I have seen comments from taxon specialists - can’t confirm without flowers or fruit - I need flowers or fruit.
Your way is correct.
1 Like
This is how I have been using the orange button.
If the evidence is definitely not sufficient to confirm the species (specific key features are not visible) and there are other species in the area that cannot be ruled out, the ID should not be at species level. The orange button pushes it back to genus. The green one does not.
There have been some previous discussions about the wording and whether it expresses what it is intended to, e.g. here: https://forum.inaturalist.org/t/doesnt-inat-commit-a-basic-logical-error-when-an-identifier-suggests-a-higher-order-taxon/44144
4 Likes
I don’t see the point of writing something and interpreting it differently. If there is a conflict between the wording and the intended interpretation, that should be sorted out. I think the current wording is clear and makes sense, and I don’t know why anyone should assume a different intent behind it.
Great, this is my interpretation, too. Now we all conflict with the Etiquette. :)
This is what I thought we were supposed to do too. There are some fungi that can’t be differentiated without viewing spores under the microscope. If they just pick the species the CV suggests that looks like it, they could be right or they could be wrong. I have been pushing it back to genus and marking as good as it can be.
3 Likes
Thank you so much for this! I’ve been struggling with identification etiquette for awhile since I began and for the last week I’ve been having people message me about my poor etiquette and this from really helps me out
I have also interpreted the orange button as @lappelbaum @spiphany @regnierda and others have suggested. If someone posts a photo that is inherently unidentifiable to species-level, not due to lack of identifier expertise but due to a lack of necessary features being observed, it should not have a CID at the species level. Even if there’s a chance the suggested species may be right, if there’s a reasonable chance that it’s not right and there’s no way to confirm/deny, that’s when I smash the orange button. To fail to do so would just leave tens of thousands of observations of difficult species complexes randomly assigned to one species or another with no rhyme or reason to which name is on which species. If species A and species B look the same in most photos, then whether a given photo gets posted as species A or species B is essentially random.
I’ve encountered users who were basically pulling their hair out trying to figure out why one set of 500 observations are labeled species A and 500 others are labeled species B when there’s no discernable difference, when trying to work out which species their observation is. The simple answer is often “none of those 1000 observations should be at species-level; there’s literally no visible difference between the two sets of photos”. There might in fact be 500 of each species posted, but which photo gets assigned which name in a cryptic species complex is a crap shoot.
While it does rub some people the wrong way to have their observations bumped “back to genus” when their ID might be right, I think this is ultimately the right thing to do. It’s a good learning experience to find out that what I’ve been calling one species is in fact a whole complex of cryptic species that I can begin looking into and paying closer attention to. I strongly oppose the idea that an ID should be left alone if you can’t definitively disprove it. If that were the case, I could post a blurry photo of a fly and call it any species in the world and say “please don’t disagree with me, because I might be correct for all you know”.
I like the current wording of the orange button a lot- it’s not asking “are you sure the finer ID is wrong”; it’s asking “are you sure the finer ID is unsupportable by the evidence provided”. The fewer features are visible on a photo, the more possible IDs could be correct, but that shouldn’t mean that unidentifiable observations get whatever species name the observer wants to give them.
Essentially, I’d say the burden of proof is on the person supplying the finer ID to defend it, and if they can’t provide evidence to do so, the most honest thing to do is back it up to a coarser option.
12 Likes
I’ve used one of those buttons just once. IIRR someone had identified a spider as Argiope keyserlingi (St. Andrew’s cross spider), but the stabilimentum looks like a roughly circular area about as big as the spider, whereas the St. Andrew’s cross spider is named for the stabilimentum that looks like a big X extending well beyond the legs, which are held like the crossing part of the X. I don’t remember the wording on the buttons, but it was confusing. I managed to disagree with the A. keyserlingi ID, which is what I wanted to do.
What is the etiquette for adding observation fields?
1 Like
The etiquette part of the article is great!
