The man who came to dinner came not in black tie, that badge of bourgeois stratification, but rather, as befits the leader of the vanguard of the proletariat, in the sort of plain worker's garments tailored for him on Saville Row. This man is a disarming fellow. He and his host have signed one agreement and are headed for a much more significant one.

Supporters and critics of the agreement on intermediate nuclear forces agree that it is less important for the hardware involved than for the political context in which it occurs and the political and strategic tendencies it expresses and intensifies. These are illuminated by the coming agreement to eliminate about 50 percent of the warheads on strategic systems.

Such reductions might serve the only important purpose of arms control -- enhanced stability of deterrence -- if the remaining warheads are on the right mixture of systems, and if subsequent presidents do not feel compelled to seek, as obligatory encores, even more destabilizing deep reductions. Furthermore, a 50 percent reduction might enhance the feasibility of strategic defense by limiting the intensity of the potential barrage to be defended against. And a first strike to be defended against after a 50 percent reduction might require strategic defenses very different and less baroque than the one using hundreds of satellites to destroy heavy SS-18s aimed at silo-based U.S. missiles.

However, there are two problems. Strategic defense may be dying on the vine. And the United States resists a decision that would alter the configuration of the U.S. deterrent, and hence Soviet targeting, and hence the suitable strategic defense. That decision is to deploy mobile missiles.

The Reagan administration promised to close ''the window of vulnerability,'' the vulnerability of U.S. land-based ICBMs to a disarming Soviet first strike. The administration may be opening that window wider and jamming it open permanently with the coming agreement.

Negotiations are focused on the question of ''sublimits'' within the limit of 50 percent of current totals. The crucial sublimit is said to concern the Soviets' heavy and highly accurate silo-busting missiles. But the emphasis on that sublimit is a residue of SALT II, when U.S. strategic planners assumed the modernization of the U.S. ICBM force with 200 MX missiles -- deployed in some mode that would present Soviet targeters with multiple aim points, by using some mobility and protective shelters.

Now, however, because of the mutual assured destruction of the MX programs at the hands of Congress and Reagan, we can count on only 50 MXs, and those sitting like fat ducks in old silos. Even kept in readiness on rails, they would be vulnerable to surprise attack. After 50 percent reductions of warheads, the ratio of Soviet warheads to U.S. land-based missiles probably will be worse. Even if the sublimits are fiddled so that the ratio becomes a bit better, it is inconceivable that it will become sufficiently better to make a few hundred U.S. silos secure.

We cannot count on the survivability of bombers, which could be destroyed on their runways by missiles launched on short, low-trajectory flights from submarines near U.S. coasts. Such missiles could also destroy a significant portion of the third leg of the strategic triad, those ballistic missile submarines in port at the time of attack. We would be left with nuclear-armed submarines at sea -- a few Tridents and whatever other submarines carry cruise missiles, unless the administration negotiates away these cruise missiles too.

The answer to the vulnerability of land-based ICBMs is mobility. But mobile missiles are hard to count. Counting is crucial to arms control, and arms control is, under Reagan as under Carter and Ford and Nixon, the ''centerpiece'' of U.S.-Soviet relations. Therefore mobility is considered a problem, not a solution.

Granted, the Soviet Union has three times the U.S. land mass in which to move and hide mobile missiles. And it has no domestic opposition of the sort (from some of Reagan's conservative friends out West) that blocked an MX basing mode involving mobility. (The Pentagon calls such opposition ''a public interface problem.'') But the United States is not Andorra; it is big enough to accommodate mobility. The question is: How serious is it about survival?

Does the president understand why the Soviets are suddenly saying SDI is a deferrable problem? They recognize the slow strangulation of a complex collaborative scientific enterprise. Strangulation is the result of three factors.

One is Reagan's acquiescence in a ''temporary'' (seven to 10 years -- the number hardly matters) undertaking not to exercise the right to withdraw from the Antiballistic Missile Treaty. The second is Congress' determination to enforce a narrow interpretation of what that treaty allows in the way of testing of defensive systems. The third is the Reagan deficit.

Nothing is more permanent than a ''temporary'' arms-control arrangement. Any deviation would be denounced as contrary to the ''spirit'' of the arms-control ''process.'' The narrow interpretation is, therefore, forever and will prevent the sort of tests necessary to prove strategic defense feasible. In the context of permanent fiscal constraint, Congress will inflict on strategic defense an anemia that will drive scientists to other projects that are not permanently in the subjunctive tense.

So after the INF agreement weakens the U.S.-NATO link and enhances Soviet conventional-force advantages, the next agreement may enlarge the vulnerability of U.S. strategic forces and the leverage of Soviet conventional forces. Gorbachev is a disarming fellow.