Should "As good as it can be" trigger Research Grade?

I feel “as good as it can be”, when marked by the observer, shouldn’t automatically trigger the Research Grade attribution and wonder what is the logic behind that. I have many insect observations that even though in theory they might be able to go down to species, genus level is as good as it can be for the photo, in that case, I generally mark it “as good as it can be” so I don’t spend more time on figuring it out, but I don’t feel that should become “Research Grade” if they can still be further id-ed. I feel this path to “Research Grade” should be separated from the id-ed to species path. Personally I’d like to see a “Highest possible Grade” instead of “Research Grade” for this type. Just checking with community and see if others have similar concerns or I missed something.


It looks like you’re using the as good as it can be mark as a sort of “note to self” if i understand correctly. I think most people aren’t using it that way. Perhaps you could instead use a different way to annotate your observations, like observation fields? Then you wouldn’t have to worry about RG at all.


I agree with the way “as good as it gets” triggers RG. RG basically just means that the observation gets exported to GBIF. If there’s community support for an ID and the observation would be useful to researchers and GBIF is willing to host it, I’m in favor of the observation data being made available by it going to RG.


I like the way “as good as it can be” sends observations to Research Grade. Assuming the identification is at a low level (genus or lower, I think) and at least two people have agreed to that identification. This is a way to (1) get observations out of the Needs ID queue when further ID’s won’t help and (2) preserve what useful information the observation does have and make it available to researchers, who may be working on the whole group and thus may want to see it.

I personally don’t mark an observation “as good as it can be” if I think other people might be able to improve the ID. I only do it when I’m confident that either the species is a taxonomic mess that can’t be solved with the photos, or its a 10-pixel bird in flight that no one will ever be able to identify. (Or equivalent.) If I personally don’t want to see the observation again but it might be identifiable by someone else, I mark it “Reviewed.”


I’m always skeptical that an observation marked “as good as it can be” is actually identified to a level that’s as good as it actually can be.

If it’s at genus level and robustly agreed upon I don’t mind if it gets RG and exported though.


Thanks for all the responses - very helpful!

I only ever mark a sighting “As good as it can be” when I know that the sighting involves something that is yet to be formally described. So it is usually something known to science and may have been placed in a family or genus but is yet to be given a species name. In which case, on current knowledge, the sighting is “As good as it can be”

Tribe too. And I think it shouldn’t.

For example: this observation has different species in the photos, Convolvulus arvensis and Calystegia sepium.

The recommendations from the help page are: « it’s best to identify to the lowest taxonomic level that fits all photos » and « mark the observation as “Based on the evidence, can the Community ID still be confirmed or improved? No, it’s as good as it can be” in the Data Quality Assessment section. »

So, we have an observation RG at tribe level, exported to GBIF but with two different species in the observation.


it’s any rank lower than family

1 Like

It should be any rank below genus (included).

Why is this a problem?

1 Like

It’s two different species, that could (and should) be split in two different observations. But the observer doesn’t care because it’s RG.


Interesting, and I quote because what you said is not what I expected, but generally in alignment with the opinion of others.

When I think “research grade” as a general term, I’d expect that an observation was identified by an expert, and as a result is a reliable taxonomic observation, with some reasonably high certainty that the ID is correct, and if used for research doesn’t require extensive confirmation.

That said, my research using BOLD has revealed numerous erroneous IDs, most made by researchers generally considered to be well versed. Same said use of Observations on iNat are rife with errors, with some filtered searches being 100% wrong, despite the fact that two or more people have agreed to the ID.

In exporting “as good as it gets” (according to “somebody’s” opinion) to GBIF, isn’t that a disservice to research?

Also- what do you mean by “GBIF is willing to host it”? Isn’t it automatically uploaded, without further review? Perhaps I misunderstand your point.


I think that if an observation goes to RG at a level above species, the probability that it is correct (at that level) is high. So it’s not a problem for researchers. Identifying the observation to species when ID should stop at a higher level would be a disservice to researchers.

One advantage of being able to send observations to RG at a higher than species level is that it reduces the incentive to ID to species (correctly or not) just to get to RG.


People I have talked to who research undescribed fungi and galls (in USA) have ask that these species be left at genus level and NOT marked “as good as it can be”. I don’t recall exactly why.

I mark things “as good as it can be” when the photo is too poor quality to ID any further or the species can only be IDed with characters not represented in the photo or notes. Sometimes this requires bringing it back up to genus level first.

1 Like

A frustrating observation. The observer has produced an interesting comparison of two species growing together. I wish the third photo, comparing the open flowers, were placed first and the whole observation duplicated as you suggest, with one copy identified as each species. However, that’s up to the observer.

Given how misleading either of the first two photos would be (though technically containing flowers of both, one withered), I’m OK with this somewhat misleading record going to RG as Convolvuleae and getting out of our way. But it could easily be made more useful. I think that’s the problem, not the fact that it went to RG at the level it did.


GBIF has basically agreed that observations with a community ID (as determined by iNat’s algorithm) at a certain level are good enough to host there. The records are uploaded by iNat, but really it’s that GBIF has agreed to host observations that meet certain criteria - iNat can’t force them to do this (Edit: I can’t find a source saying whether the taxonomic level cut-off for RG export to GBIF is a GBIF or iNat criterion - so it’s possible iNat could decide to export even more observations and GBIF might be amenable). There is no ID review on the GBIF side that I’m aware of, but GBIF could just decide iNat data isn’t something that they want to host (or they want a different subset or whatever).

I am certain that there are lots of erroneously IDed observations from iNat on GBIF, just as I’m sure that there are lots of data from specimens incorrectly IDed in natural history collections on GBIF.

Research has shown that iNat observations often have similar error rates to NHC specimens (though I suspect this has a lot of variation across different taxa).

As with any open dataset, it’s up to researchers to do their due diligence and use it appropriately. If researchers want to exclude iNat data because they don’t have confidence in it (which is certainly reasonable in some situations!), they’re free to do so.


IMO marking “as good as it can be” should used sparingly and only when there is a high degree of certainty that a lower ID is impossible.

At this point, I personally use it almost exclusively for definitive hybrids without iNat taxa. That’s mostly because the last time I created a flag for a hybrid taxon I was advised that creating new hybrid taxon was not desired and to ID to lowest possible taxon, add an observation field for hybrid name and mark “as good as it can be.”

Even in the taxa that I know exceptionally well I rarely mark “as good as it can be” based on the idea that none of the keys I use for identification are visible, on the off chance future changes with information/key might change that. I suppose that the truly unidentifiable images that are too distant/blurry to ID past some high taxon like Class probably warrant the “as good as it can be” as well but I’m unlikely to bother to make the extra clicks to do so.

1 Like

It’s the observer’s prerogative to do what they want with the observation. The status (RG at genus) is not factually incorrect.

Appreciate that the observer contributed an observation. Don’t get upset that they didn’t go as far as you would have preferred.