Eric Anderson of evolutiondebate.info just sent me an interesting essay I’m sure UD readers will enjoy, so I reproduce it below. Eric is a regular commenter at UD, and he is a very insightful fellow who writes extraordinarily well. (Since I design computational algorithms as both a profession and a hobby I particularly enjoyed his essay Bits, Bytes and Biology: What Evolutionary Algorithms (Don’t) Teach Us About Biology, concerning the Avida program, and I highly recommend it to UD readers.)
Are We Alone? Identifying Intelligence with SETI
I just got back from a presentation this morning by Dr. Seth Shostak, Senior Astronomer at the SETI Institute, on the topic: “Are We Alone?”
By way of background, I had an email exchange with Dr. Shostak some time ago regarding Guillermo Gonzales, so I was somewhat guarded about what to expect from his presentation.(1) I was hoping to have the opportunity to perhaps ask a question or two from the floor, but in fact was able to do much more than that. Dr. Shostak not only took my main question from the floor, but was kind enough to spend several minutes with a few of us afterwards, taking additional questions and providing follow up.
Dr. Shostak is a master at public outreach. Although I would have preferred a bit more meat, his presentation was definitely at the right level for the general, well-educated public. He was thoughtful, provocative, and most importantly, funny. Not once, but numerous times, he had the audience laughing – either from a witty turn of phrase or from a parody about the public’s impressions of SETI’s work, such as some fun he had with Roswell (“Why bother looking in the far reaches of space for aliens? They’re already here!”). I am confident he has given the same presentation numerous times, but he nevertheless delivered it with a great deal of energy and a genuine sense of enthusiasm about the work he is involved in. His demeanor was not reserved for the prepared presentation either – with the small group afterwards he exuded the same charm and enthusiasm.
My question from the floor was essentially as follows: “How do you distinguish between an artifact of intelligence – a signal produced by an intelligent being – and a naturally-occurring phenomenon? And how confident can you be in your assessment?”
Dr. Shostak’s answer was essentially that you look around to see if you can identify other similar signals from various places in the galaxy, and if so, you have probably confirmed that the first signal was just a naturally-occurring phenomenon. This was certainly not the answer I was expecting nor even the direction, as I had anticipated he would respond by referring to information content, probability calculations, etc. Indeed, his response suggested essentially a negative filter: if it doesn’t appear to be a naturally-occurring phenomenon, then it might be an intelligent signal. I concluded that Dr. Shostak had either (i) misunderstood my question, or (ii) only partially answered, focusing on the negative side of the filter, rather than the positive identification characteristics used to detect a signal produced by an intelligent source.
After the presentation I fortunately had the chance to dig deeper. I indicated to Dr. Shostak that while he had provided an example of the kinds of things that might be done to determine whether a particular signal is naturally-occurring, I was really interested in the positive side of the equation by detecting design in the signal itself. Dr. Shostak responded by offering another specific situation: You see a radio wave coming from the location of a sun-like star; you know that type of star doesn’t produce that kind of signal, so that increases your confidence that it is probably coming from a planet around that star.
That again seems to be a purely negative filter. He did, however, add something that I think looks more like a positive indicator: namely, you check the periodicity of the signal to see if it exhibits some of the characteristics you would expect from a planetary origin, specifically, Doppler effects related to both the planet’s rotation and orbit.
I still wasn’t satisfied. Following a few more questions, I circled back to my central query: Isn’t there something in the signal itself that would allow you to determine that it is produced by an alien intelligence? The answer, in short, was no.
How Can We Resolve This?
Here is the reason for that surprising answer: in most cases(2) the resolution of signals across interstellar distances is not sufficient to identify the message itself, only the kind of signal. As Dr. Shostak explained it, SETI is doing the equivalent of a long-term exposure, and the signal is just too faint to tease out individual components. Thus, SETI can identify the location and type of signal, but not a message itself.(3)
Think of it this way (my example, so any failure of analogy is mine, not Dr. Shostak’s): In long exposure astrophotography, you leave the shutter open for a lengthy period of time, and in many cases you then further stack multiple exposures to obtain a useful image. The signal is so far away and so weak by the time it reaches Earth, that long-term exposure is needed just to obtain the faint signal. This works relatively well for astrophotography because the objects that are being photographed are typically stable over minutes, hours, days or even years. In other words, the same picture or “message” is being transmitted continuously – the equivalent of a galactic still life.
