What’s left—the strongest candidate signals after three surveys of the sky—are 11 “events.” They satisfy all but one of our criteria for a genuine alien signal. But the one failed criterion is supremely important: Verifiability. We’ve never been able to find any of them again. We look back at that part of the sky three minutes later and there’s nothing there. We look again the following day: nothing. Examine it a year later, or seven years later, and still there’s nothing.

It seems unlikely that every signal we get from alien civilizations would turn itself off a couple of minutes after we begin listening, and never repeat. (How would they know we’re paying attention?) But, just possibly, this is the effect of twinkling. Stars twinkle because parcels of turbulent air are moving across the line of sight between the star and us. Sometimes these air parcels act as a lens and cause the light rays from a given star to converge a little, making it momentarily brighter. Similarly, astronomical radio sources may also twinkle—owing to clouds of electrically charged (or “ionized”) gas in the great near-vacuum between the stars. We observe this routinely with pulsars.

Imagine a radio signal that’s a little below the strength that we could otherwise detect on Earth. Occasionally the signal will by chance be temporarily focused, amplified, and brought within the detectability range of our radio telescopes. The interesting thing is that the lifetimes of such brightening, predicted from the physics of the interstellar gas, are a few minutes—and the chance of reacquiring the signal is small. We should really be pointing steadily at these coordinates in the sky, watching them for months.

Despite the fact that none of these signals repeats, there’s an additional fact about them that, every time I think about it, sends a chill down my spine: 8 of the 11 best candidate signals lie in or near the plane of the Milky Way Galaxy. The five strongest are in the constellations Cassiopeia, Monoceros, Hydra, and two in Sagittarius—in the approximate direction of the center of the Galaxy. The Milky Way is a flat, wheel-like collection of gas and dust and stars. Its flatness is why we see it as a band of diffuse light across the night sky. That’s where almost all the stars in our galaxy are. If our candidate signals really were radio interference from Earth or some undetected glitch in the detection electronics, we shouldn’t see them preferentially when we’re pointing at the Milky Way.

But maybe we had an especially unlucky and misleading run of statistics. The probability that this correlation with the galactic plane is due merely to chance is less than half a percent. Imagine a wall-size map of the sky, ranging from the North Star at the top to the fainter stars toward which the Earth’s south pole points at the bottom. Snaking across this wall map are the irregular boundaries of the Milky Way. Now suppose that you were blindfolded and asked to throw five darts at random at the map (with much of the southern sky, inaccessible from Massachusetts, declared off limits). You’d have to throw the set of five darts more than 200 times before, by accident, you got them to fall as closely within the precincts of the Milky Way as the five strongest META signals did. Without repeatable signals, though, there’s no way we can conclude that we’ve actually found extraterrestrial intelligence.

Or maybe the events we’ve found are caused by some new kind of astrophysical phenomenon, something that nobody has thought of yet, by which not civilizations, but stars or gas clouds (or something) that do lie in the plane of the Milky Way emit strong signals in bafflingly narrow frequency bands.

Let’s permit ourselves, though, a moment ofextravagant speculation. Let’s imagine that all our surviving events are in fact due to radio beacons of other civilizations. Then we can estimate—from how little time we’ve spent watching each piece of sky—how many such transmitters there are in the entire Milky Way. The answer is something approaching a million. If randomly strewn through space, the nearest of them would be a few hundred light years away, too far for them to have picked up our own TV or radar signals yet. They would not know for another few centuries that a technical civilization has emerged on Earth. The Galaxy would be pulsing with life and intelligence, but—unless they’re busily exploring huge numbers of obscure star systems—wholly oblivious of what has been happening down here lately. A few centuries from now, after they do hear from us, things might get very interesting. Fortunately, we’d have many generations to prepare.

If, on the other hand, none of our candidate signals is an authentic alien radio beacon, then we’re forced to the conclusion that very few civilizations are broadcasting, maybe none, at least at our magic frequencies and strongly enough for us to hear:

Consider a civilization like our own, but which dedicated all its available power (about 10 trillion watts) to broadcasting a beacon signal at one of our magic frequencies and to all directions in space. The META results would then imply that there are no such civilizations out to 25 light-years—a volume that encompasses perhaps a dozen Sun-like stars. This is not a very stringent limit. If, in contrast, that civilization were broadcasting directly at our position in space, using an antenna no more advanced than the Arecibo Observatory, then if META has found nothing, it follows that there are no such civilizations anywhere in the Milky Way Galaxy—out of 400 billion stars, not one. But even assuming they would want to, how would they know to transmit in our direction?

Now consider, at the opposite technological extreme, a very advanced civilization omnidirectionally and extravagantly broadcasting at a power level 10 trillion times greater (1026 watts, the entire energy output of a star like the Sun). Then, if the META results are negative, we can conclude not only that there are no such civilizations in the Milky Way, but none out to 70 million light-years—none in M31, the nearest galaxy like our own, none in M33, or the Fornax system, or M81, or the Whirlpool Nebula, or Centaurus A, or the Virgo cluster of galaxies, or the nearest Seyfert galaxies; none among any of the hundred trillion stars in thousands of nearby galaxies. Stake through its heart or not, the geocentric conceit stirs again.

Of course, it might be a token not of intelligence but of stupidity to pour so much energy into interstellar (and intergalactic) communication. Perhaps they have good reasons not to hail all comers. Or perhaps they don’t care about civilizations as backward as we are. But still—not one civilization in a hundred trillion stars broadcasting with such power on such a frequency? If the META results are negative, we have set an instructive limit—but whether on the abundance of very advanced civilizations or their communications strategy we have no way of knowing. Even if META has found nothing, a broad middle range remains open—of abundant civilizations, more advanced than we and broadcasting omnidirectionally at magic frequencies. We would not have heard from them yet.

On October 12, 1992—auspiciously or otherwise the 500th anniversary of the “discovery” of America by Christopher Columbus—NASA turned on its new SETI program. At a radio telescope in the Mojave Desert, a search was initiated intended to cover the entire sky systematically—like META, making no guesses about which stars are more likely, but greatly expanding the frequency coverage. At the Arecibo Observatory, an even more sensitive NASA study began that concentrated on promising nearby star systems. When fully operational, the NASA searches would have been able to detect much fainter signals than META, and look for kinds of signals that META could not.

The META experience reveals a thicket of background static and radio interference. Quick reobservation and confirmation of the signal—specially at other, independent radio telescopes—is the key to being sure. Horowitz and I gave NASA scientists the coordinates of our fleeting and enigmatic events. Perhaps they would be able to confirm and clarify our results. The NASA program was also developing new technology, stimulating ideas, and exciting schoolchildren. In the eyes of many it was well worth the $10 million a year being spent on it. But almost exactly a year after authorizing it, Congress pulled the plug on NASA’s SETI program. It cost too much, they said. The post-Cold War U.S. defense budget is some 30,000 times larger.