A Kind Reminder To Only Identify What You Are Sure Of!

Recently, I’ve been going through a lot of old identifications and research grade observations. When they are ided incorrectly and when I fix them I often get comments like:

“it’s not my fault I’m crap at spiders” etc.

I don’t want to point out any names to avoid and rudeness, but remember, if you made the ID fully knowing that the id was incorrect or that it could easily not be you identified it as, it… well… kind of is your fault. Don’t worry though, I’m not blaming anyone we all make mistakes and the best thing we can do is admit it, apologize if nessicary, and then move on and make up!

As a general rule something that can really help with your identifications is, even if it be for a split second, “do I know this is this?” if not just add a broad ID. Even if you are using the CV I suggest not adding a species ID to avoid it being incorrectly agreed upon and thus creating incorrect data. Always identify and agree upon observations that you can not because some other user you think is experienced did. That kind of independent identification is way, way more valuable than quickly clicking “Agree”, because it can be used to correct errors. It’s kind of like doing a long test, writing down the answer, then throwing out your previous work and redoing the whole thing from the beginning. If you arrive at different answers, then you know you must have made a mistake somewhere, and if you arrive at the same answer then you can be more confident that you did not make a mistake. If the person next to you just copied your answer you’d have two identical answers that cannot increase your confidence that the answer is correct.

Again, don’t worry! We all make mistakes just remember to be careful.

My profile page has more about this: https://www.inaturalist.org/people/3911628


This circles back to one of my recurring thoughts that for certain groups, recommending or even encouraging species-level identifications is the root of many false IDs. When I started iNat I felt that it was expected to be able to choose some species for every photo. I’m just starting to go back through my original observations and yikes… Fungi, lichens, and insects are difficult for non-experts to ID to genus, never mind species.


Several people I know in real life have smartly told me a paraphrase of Cunningham’s Law with regards to their philosophy of identifying their own observations on iNaturalist, usually using the CV suggestions. It annoys me because they’re actually right in that they are more likely to get a reaction, but they’re also more likely to get false Research Grade observations, and that’s not worth it to me.


There is such a thing as using one’s best judgment.

There is also such a thing as being upfront that one is in the learning process.

Finally, there is such a thing as explaining what resources you used and how you reached your conclusion.

All of which, I consider equally as valid as “being certain.”


It depends who, supports my tentative ID. For obs which have previously been ignored, or avoided, Cunningham can be useful.


I don’t feel too bad making an initial guess at an ID, guided by computer vision. But I try to be much more careful with my confirmatory IDs (and annoyed when someone instantly agrees with my ID without doing their own due dilligence).

But these standards are organic and continually shifting. It may be that the guidelines I’ve internalized don’t scale well, as iNat becomes more popular. It’s so exciting getting an observation to species, and I’ve repeatedly fooled myself going “oh maybe the new c.v. can identify my observation even though no person has been able to for years”. And sometimes it works! But usually what’s happening is that my observation didn’t capture the key features needed to make an ID, and I’ve wasted someone’s time correcting me.


I’ve seen this so happen so many times in regards to Narcissus pseudonarcissus. Every spring in the UK thousands of observations are uploaded and people constantly identify cultivars and hybrids as N. pseudonarcissus, sometimes these cultivar observations reach research grade status.

I usually tell them that it’s a cultivar when I ID it back to genus level and most of the time point them in the right direction of googling a species to see what it looks like before posting the ID. I’m not sure how much impact this has as I feel I’ve done it a lot, but I have noticed the odd few users learn from what I’ve taught them in future observations.

I myself have been corrected by other iNat users before. I once uploaded a red elfcup mushroom within the genus Sarcoscypha. I like many others in the UK thought it was simply Sarcoscypha austriaca, but it turns out that species is part of a species complex with Sarcoscypha coccinea. The two can only be identified to species level with a microscope. After learning this I continued to use this knowledge in future ID’s. I spread this knowledge to every new user I can when correcting their ID’s back to genus level. I feel like that must be making a difference.

