The only thing special about my refutation of Anfinsen in 1989, it turns out, is that it was a non-biological refutation, pssibly the first. A little statistics drawn from the meager dna and protein sequences available in 1989 (they fit on a few floppies) was enough to prove that codon choice, when there is more than one codon for an amino acid, influences the secondary structure of the protein at that loocation.
As to stop codons, if you had three halt instructions in an unknown assembly language, the decision that they are all the same and interchangeable might be assumed risky by programers, and there is biological evidence that this decision is wrong. (just google).
This would be important if the unknown assembly language was for some important program which we knew very little about and was of great consequent to our existence..
But since it's jst the genetis code, hell, lets just hack it and see what happens!