Congress is on the way home. The U.S.-Soviet SALT deal is about to be struck. So you will soon hear a lot of assertions about what Congress's legislative handiwork or the arms-control agreement will do - self-confident, tempting-the-gods type assertions that the tax bill will certainly have this effect, and the cruise-missile protocol that.

Be leery. The record on all this is terrible. As evidence I will cite a couple of irrefutable assumptions that I and several thousand other political observers held throughout the past decade, which have lately been, if not refuted, at least roughed up beyond recognition. One has to do with welfare, the other with the U.S.-Soviet strategic relationship. But they are intimately connected in that they illustrate the same point; the risks of taking "paper" policy too seriously, and the crash that must follow when such policy has taken account of everything in the world that is relevant but human behavior as we have come to know it over the centuries.

I'll take welfare first and that familiar incantation that all critics utter as a kind of grace before settling down to the serious business of proposing a reform. It is that the present welfare system "discourages work and puts a premium on family breakup." This, of course, may be true. The problem is that extensive experiments with the kind of income plan many of us have favored to replace the current welfare system have turned up more of the same in our reform. The welfare recipients on the experimental plan substantially reduced their work effort - i.e., their taking and keeping of jobs. And - here was the genuine surprise - there was an astonishing increase in the amount of family breakup under a reform designed to have precisely the opposite effect.

One hardly needs to be told that these test results are the subject of hot controversy. Social scientists, welfare officials and politicians are far from agreed on what they mean; and as they try to decide, the decibel count is bound to get higher. But I don't think you have to judge the ultimate policy implications to spot the weakness in the thinking that has been confounded. It is that drawing-board, formula-crazed mentality that is the occupational hazard of political and social analysts everywhere.

To be fair about it, everyone always knew that giving people money was not exactly the best way to encourage them to take jobs, especially the kinds of jobs that are likely to be available to welfare recipients. But the documentation of a work drop-off under a more equitable and generous payment scheme was still startling. What accounts for it? Surely one answer is that real live people in these circumstances do not make closely calibrated judgments like those reached by the drafters of the program who were trying to provide work incentives along with the added money. Welfare recipients, that is, are no more likely than anyone else to fine tune their decisions to work (or not to work) to some relatively inconsequential benefit - 29 cents less for the second half hour, rather than 37 cents less, and so forth. If the work-generated income doesn't make an appreciable difference, they will say, as anyone would: the hell with it.

That the incidence of marital split-ups and one-parent familes will be increased when you don't make marital split-up or one-parent status a condition for getting welfare (as it is in many places now), is something else again. I will leave the probing of that mystery to others, noting only that it did get me to thinking about a particular piety that I have expressed countless times myself - and heard others express as often. It is that under the current system, vast numbers of men are forced to abandon their families in order to support them, since welfare payments will not be made if they are there.

On reflection, this now seems to me little more than another example of that combination of condescension and sentimentality with which the social analyst views the poor. For the explanation of why these men are abandoning their women and children has theoretical merit, but no human plausibility whatever. The odd case here and there can perhaps be explained this way, but surely it goes against everything we know about human behavior to suppose that men who truly care for their families are deserting them in vast numbers - vanishing from their radar screens altogether - so that the families may get their welfare checks. It doesn't figure, doesn't work.

Whatever job you feel moving, as we are about to, from the depressing world of welfare payments to the even more depressing world of nuclear arms, should be cushioned by the familiarity of the surroundings. For we are once again in the presence of pulverized assumptions about the way individuals and institutions can be expected to behave.

You will have read and heard over and over, in relation to the current SALT talks, that a large part of the problem is the vulnerability of each side's land-based intercontinental ballistic missiles to attack by the increasingly accurate missiles of the other side. I emphasize the word "accurate" because for about a decade it was not just desire or hope, but actual policy in this country that such threatening levels of accuracy should not be achieved. The Soviets, the thinking went, wouldn't be able to achieve them any time soon, and we would deliberately forgo the feat. We would stop where we were. Official doctrine asserted this, political observers commended its wisdom - and hollered up a storm when it looked as is some weapons system or military decision was contravening the policy. So the policy held - on paper. It was only the weapons that got more accurate, as distinct from the commentary about them.

There has been a fascinating series of articles in Science magazine describing the "technology creep" responsible for the change; improvements across a range of related fields that necessarily refined and bettered the quality of the affected weapons. To read it is to marvel that you ever thought things could be otherwise. People don't leave off these quests for scientific and technological answers when they are halfway there; there is a drive to completion that is, and always will be, at least as strong as any policy directive that seeks to thwart it.

I stress the point because just as surely as the upending of welfare-reform expectations will be attributed to the fundamental badness of the poor, so, too, this complicating improvement in the quality of missiles will no doubt be attributed to the wicked military on both sides. Yet devil theories will only perpetuate and cause us to repeat the kind of errors I've been talking about - and confessing to. Keep that in mind when they start telling you how everyone from consumers of energy in Ohio to Saudi sheiks and Soviet military planners will behave under a set of new rules. Before you ask whether it's good, ask whether it's even plausible.