Motivated by the RNA world hypothesis, Michael Yarus and colleagues have proposed a model for the origin of the ‘universal’ genetic code, in which RNA aptamers directly template amino acids for protein assembly. Yarus et al. claim that this “direct RNA templating” (DRT) model provides a stereochemical basis for the origin of the code, as shown by the higher-than-expected frequency of cognate coding triplets in aptamer amino acid-binding sites. However, the DRT model suffers from several defects. These include the selective use of data, incorrect null models, a weak signal even from positive results, an implausible geometry for the primordial RNA template (in relation to the universally-conserved structures of modern ribosomes), and unsupported assumptions about the pre-biotic availability of amino acids. Although Yarus et al. claim that the DRT model undermines an intelligent design explanation for the origin of the genetic code, the model’s many shortcomings in fact illustrate the insufficiency of undirected chemistry to construct the semantic system represented by the code we see today.