Strange bug as AFAIK none of the checklists that prove provide this name make this link to the dipteran fly family.

I looked also and came to the same conclusion.

My suspicion is this is a bad algorithmic assembly of the nub itself, due to Valonia vs Val[l]onia.

Valonia complicata Walker, 1856
Valonia planiscutellata Frey, 1964

My hunch being that those genera were created in the backbone, and then when the otherwise unseen Vallonia genus was added it was too aggressively fuzzy-matched and all those snails were brought in under the existing flys of Valiona.

Or something along those lines.

That seems likely; the more recent changes to the NUB matching allow double letters and single letters to be considered similar, but reduced the similarity of other spelling changes.

Good spot, currently both Valonia (one l, fly http://www.gbif.org/species/8707617 ) and Vallonia (two l's, snail http://www.gbif.org/species/2296006 ) are treated as children of the fly family Platystomatidae http://www.gbif.org/species/9504. Perhaps one for the https://github.com/gbif/backbone-patch to resolve...

Thats likely the cause, yes. And not easy to fix I bet. The reason the 2 genera exist is they differ in authorship.
Maybe we need to restrict the name normalization that reduces double chars to a single one to epithets only and leave the genus as is. I'll add a test and see how that can be adressed. The backbone patch unfortunately won't help here as its names also undergoes the name normalization

Backbone building fixed in https://github.com/gbif/checklistbank/commit/176a739b1e7823082d1d302833c7a7cdd4933fbc to just apply character dedoublication for bi/trinomials. Gonna be fixed live with the next nub build

See also https://github.com/gbif/checklistbank/issues/19