My first two links are indeed about transcription! Rereading my comment in context now I'm realizing it was a bit unclear. I was responding to "the genetic code is rarely discussed as a potential source of error".

However, I believe my third link is about environmental stress triggering intentional mistranslation (mRNA to protein). From that paper's figure 3:

> Proteins arising from “statistical proteomes” have various folding and binding properties, resulting in phenotypic diversity in the host organism.

I haven't bothered to pull up the related references (5 obvious ones) to assess their strength though.

The papers you linked seem to be claiming that codon encoding preferences (which vary by gene category) are in fact due to (or merely correlated with?) GC content in mammalian genomes (as opposed to a number of other previously proposed mechanisms). This is surprising because individual tRNA abundance varies by cell state and type, so that would have been the obvious (but apparently wrong) explanation. It's doubly surprising because a number of single celled organisms utilize the mismatch between codon preference and tRNA availability in order to regulate protein translation, but these papers are claiming that's not a significant factor in mammals.