Personally, I prefer numbering references and citing them in the text by something like
see A. Miller and P. Smith  for further details
see Miller–Smith  for further details
However, some publishers, and some coauthors, insist on using letters like
see [MS] for further details
which is short and maybe more intuitive than
see  for details
but quickly prone to becoming ambiguous.
The longer the article, the more likely it is that you need to cite different authors with the same initial, at which point the obvious rule of using just the first letter of the surname breaks down. At this point many authors just seem to add further letters from the surname quite randomly.
Is there a sensible set of rules for determining acronyms for publications?