President Reagan's MX decision is being assailed by some as too much and by others as not enough. But the correctness of Reagan's decision rests on one simple question: does his plan reduce vulnerability?

Those who feel no new missile system is warranted need to realize that, given political realities, we are going to have some form of MX. The only issue is what kind of basing system is adopted.

Those who want to dig up Utah and Nevada need to recognize that the political realities rule it out. The issue, therefore, is whether the president's plan to put 100 MX missiles into hardened silos will reduce their vulnerability to Soviet attack over the next several years. And the key to that is whether our silos can be made so hard that the Soviets cannot count on being able to knock them out.

Hardening is nothing new. When we first put Minuteman missiles into silos in the 1960s, we only hardened them to withstand 300 pounds per square inch (psi) of blast overpressure. As Soviet missile accuracy improved, we recognized a need for harder silos and rebuilt them to withstand 2,000 psi, assuming that would be good enough for a decade -- which it was.

The mathematics of destruction show that the Soviets can take out 80 percent of our 2,000 psi silos if their missiles have an average accuracy of 780 feet. This uses the standard assumptions: that 90 percent of their missiles work, and that they aim two of their half-megaton warheads at each silo.

The administration now proposes to put MXs in those silos and harden them to 5,000 psi -- substantially more than either the 2,000 psi of current silos or the 600 psi to which the missiles in Utah and Nevada would have been limited by virtue of being above ground.

There is little doubt it is feasible to harden our silos to 5,000 psi; the Russians have already done so with theirs. The more important and more difficult question is whether the Soviets could improve their accuracy enough to overcome the increased hardness. To knock out 80 percent of a complex of 5,000-psi silos, the Soviets need to get their accuracy within 600 feet. This would represent a 30 percent improvement over today's accuracy. That doesn't sound like much, but we've reached the point where even the smallest improvements in inertial guidance are very hard to come by.

Look at the record. The Soviets started with a missile that missed by an average of 6,000 feet. They cut that to 3,000 feet after seven years. It took almost nine more years to halve the average miss again, to 1,500 feet. The next improvement, a 700-foot jump, took only two years, to 1975, but the Soviets accomplished it largely by buying machines from the West to make precision ball bearings. Since we don't have any better machines, they can't repeat this leap. In fact, since 1975 their accuracy has not shown significant improvement, although they have deployed more and more missiles with the 1975 accuracy -- hence the vulnerability problem we now face.

This certainly doesn't rule out future chances of accuracy improvement. If we harden our silos to 5,000 psi, we will be encouraging the Soviets to pour more money into improved accuracy. But even if they could quickly find a way to do it, they would face test and deployment programs that together take about seven years. To be sure, it will take us time to harden our silos, but it will take the Russians time to design a more accurate warhead, more time to develop it and still more to deploy it.

This discussion eassumes that all the accuracy numbers I've quoted really mean something. In fact, Soviet accuracy figures are themselves accurate only in the ideal conditions of a test environment. There is a real question whether we can ever expect operational accuracy within a few hundred feet for any inertial guidance system.

Falling warheads must pass through many layers of atmosphere, some with swift winds like the jetstream that can blow them off course. Test missiles are fired by specially trained crews, not the ordinary Russian conscript. The test accuracies are schieved by adjusting during repeated shots over thoroughly instrumented test ranges along an east-west axis. Without test firing across the North Pole, the trajectory missiles would fly in a war, no one can be sure the right adjustments have been made.

Finally, like us, the Soviets use multiple-warhead missiles. The missile does not release its warheads all at once. Instead, a "bus" flies along, dropping warheads one at a time. Every time the bus maneuvers and releases a warhead, a new element of error is introduced. The high accuracies we talk about represent the test accuracy of the first warhead on each missile, but for the first strike against silos, the Soviets would need to use the later and less accurate warheads.

No one can be absolutely sure about ICBM survivability in fixed silos, but what is known suggests hardening will buy us six to eight years -- maybe more. And that applies to test accuracies. Operational accuracies are very much less reliable.

At any rate, if we can't be sure, neither can the Russians. One more level of uncertainly will be added to a Russian first-strike scenario that has already been described as "chancier than drawing to an inside straight."