"research level" not always reliable

The “research level” flag shows agreement between ID from at least two independent users and as such is a simple and efficient way to give evidence of ID’s that are likely to be reliable.
However IMHO there is a frequent problem with it.
Let’s suppose that I post an unknown (to me) ladybird and I ID it to family, that is Coccinellidae, then someone else refines the ID to, let’s say, Harmonia axyridis.
I didn’t know that it was an Harmonia axyridis, so I should not add my own ID Harmonia axyridis after the ID of the other user.
That is not my behaviour, but I find this scheme frequently on iNat, leading to questionable “research level” ID’s, relying actually on one user only.
Is there some way to improve “research level” definition to overcome this issue ?


It’s been widely discussed, so you can read on that quite a lot. RG just means observation can be shown in GBIF and other platforms and there’s a community consensus, there’s no way to stop people from agreeing, but you can write to them, or discuss the change of the system of two ids or new level of status in topics about it. 2 id system is needed for rare taxa and can’t be changed to 3 or more.


Welcome to the Forum. What you do in these cases is very correct. This topic is one where many people have similar views (from previous topics). But, for now, the only solution is to keep doing the correct thing. Possibly, in future updates, better onboarding/training will help or maybe another strategy in the user interface will be developed.


The most frequent RG misidentifications I’m coming across are identified by computer vision and confirmed by a friend of the observer.

On the other hand, I’m identifying little known South American katydids (sometimes species never photographed alive before), and unless the observers accept the ID, they usually won’t reach RG.


Yes, you’re correct. The good thing about RG is that it can be rescinded by sufficent numbers of explicit disagreements, so errors can quite easily be corrected.


Welcome to the Forum @bssmnt52 !

For the previous (now closed) discussion of “Research Grade,” see this topic:


For cases like this, where there’s special value to getting an observation to research grade, I’d suggest climbing up through the taxonomic tree for the observation and skimming the “top observers” list until you find someone with ID skills who might be able to help in confirming the ID.


You might feel better knowing that iNat has examined a sample, using experts, and found that the community taxon identification was correct about 92% of the time:


That was back in 2017, so perhaps it would be worth repeating to see if that’s changed.


This often happens with iOS users who don’t use the website. I don’t have internet connection other than my phone and that really slows down the learning curve.


Hi @bssmnt52, using your example as a practical case, if you ID an observation to family (Coccinellidae) and an expert IDs to species (Harmonia axyridis), even if you had no idea that species existed but another iNat user suggests it, you do your research and are able to confirm the species ID on your own, why wouldn’t you agree with the ID? In that case you aren’t relying on a single person, and this is something that I personally do regularly.
In some difficult to ID species like Accipiters or some cases of Anoles with poor photographs, I usually don’t agree with suggested IDs even if they seem accurate because I do not have the knowledge nor do I know tips and tricks to separate similar species. There I agree that iNat users should not be “blindly agreeing,” but about my first point, do you see something wrong with that?


It’s not something to be terribly concerned about. Misidentifications are rampant in scientific museum collections also. Most (not all) scientists are aware that specimens in both museum collections and on iNat are potentially misidentified and deal with it accordingly. Obviously, we strive for absolute correctness, but it is a state that will never be reached regardless of strategy.


Absolutely. When I go to a herbarium I’ve never been to (museum collection of dried plants) for my research, it seems that about 5-15% of their tropical grasses are misidentified.


Indeed herbarium records can be wrong or outdated - lots of old records that haven’t been reexamined yet after taxonomic splits and merges, for example.


Yes, I think this would be interesting to run again! Additionally, it might be insightful if we also look at different subcategories as well. 92% reflects overall, but what about just for plants, which often are lot harder to ID? Would the number be lower?


Plant ID’s vary from species to species. Some are nearly 100% correct. Some actually are 100% correct. Some have a high rate of error.


All plants I identify are in the 100% category and you can take my word for it :grinning: No need to check


:grinning: I’m sure they are! (Wish I could say the same for mine!) However, I was referring to how those plants were before I ID’d them.


Honestly, iNaturalist is probably much more accurate than it should be for the millions of observations and users that contribute, but “RG” always requires a grain of salt. Same on bugguide, which is run by a mix of amateurs (like me) and experts.


Or even just taxa in areas where there aren’t a lot of iNat users.

An enormous amount of my observations never get confirming or refining identifications due to my area and the species.


When I started utilizing iNaturalist I did not know what I was looking at so uploaded with exceedingly general identifications. But over the last few years I have learned a tremendous amount so have begun to go through my older observations and refine my identifications to the level that I would upload them were I to observe them now.

I would resist the urge to seek a unilateral solution; what may be perceived as “an issue” may instead represent an observer returning later having furthered their education.