medical and business school students do. There would be some contraction of our research dreams, for sure. And in the short term we would lose some of the incredible talent from Asia that has been attracted to our graduate programs. It will not be easy on us. But is cutting education easy?

I imagine that if the industrial lobbying pressure on immigration could be resisted, the changes would lead in the long run to an increase in pay for U.S. science Ph.D.s. Through that economic incentive, the wonderful increase in the number of women Ph.D.s we have witnessed might be complemented by an increase in the number of American men drawn to our profession; there have been fewer and fewer of them in the last decades.

I think we would then have a more rational system, and a way to deal with future budgetary constraints that's equitable to both research and education.

Roald Hoffmann is the Frank H.T. Rhodes Professor Emeritus of Humane Letters at Cornell and the 1981 Nobel Laureate in chemistry.

This article appeared in a modified version in The Chronicle of Higher Education (55:35). For a discussion of the proposal, with disagreements from Cornell colleagues, see Jeffrey Mervis, "Reshuffling Graduate Education," Science 325, 31 July 2009, p.528-530.