The solution presented in this anonymous, undated article is relevant if you’re constrained to only using LaTeX. But you’re not, and there are certainly easier ways to go about such things. We have Python and other higher level languages that we can use to pre-process source and extract metadata. This type of problem is made much easier if you write in Markdown, putting authors, affiliations, and other metadata in (say) YAML blocks, and use Pandoc to convert to LaTeX (or anything else). You can write a filter in Python, Lua, or Haskell to rearrange the metadata the way the author wants. This would be a far more pleasant experience, and easier to extend and generalize.