Can nuclear war be ''won?'' Could either of the two superpowers confidently launch a strategic first strike, then conclude what would almost certainly be unprecedented violence on terms favorable to itself?
US nuclear strategy and force-modernization is based on the assumption that the Soviet Union does have the ability to destroy American land-based missiles in a first strike, or at least threaten to do so and thereby force Washington to capitulate. Moscow apparently fears the same of the United States.
But evidence gathered by nuclear weapons experts at the Massachusetts Institute of Technology shows that the outcome of a first strike by either side is far from certain, and that this uncertainty creates a strong deterrent. At the same time, the recently concluded two-year MIT effort warns that new generations of nuclear weapons just over the technological horizon could remove many of these uncertainties and should therefore be resisted as ''profoundly destabilizing.''
US officials insist that a nuclear war cannot be won. ''But we do have to recognize that we're up against a potential opponent who, from all the evidence, thinks it can be won,'' Defense Secretary Caspar W. Weinberger told the Monitor recently. ''And that evidence is: the weapon's they've acquired, the hardening of their targets (missile silos and command centers), the refiring capability, putting people in building shelters and a lot of money for civil defense, their doctrine expressed in the things that are available for publication, and their very belligerent statements.''
The President's Commission on Strategic Forces concluded that the so-called ''window of vulnerability'' used to justify new US strategic weapons was not as worrisome as some had claimed. The new MIT data further clouds this threatening window of time during which the Soviet Union over the next few years allegedly could attack US strategic forces with confidence of success.
The complex and detailed MIT report, titled ''Ballistic Missile Guidance and Technical Uncertainties of Countersilo Attacks,'' was written by Matthew Bunn and Kosta Tsipis. Dr. Tsipis heads MIT's Program in Science and Technology for International Security.
Major uncertainties about missile accuracy and reliability, warhead destructiveness, target hardness, gravitational and atmospheric variances, plus effects of ''fratricide'' (incoming warheads destroying or throwing each other off course) make the outcome of attack impossible to predict accurately , they assert.
''The technical uncertainties in any strike would be far greater than is commonly perceived in the United States,'' they write. ''In general, the results of an idealized attack are considered, while the uncertainties are ignored.''
Missile flight testing over relatively short distances (which is especially true for the Soviet Union) ''has a significant effect on almost every source of error in the system,'' report the MIT researchers, ''and in some cases, this effect is difficult to predict. . . . It would be essentially impossible for a Soviet planner to have reasonable confidence of being able to destroy significantly more than half the US land-based force.''
Since the Soviet Union is much more reliant on land-based missiles than the United States (75 percent to 25 percent), this puts Moscow in a particularly worrisome position. This is no reason for complacency, however.
''The field is wide open for drastic improvements in missile accuracy, which could have an enormous impact on the outcome of countersilo attacks, and the confidence with which such outcomes can be predicted,'' states the MIT report. ''In the absence of arms-control limitations . . . the development of such a dangerously unstable situation cannot be ruled out.''