About the problem of author strings in the middle of the scientific names in autonyms: Perhaps the debate is argued too much from the providers side. My proposal to add a scientificNameWithAuthorship is based on consumer use cases.
verbatimScientificName is meant to give the consumer no guarantee whatever on what form the name has. This is good, because it guarantees that a maximum of records can be served from providers having no guarantees themselves. However, it does not allow the consuming applications or services to make decisions.
I believe the majority of a consumers either need a scientificNameWithoutAuthorship (output policy, name-matching policy) or a scientificNameWithAuthorship (unambiguous name representation policy, name-matching policy).
The latter use case cannot be served with the modification following Markus's and David's proposal. This means that every service intending to match or display unambiguous name strings with authors needs to do parsing. Furthermore, providers that know that they have canonical scientific names with authorship cannot transmit information about this fact.
I even doubt, whether the use cases for an isolated scientificNameAuthorship string are very frequent (although they certainly exist). I therefore propose to emend DwC with:
verbatimScientificName scientificNameWithoutAuthorship (which might continued to be called scientificName) scientificNameWithAuthorship
and drop the scientificNameAuthorship to reduce complexity.
Gregor