In 2013, a paper titled “The ‘Wow! Signal’ of the terrestrial genetic code”appeared in the journal Icarus. A follow-up paper by the same authors, “Seti in Vivo: Testing the We-Are-Them Hypothesis,” was published in 2017 in the International Journal of Astrobiology. These papers argued that the standard genetic code showed evidence of having been designed by an extraterrestrial intelligence.
In the second paper, the authors explicitly distance themselves from intelligent design. They say that considering their hypothesis a form of intelligent design is a “common misconception.” They claim that they are not engaged in a form of intelligent design because they do not argue against natural evolution. To be fair, they presumably do not agree with the critiques that intelligent design theorists often make of Darwinian theory. But intelligent design is a broad tent and would include those who, while accepting the efficacy of Darwinian processes, think there is evidence of extraterrestrial intelligent design.
Furthermore, the authors use the same kind of technique that underlies the design inference. They point to features of the genetic code which are highly improbable and cannot be explained by any natural mechanism. However, as we will see, they would do better to pay more attention to work done by intelligent design advocates in understanding the basis for drawing design inferences.
Drawing Design Inferences
So what exactly is this improbable feature they have identified? The genetic code maps triples of nucleotides to amino acids. However, not all of the nucleotides are equal. Rather, in half of all codons, the first two nucleotides determine the amino acid regardless of the third nucleotide. These can be called the undivided boxes.
The amino acids all have a common core, but with a unique side chain. Well, almost: proline differs slightly and that will be important in a minute. For each one, we can count the number of nucleons, that is protons and neutrons, in the core and side chain. The standard core has 74 nucleons. Proline, the exception, has only 73. However, for what follows we will “pretend” it has 74, taking one of the nucleons from its side chain.
A Surprising Pattern
How many nucleons are in the side chains for all the amino acids encoded by the undivided boxes? 333. How many are in the cores? 592. How many altogether? 925. These numbers may seem banal, however:
- 333 = 37 * 42
- 592 = 37 * 52
- 925 = 37 * 62
The numbers thus display a surprising pattern. But at this point, those familiar with such numerological tricks will likely grow suspicious.
The authors try to bolster the evidence by pointing to other patterns. Take all of the codons with no repeated nucleotides. How many nucleons for amino acids for such codons that end in G or T? 307. How many nucleons for amino acids for such codons that end in A or C? Also 307. Now take all of the codons with two nucleotides the same while the third differs. How many nucleons in the amino acids where the repeated nucleotide is A or G? 999. How many of the same group where the repeated nucleotide is C or T? Also 999. The authors point to some additional patterns, but the flavor is the same. By applying certain numerical operations on nucleon counts of amino acids, we can find various surprising patterns.
Missing in Action
What the authors have missed is specification. They can show improbable patterns, but in order to draw the design inference that they want, they would need to show that the pattern is independent. The technique of description length as emphasized in the second edition of The Design Inference helps us to see this.
The fundamental problem with these methods is that they involve lengthy descriptions. Each of these patterns required several steps to expose. Those steps become the description length and have to be counted against the improbability. Each step along the way, including selecting only the undivided boxes, adding up nucleon counts, pretending that proline has a standard core, multiplying by 37, and taking the square, takes up description length. Once that has all been taken into account, no improbability remains.
Adding up additional patterns does not work either. Such an approach tries to make up on volume what it loses in each individual case. While each additional pattern makes the probability smaller, it also makes the description length longer. As a result, the case is usually no further ahead.
The idea of finding an intelligent message encoded in the standard genetic code is certainly a fascinating possibility. However, work done by intelligent design proponents in understanding the logic of design inference shows that this particular case misses the boat.