On the procedural side, I would add suggestion to be careful with Compare, the same way caution is suggested for CV. Some widespread taxa do not appear on Compare. I learned this the hard way.
Adding a suggestion for the identifier to look up relevant keys and other resources would also be helpful for novice identifiers.
When we talk about evidence, is this a murder trial or civil? A lot of variable taxa and genera can only be identified on the balance of evidence and very rarely with full confidence.
Bumping up an observation is fine but I think doing so without a comment is rude. I won’t do it.
Then again, my aim is to find underrepresented taxa in the field and within other people’s observations.
1 Like
Yesterday I found a picture of an orb web and identified it as Araneidae; I’m no spider expert, so I just put it in a family of orb weavers. Then a spider expert identified it as Araneidae and marked it “as good as can be”. Without a closeup picture of the spider, there’s no way to futher identify it.
Generally following along on #5. “Agreeing” is not the same as deferring to someone else’s opinion. To “agree” with an ID is to see the characters upon which the ID is based, either independently or with the guidance from the first identifier’s annotations, and not just to assume that the other person is an “expert” who knows what they are seeing and must be correct - I will call the latter “Monkey See, Monkey Do”. This is not intended to insult anybody, just to emphasize that identifying organisms requires some knowledge and intention regarding the subject matter, not just congenially clicking a button.
The active recognition of characters that support an ID before “agreeing” is especially important, because once an observation receives two uncontradicted IDs as the same species, it becomes “Research Grade”, and drops off the list of observations that “need review”. “Research Grade” observations from iNaturalist are aggregated by GBIF (the Global Biodiversity Information Facility) and become georeferenced data points for the distribution of the species to which the observation has been identified in that database. Those data points are assumed by many GBIF users to be authoritatively correct, and can have implications far beyond the iNaturalist community, including in decisions about regulation of invasive species.
A phenomenon contingent upon Monkey See, Monkey Do is the “Lemming Effect”: when several observations - perhaps including the reference photo in the icon for the species- of a given taxon become “Research Grade” for a given species ID, those observations can be used as points of reference for subsequent identifications - a sort of crowd madness or AI hallucination that is character based (this new image has the same characters as another one that is “Research Grade”), but premised on a false foundation of evidence. This results in potentially large numbers of observations being consistently misidentified, leading to apparent false distribution patterns on the maps in iNaturalist (and GBIF).
An example is the Cyllopodini, a tribe that includes several genera of day-flying yellow and black mimetic geometrid moths (along with co-mimics in Riodinidae, Crambidae, Dioptinae, and Arctiinae), that are quite difficult to tease apart. Some of the “species” of these as identified in iNaturalist are not what they have been identified as (quite a few identified to the wrong family of Lepidoptera), and there are multiply “corroborated” false IDs of species that occur on the eastern slopes of the Andes that have been identified as species from the Atlantic Coastal Forest of Brazil, leading to false patterns of geographical distribution. Once these patterns become established, it takes a concerted effort by more than a single identifier to correct them (it is “majority rule”, even to get back to “needs ID” status if there have been more than one “agree” to the incorrect ID). I have started to try, to address the cyllopodine mess using the collection at the Smithsonian and the original literature to try to correct errors, but it is an uphill battle.
I don’t know if these issues have been discussed in this forum elsewhere or not - there is a long tail on this string. Regardless, as of 2025, Monkey see, Monkey do and the Lemming Effect are still problems.
3 Likes
We also have the Geomodel Anomaly tool option.
https://www.inaturalist.org/blog/99727-using-the-geomodel-to-highlight-unusual-observations
Or - if you write a journal post - you may be able to inspire some competent help.
We Homo sapiens are social creatures.
If the “Agree” button was replaced with “I come to the same conclusion” there would be less blind agreements.
If the CV suggestions were called CV “guesses based on looks and location” there would be less reliance on them.
