Virtually all SETI experiments probe the skies looking for broadcasts from afar: radio or light signals that would tell us that someone as sharp-witted as ourselves is out there. But could it be that while we use binoculars to scan the cosmic sea, bottled messages have washed up unnoticed at our feet?

As Robert Roy Britt reports elsewhere, the Aug. 25 cover story in the journal Nature suggests that the most efficient method of sending messages between the stars is not to broadcast them, but to use snail mail. Rutgers University researchers Christopher Rose and Gregory Wright reckon that clever aliens aren't going to prattle into a microphone. Instead, they'll inscribe their messages onto some hunk of matter (film, floppy disks, and flash cards are simple examples of information-bearing media from our own technology), pack it all into an interstellar rocket, and launch it towards their extraterrestrial pen-pals.

The two computer scientists claim that, compared to broadcasting radio to someone else's solar system, going postal could be enormously cheaper, requiring only a trillionth as much energy (or thereabouts) for the same message.

Does this mean that SETI experiments are misguided? Should we be using rakes instead of telescopes to search for messages from other worlds? Is it possible that an advanced civilization has littered solar systems like ours with packaged dispatches we have yet to find?

Answering these questions requires considering a few realistic scenarios for interstellar communication.

First off, there's the indisputable fact that physically transporting information can be quite efficient. Imagine packing a tanker ship with DVDs, and sailing it to Australia. You could jam approximately 10 billion disks into the tanker, which might take a week to cross the Pacific. That's an average 'data rate' of 600 trillion bits per second, and a cost per bit of roughly 0.02 trillionths of a cent! Those are impressive numbers that not only blow away your internet dial-up, they clobber broadcasting, too: sending the same amount of information with a TV transmitter would take two million years.

OK, bussing the bits might beat broadcasts in some circumstances. But what about interstellar communication? Consider an example derived from Rose and Wright's article: a message sent by missile mail to a recipient 100 light-years away. Suppose the delivery rocket travels at one-thousandth the velocity of light, faster than any of our own spacecraft, but hardly an unthinkable speed. Clearly, it will be 100 thousand years en route. Suppose that during all that time, your telecommunication colleagues have a powerful transmitter switched on, using an antenna comparable to the stadium-sized Arecibo dish to beam radio waves at the same recipient. Both schemes are assumed to send an equal number of bits, and both take the same amount of time (100 thousand years) to deliver them. But the cost per bit - in terms of energy - will be 100 billion trillion times less for the rocketed message, according to Rose and Wright.

This example certainly seems to suggest that snail mail beats hail mail by a large margin. But it's worth checking out a few important details of this argument. To begin with, the Rutgers researchers assume that the encoded message is very efficiently packaged, with a density of 2 million billion billion bits per kilogram. That's the information density of single-stranded RNA (like a polio virus), in case you wondered where the number came from. Of course, it might not be entirely obvious how to encode information at this enormous density (the aliens' problem) or decode it (our problem), but the point is that the total information on all the hard disks in the world, if packed this tightly, would weigh less than a single gram! Yes, that's right: you could send the contents of all the libraries on Earth in one envelope, if you could package it as efficiently as Rose and Wright have assumed.

Ergo, it should be obvious that the postal rocket will be far bulkier than any reasonably sized message it will carry. But this necessary packaging (the rocket) must be sent, too, and that takes energy. In addition, there's a real delivery problem. The nearest star systems of interest to extraterrestrial correspondents (for example, those known to have planets with biology) would probably be at least 100 light-years away.

As SETI astronomer Frank Drake points out, it's not easy - indeed, it's excruciatingly hard - to launch a spacecraft with adequate precision to make a soft landing, or go into orbit around, a planet that's that far off. During the 100 thousand years transit time, the planet's motions, and consequently, its position, will be slightly perturbed in complex ways by gravitational interactions within its solar system. The only hope of a precision arrival is to use a 'smart' rocket that can maneuver once it reaches the vicinity of the target. But maneuvering requires sensors, circuits, and fuel, and that adds to the weight of the rocket, further decreasing efficiency.

Such practical considerations boost the cost of message delivery. At the same time, there are plenty of ways to lower the cost of broadcasting. A larger antenna, for instance (composed of an array of widely separated dishes) could direct its radio energy at the recipient's inner solar system. At 100 light-years, that simple stratagem would reduce the energy cost by about a million at microwave frequencies, compared to the Arecibo-sized antenna. But the real trick in broadcasting bits into space, rather than blasting them, is to transmit light, rather than radio.

Radio is a great way to get another world's attention, but if you really want to send your cosmic correspondents your encyclopedia, you can do it faster on a light beam. It's technically feasible to semaphore 10 gigabits per second this way, which means that the texts of all the books in the Library of Congress could be rattled off in less than a day.

Sure, a light beam, even a well-aimed one, would require a good deal of power if loaded with information. You'd need ten trillion watts, even if you concentrated it on the recipient's inner solar system. On the other hand, you might be able to use solar energy as your power source, limiting the costs of the project to initial construction and maintenance.

But the real point is this: even with messages that are as large as all Earth's libraries combined, the delivery time from sender to receiver in our example is scarcely more than a century. Snail mail might be able to convey more info, yes, but would take a thousand centuries to do so.

So while Rose and Wright make an interesting point, it seems only reasonable to expect that a lot of interstellar messaging is going to be broadcast rather than delivered. Sometimes it's better to eschew the pony express, and saunter down to the telegraph office.