Yes, the old backbone ignored authorship pretty much and required binomial names to be unique (it did cope with genus homonyms).
This is one of the major goals for the improved backbone.

http://dev.gbif.org/issues/browse/POR-2467
http://dev.gbif.org/issues/browse/POR-3022

We hope to advertise a first public preview of the new backbone at the end of this week.

Luckily this seems to be fixed in the upcoming preview. There are 3 names now with different authors, all of them being synonyms:
http://www.gbif-uat.org/species/search?q=Hedyotis+microphylla&dataset_key=d7dddbf4-2cf0-4f39-9b2a-bb099caae36c

With Hedyotis microphylla Merr. being a synonym of Praravinia microphylla (Merr.) Bremek.:
http://www.gbif-uat.org/species/8171958

... or in the new nub browser (details page waiting for an index rebult still)
http://mdoering.github.io/nub-browser/app/#/search/Hedyotis%20microphylla

"There are 3 names now with different authors, all of them being synonyms:" I think it's more complicated than that. For example, http://mdoering.github.io/nub-browser/app/#/taxon/2907578 lists these names as synonyms, but IPNI has:

Hedyotis microphylla Merr Philipp. J. Sci. 1(Suppl. 3): 239 1906
Lucinaea microphylla Merr. Mitt. Inst. Bot. Hamburg vii. 286 (1937).
Urophyllum microphyllum Merr. Philipp. J. Sci., C 12: 161 1917

These are three different names, and Praravinia microphylla (Merr.) Bremek. has the basionym Urophyllum microphyllum Merr. Philipp. J. Sci., C 12: 161 1917 (see http://biodiversitylibrary.org/page/8370141 ) NOT Lucinaea microphylla Merr. as shown in http://mdoering.github.io/nub-browser/app/#/taxon/2907578 - PS the browser is nice but doesn't work in Safari

I think we're hitting the limits of matching names, IPNI has most if not all of this as explicit links with IDs which should help avoid these situations.

That's the hopefully rare case I was waiting for when the same author described several species with the same epithet in the same family.

The trouble is that the old IPNI copy we have does not provide any basionym information: http://www.gbif-uat.org/species/105568293/verbatim
So in this case the basionym is derived by looking at the authorship. And because the authorship information particulary of zoological names is often missing brackets, our algorithm considers multiple names with original authors as potential recombinations.

If there are more cases like this we should probably not create new basionym relations programmatically in such problematic cases. We would then end up with all of those names being accepted as IPNI does not indicate a synonymy.

The only good solution would be to have IPNI or someone else providing us that information

See POR-2989 for ideas about improving a too eager programmatic basionym grouping

@Markus, the old IPNI DWCA does have some basionyms present. For example, urn:lsid:ipni.org:names:360445-1 appears once as a taxonID and once as a BasionymID.

http://ipni.org/urn:lsid:ipni.org:names:360468-1:1.3.2.1

There's an original name usage on the record: http://www.gbif.org/species/103257968/verbatim

However, I don't remember how complete IPNI's basionym links are. It could be that they don't have them in their database. (I really can't remember.)

The upcoming August 2016 backbone has this fixed:

• 7713093 Hedyotis microphylla DC. ACCEPTED within Hedyotis L.(Rubiaceae)
• 2892803 Hedyotis microphylla Kunth SYNONYM of Arcytophyllum filiforme (Ruiz & Pav.) Standl. (Rubiaceae)
• 7912822 Hedyotis microphylla Merr. DOUBTFUL within Hedyotis L. (Rubiaceae)
• 2899451 Lucinaea microphylla Merr. SYNONYM of Schradera polysperma (Jack) Puff, R.Buchner & Greimler
• 5338582 Urophyllum microphyllum Merr. SYNONYM of Praravinia microphylla (Merr.) Bremek.

Again we are missing any basionym due to an old IPNI copy (http://www.gbif.org/species/105579779) and the less eager basionym grouping. Should be adressed in POR-3160