If Compare would read “Some more of our guesses, check your field guide” there would be less uncritical followers.
Identifications are classified as “Leading” as in leading to RG. Do we need every observation to be identified? Why?
When I post a different ID than what is already there, the interface says I disagree. Nope, I am just submitting my conclusions regardless.
When the wording of the apps intentionally or unintentionally pulls on the social conditioning strings, the results are as expected. No point complaining.
Want better observations? Provide better guides - by genus.
Want better identifications and more identifiers? Point to up to date resources. There will be incorrect IDs still but that’s fine. iNat is not a medical diagnostic app.
As I said before, the etiquette part is great. Keep this place civil and friendly, as is.
3 Likes
When I have concerns that an ID should have additional confirmation, I will sometimes vote “Yes” on the Data Quality Assessment question,
Based on the evidence, can the Community Taxon still be confirmed or improved?
This keeps the observation in “needs review” status until the vote is removed or counteracted. (Note: use this option judiciously and follow all notifications on such observations, since otherwise there is no automatic notification that further confirmation has been received.)
This is the fault of said users, and should not be a deterrent to iNaturalist users offering their best ID suggestions, even if they are not 100% certain or not “experts.”
The exact same problem applies to all data sets in GBIF. It’s like assuming that all specimens in a museum dataset in GBIF are correctly identified. No researcher worth the label should ever be making such an assumption. “Research Grade” in iNaturalist is the museum equivalent of saying “a specimen has been deposited and filed under this name.” There is still no guarantee that the name is correct, though it’s more likely than not based on iNaturalist’s own studies of RG ID error rates.
8 Likes
I think this has changed at some point in the recent past, as I now get notifications for those obs that I click ‘can the Community Taxon still be confirmed or improved? yes’
YMMV, depending on taxa and areas of choice – I wouldn’t say ‘exact’.
In my (restricted) experience, “Citizen Science sites” – there are 4 of them usable here, iNat being the major and comparatively “less wrong” one – are notably more prone to exporting to GBIF amounts of “collective hallucinations” (i.e. add occurrence data for vastly exotic or ecologically dubious taxa, never to be found once checked in the field) than dedicated herbaria and regional/national scientific organizations (… I first joined iNat in the hope of helping fix such citizen science sites: silly me!). This is why for our regional studies, the citizen science datasets are usually unselected prior to ingesting GBIF data - a preemptive measure to save valuable time.
edit: combined msgs
I disagree.
The green button is “I don’t know [based on what is provided]… but I’m certain it is genus/family/etc XYZ.
The orange button is “I know for a fact this is not the species you identified it as…”
If you are using the orange button when it could be that species but there isn’t enough information for “you” to be fully confident, then you are effectively becoming a gate keeper.
1 Like
Not at all. No gatekeeping involved.
If I know for a fact that the evidence is not sufficient to determine which species it is based on current knowledge about how to distinguish them, than it is not just my personal opinion that a species ID is incorrect or “my” lack of confidence, but an objective lack of the information needed for ID. Using the orange button is the only way to change the community ID to a higher level. It would be incorrect (overly precise) to arbitrarily suggest a species if there is no way to know what species it is. Therefore it is correct to say “I know for a fact that this species ID is wrong” (this is not the same as saying it is not that species; merely that IDing it as such is incorrect).
If the observer has additional information that they have not provided or if the state of knowledge changes (again, not my personal knowledge, but the knowledge of the community as a whole, including experts), such an ID can of course be revised.
7 Likes
That’s a somewhat different interpretation of the orange vs green disagreement button than I’ve understood. Not necessarily wrong, just different. If I push the green button, it means the previously proposed species could be right but there might not be enough evidence for me (or the community) to say it’s that. So I’m proposing a higher taxon, such as genus. The orange button means I’m saying it’s NOT that species (the evidence shows it’s not) but I’m not sure which of the other species it could be within the taxon I bumped it to.
4 Likes