Editorials

The title I've given this piece is really false advertising. The actual answer to "how to be understood by intelligent aliens?" will remain unanswerable until we've actually been understood by a real-life alien and learned how that worked. And the answer would depend on the alien concerned. How we would communicate with intelligent jellyfish living under the ice of Europa would be more like how we currently communicate with dolphins than how we would communicate with aliens who we first saw as they lit up fusion rockets a light-month away.

So let's narrow the question to a more traditional SETI one. How can we design a message to be transmitted into space that will be intelligible to someone who knows as little about us as possible while still being able to receive the message? This question has been thought about for a very long time, but there are a couple of points about it that I'd like to make.

One basis of all forms of communication is having some common knowledge, which is then used to convey new information. We would by definition have some common knowledge with any alien watcher who picks up transmissions from Earth - be those at any frequency and either deliberate or leakage. Said watcher will need to know some astronomy, some physics, some engineering, and likely a fair bit of math. I don't see any way around that, unless someone can come up with a plausible science fiction scenario where a population evolves non-technological narrow-band high-sensitivity receivers - be those receivers radio or optical or something else. This common knowledge base need not be extensive. I can envision a watcher from a culture in what we might call the Iron Age, except with nearly-parabolic metallic cooking mirrors that lead to a strictly limited phenomenological understanding of radio astronomy. But there will at least be a starting point.

To convey new information, we need a standard way of representing ideas - a vocabulary. Many different groups have come up with ways to do this: the Arecibo Message, the various Cosmic Calls, Lincos, my own test message designs, among others. All of these follow a seemly-inevitable approach: build from what should be common knowledge, convey the relationships between ideas with a series of examples until the meaning of the symbols isunambiguous, and then progressively add more information. With enough data, a sufficiently capable watcher would be able to interpret almost any such scheme. The question then becomes what ways of encoding a given set of ideas are most efficient and easily understood.

This is where I disagree with the composers of the Arecibo Message and of Cosmic Call (and also with Michael Chorost's Guest Editorial last month). Those messages were based on two-dimensional images, monochrome, represented as zeros and ones. Certainly a watcher could interpret those messages, but it might not be so instinctive as it is for most humans. Perhaps the watcher is blind, and listening rather than watching. Perhaps they see in full three-dimensional volumetric rendering and don't immediately parse two-dimensional arrays. And while pictures are very good for representing certain types of information - such as human figures, chemical structures, and planetary surfaces - they're relatively low in information density. Simple diagrams often contain a lot of white space.

Message designs based on constructed languages, with some elements based on computer programming languages to allow for arbitrary data structures, can be much more flexible and efficient in terms of the number of bits required to convey many ideas. For example, here's a section from one of the test messages that I've variously subjected astronomers, engineers, a historian, and Caltech undergrads to as proxies for actual aliens:

00777700 10000000 37004001 10000000 00666600
00777700 10000001 37004001 10000001 00666600
00777700 10000002 37004001 10000002 00666600
00777700 10000003 37004001 10000003 00666600
00777700 10001237 37004001 10001237 00666600

This may not seem like much, but those 600 bits include a number of important concepts. Did you figure them out?

We have the basic structure of the message as being based around words (each 24 bits, here expressed as 8 characters in base 8), the words that separate out the beginning and end of a statement, the word for "=", and the convention for representing positive integers (although the last is a little less obvious).

When I've compared messages based on constructed languages to those based on images in blind tests with human watchers, I've found that both are readily interpreted but the image-based ones take about three times as many bits to convey equivalent information. That means we can either convey that more information with languages designed to be intelligible than with purely pictorial ones, or we can repeat whatever we want to say several times to guard against the watcher missing much of it.

But all of this is primarily my personal opinions. After all, we're trying to mind-read intelligent aliens who plausibly should exist, but of whom we know exactly nothing. Again, this will only truly be resolved when and if we actually find some.

Disclaimer: The opinions expressed in editorials are those of the individual authors, and do not necessarily reflect the position of The SETI League, Inc., its Trustees, officers, Advisory Board, members, donors, or commercial sponsors.