However, think what would happen if instead of a still life the interstellar “picture” you were photographing changed many times per second, like a high speed slide show or individual frames of a video. At the end of the long exposure you would have definite evidence of a signal – of type x, from location y, with average strength z – but the entire picture would be washed out and utterly useless from a standpoint of trying to determine what any individual frame of the show looked like. Similarly, with radio signals the message is encoded in small changes to the signal that vary many hundreds or thousands of times per second. As a result, if you can only “see” the signal by staring at it for an hour, you certainly will not be in a position to identify individual pieces of a message encoded in bits that vary thousands of times per second. Any message that may have been contained in the signal will be completely washed out from the long-term exposure required to see the signal itself.
If I am correct in my understanding of the above, this means that, barring some unusual circumstance, SETI is not able, and does not expect to be able, to read a message from an alien civilization.
Thus, it is not so much that Dr. Shostak misunderstood my original question about what kinds of artifacts need to be present in a signal to allow us to identify intelligence. Rather, my question rested on a misconception that SETI has the ability to receive a “message,” when in fact it only has the ability to identify the general parameters of a signal. SETI then uses what is largely a negative filter, albeit perhaps with some positive parameters in particular instances, to decide whether the signal is a naturally-occurring phenomenon. If it is not, then that increases the confidence that they have detected an intelligently-produced signal.
Meaning in the Message
There are a couple of implications flowing from the foregoing.
First, one oft-leveled charge against Dembski’s explanatory filter is that it is just a negative filter. I believe this assessment is not completely accurate, but even if the filter is negative, SETI gives us an example of how a negative filter can, if not provide a definitive conclusion, then at least guide tentative assessments of the artifact we are examining. (I suspect we could find other examples in archeology and forensics of purely negative filters being put to good use.) Thus, if someone takes the position that the work SETI is doing is scientific in nature – not necessarily the underlying assumptions about the likelihood of intelligent extraterrestrial life, mind you, but the actual signal detection and analysis work itself – then it seems one must also be willing to grant that use of a negative filter can be a valuable scientific tool. At the very least, use of a negative filter as one of the tools does not invalidate the entire enterprise.
Second, the relationship between the work SETI is doing and design detection in other areas is slightly different and more nuanced than what is typically perceived. Specifically, although Dembski refers to the movie Contact by analogy, and to SETI specifically, as examples of design detection, in fact SETI does not expect in most instances to be able to tease out a message. In biological terms, this situation might be likened to the initial discoveries of DNA’s existence and structure, coupled with the then-still-unproven hypothesis that DNA might contain some useful information. Thus, SETI is helpful in terms of setting a lower bound of confidence for what is needed to infer intelligent activity: a signal with the same general characteristics as would be produced by an intelligent agent, together with the absence of a natural explanation. This is a much lower threshold than Dembski’s probability bound, and it seems design detection, as applied to biology, can take a much stronger position in terms of its confidence in asserting design. SETI’s receipt of a signal across space, even without resolution of the underlying message, may be sufficient in some cases to detect intelligent activity. To pursue the analogy, the very existence of the DNA molecule, while an important signal across time, has now been resolved to the level of the underlying message. And with DNA and the other complex specified information found in biology, therein lies the much stronger positive case for an intelligent source.
I hope to follow up at some point with Dr. Shostak or other SETI personnel to make sure that my understanding of the SETI approach as briefly outlined above it is accurate, specifically in terms of the general inability to obtain the resolution necessary to resolve a message, and also in terms of the extent to which the inference to intelligent activity is based on a negative, as opposed to positive, filter.
This post is already too long, but at some later date I might perhaps also offer a few thoughts on the aspects of Dr. Shostak’s presentation that were less strong, such as the ideas about artificial intelligence and some of the underlying assumptions, including evolutionary assumptions, that get included in the Drake equation variables.
In any event, I hope the above provides a better understanding of SETI’s approach and its broader implications for detecting the artifacts of intelligence, both across the reaches of space and closer to home.
(1) As an aside, regarding other SETI impressions, I keep on my desk an article from late 2004 by Dr. Emma Bakes, Principal Investigator at the SETI Institute, that astoundingly states: “Additional experiments performed by Sagan and Khare showed that adding water to Titan haze analogs produced amino acids, so that one form of life based on these types of hydrocarbons could spawn another similar to terrestrial life. For the transformation of organic hydrocarbon chemistry to terrestrial life, ‘just add water.’” (Emphasis added).
(2) Barring, I suspect, a relatively proximate alien civilization that has produced a massively-boosted signal for the express purpose of interstellar communication, and which also happens to be pointing Earthward.
(3) Again, let me suggest that if there were an alien civilization bent on sending out a signal in order to be heard they could potentially overcome this limitation. Dr. Shostak’s practical point, however, is that if you are eavesdropping on an intelligent civilization that doesn’t have as its purpose to be heard, the kinds of signals it produces would likely not have the resolution required for transmission of a message across interstellar space.