Peptides and nucleic acids are entirely distinct entities, either from a biologist's or a programmer's point of view. How can you store them in the same field without explicit indication of whether it's a peptide or nucleic acid? For polymers composed solely of standard residues it is easy to guess, but the sequences often also contain modified residues! Who knows if IMA is a modified amino acid or nucleotide without looking at the MODRES records? (see also below)
If MODRES came before SEQRES, the previous problem could be solved (although not elegantly). But the big problem is that MODRES comes after SEQRES! Using a modified residue in SEQRES before declaring its meaning in MODRES is just like reading an uninitialized variable.
If you want free text/free key-value pairs/free hierarchies, why not using a consistent syntax for all fields?