My understanding is that the taxonomic tree used in iNat is a determined thing, but as far as what observations are identified as, it is a concensus of what IDs are applied by all identifiers. We can all put different (valid within iNat) IDs and yet none of us could be putting what the system takes as the concensus view. I agree regarding “taxonomic concept”, something that took me a while to get my head around (coming from a novice perspective)