That’s really not the great falsification you think it is. My point was simply that the presence of all sorts of “casual” observations on iNat doesn’t pollute the GBIF dataset, which is a frequent source for occurrence data used in scientific studies.
I agree with you that there may be plenty of research value in observations that don’t meet iNat’s RG criteria. But someone choosing to use these “casual” datapoints can set their own thresholds for location accuracy and confidence. In the example you give of phenology of a cultivated plant species, knowledge and accuracy of location may well be immaterial, or, if you aim to show that pansies bloom earlier in Portugal than in Norway, very important. And if knowing a precise location is important, iNat makes it easy to choose only those observations.
I guess I should have made a more precise statement. How about this?
Researchers who are concerned about the quality and verifiability of iNaturalist observations as a data source have the option to select a subset of data that, for example, excludes observations with obscured, private or missing locations. Researchers can (and should) verify that their selection of iNaturalist observations is appropriate for the analysis they intend to perform. iNat applies several filters to select data that meets the “Research Grade” threshold for export to GBIF.
Similarly, specialists who choose to assist in providing identifications on iNaturalist are free to use any combination of filters to select observations to review. For taxa where precise location data are needed for accurate identification, identifiers are free to identify at a higher taxonomic level, to add a polite comment explaining the importance of location data, or to exclude obscured and private observations from their consideration entirely. No one is obliged to spend time reviewing obscured or private observations.
Try adding &geoprivacy=open&acc_below=1000 to your Identify URLs and see if that works for you.