I've been writing columns long enough not to be terribly upset that I sometimes reach wrong conclusions on the basis of insufficient knowledge, false data, poor note-taking or congenital thickheadedness.

What I find embarrassing is to overlook some perfectly obvious point.

I'm embarrassed.

Just a week ago, I wrote a column in which I outlined a proposal by the former head of the Defense Intelligence Agency, Lt. Gen. Daniel O. Graham, for a technological solution to the nuclear impasse:

"In brief," I wrote, "he would have the United States 'put into space a nonnuclear system that could shoot down Soviet nuclear weapons over Soviet territory' -- an approach he says is feasible under technology already developed.

"Such a system would have a number of things going for it, among them that it would neutralize Soviet nuclear missiles directly, not by trying to convince the Russians that if they destroyed half the world we would respond by destroying the other half. And because the laser- based system envisioned by Graham's group (High Frontier) would attack Soviet missiles that were already launched (rather than attack Soviet territory in a counterattack), it would limit rather than spread nuclear war. Finally, since the High Frontier system would have an extremely limited offensive capability, its development would not be viewed as an arms-race-inducing threat by the Soviets."

And why am I now embarrassed? Not because a number of readers (many of whom know little more than I do about these matters) doubt that the technology exists, or that it could be done in five or six years as Graham believes. That's a problem of insufficient data. Nor is the source of my embarrassment merely that Graham's proposal doesn't really involve the use of lasers. That is the result of poor note-taking.

What embarrasses me is that I (and apparently a number of readers who responded to the column) overlooked an obvious, non-technical flaw. Jeremy Stone, smarter, more experienced in these matters, and demonstrably less thickheaded, didn't miss it.

According to Stone, director of the Federation of American Scientists, even assuming Dan Graham's scheme would work exactly as he envisions, it would be "the most destabilizing development imaginable." Why? If we could shoot down any Soviet-launched ICBMs, then the Soviets would have no defense against our own. The result would be an American technological advantage sufficient to frighten the Soviets to death -- or to rashness. Imagine our panic if the Soviets were to come up with a means of neutralizing our best weapons while preserving the effectiveness of their own. We might even be tempted to launch a preemptive strike to prevent the development of their terrifying scheme.

Would the Soviet reaction to our own technology be so different?

My blind spot is the same one that flaws most of the American debate on nuclear warfare: the assumption that the United States would never start anything, and that the Soviets know it. Thus we find it perfectly reasonable to view every new American weapons system as "defensive" -- and to expect the Soviets to share that view. That our MX missiles (now dubbed "Peacekeepers") would be installed in specially hardened silos and targeted on Russian missile silos for defensive purposes only may seem reasonable to us, but hardly to the Russians.

Stone makes one additional point: The Graham proposal would be a clear violation of the treaty against antiballistic missiles.

Stone is convinced that the way out of the madness we are in is not through "Star Wars" technology but through negotiations, as slow and frustrating as that may be.