I think the best thing we can do is reach out to those members in our community and correct their ID’s while also explaining it as clearly as possible. I reckon most should learn from the experience and may even go on to teach others within the community. If they seem particularly inept at ID though, you can suggest they ID to more basic things such as lizards, fungi, snakes, insects, spiders, etc. That way experts can find their observations and ID for them.


I agree that it’s not good practice to suggest an ID if you don’t have some reason to believe it’s correct. But at the same time, I always maintain that there’s not a single ID I’ve ever suggested for anything that I won’t withdraw if someone presents evidence to dispute it. No matter how confident one is about an ID, there’s always the possibility that there’s some other species we’ve never heard of that looks almost exactly like the one we suggested. It’s a Dunning-Kruger thing: if we’re truly ignorant of something, it’s impossible to realize how ignorant we are, because we lack the necessary knowledge to assess the depth of our own ignorance.

I’m sure we’ve all had the experience where we think we’ve identified something after hours of work, we Google it and see a thousand pictures just like ours, we see a thousand Observations on iNat in our area with the same ID, we confidently suggest the species… and someone more knowledgeable comes along 20 seconds later and says “you can’t actually rule out these other 20 cryptic species without microscopy or DNA work”, and back to genus it goes.

And there are plenty of species in all taxa where this ought to happen, but it just hasn’t yet because the experts have better things to do. Case in point, Clepsis peritana, only reliably distinguished from Clepsis penetralis by dissection of the female genitalia… currently 10,472 observations called “Clepsis peritana” on iNat, almost none of them with anything but a live photo of the moth, none of those species-level IDs warranted, but every day more of them get posted by users who think they’re sure of the ID… and the handful of Tortricid experts who know better don’t have time to go through 10,000 observations to comment “well actually…” on every one of them. And this is just one of many examples. The ID you or I most recently suggested could be in this category as well… how would we know, if we’re not up with the most cutting-edge taxonomy research for that particular taxon?

So I agree, don’t post an ID if you have reason to think it’s probably wrong, but if we were really all honest with ourselves about how confident we ought to be about our IDs, I don’t think much species-level IDing would ever get done. I’ve posted about 25,000 observations and 160,000 IDs at this point, and I wouldn’t be so bold as to say I’m “sure” about any of them. If someone told me tomorrow that American Robin was a species complex of 30 species about to be officially split based on DNA work and all my Turdus IDs were wrong given this past decade’s work that I was oblivious to, my surprise level would be exactly zero.


A very large portion of users seem to reflexively click “agree” on any ID they receive. This includes a surprising number of users who ought to know better (e.g. experienced users with at least a little scientific background, users who are active IDers themselves, etc.).

In most cases, it doesn’t seem to matter much who provided the ID and whether that person has any expertise at all in the taxon in question. It also makes no difference whether it is an ID that I have indicated I am uncertain about.

I suspect most forum users are more aware of the ID process than the average iNat user, so they are probably not the ones that this message needs to reach.

Unfortunately, I have had very little success encouraging users to not agree with IDs they cannot independently confirm. Occasionally I will get a withdrawn ID on a specific observation, but this never seems to lead to an overall change in that user’s agreeing practices.

If anyone has any recommendations about strategies that have proved effective, I would love to hear them.

(Before I get criticized yet again that maybe these users are researching their IDs: most of these users are clearly not doing so. It is generally pretty obvious whether any effort went into the ID. Please note that my comments are not directed at that small group of users who do in fact put thought into their agreeing IDs.)

There seems to be a large overlap between the reflexive agreers and those who uncritically use the CV.

This argument – that it’s fine to make wild guesses because they are more likely to get corrected – seriously annoys me because in fact I spend most of my time correcting wrong CV suggestions (at least when I am working on something like bees where the CV is generally wrong). The problem is that correcting IDs is more stressful and takes more energy than refining broader IDs, and that is time and energy I could otherwise be spending working on improving my skills and learning new taxa.

And in taxa where there are not enough IDers, a wrong ID may still sit around for months or years before someone knowledgeable looks at the observation. So there is also no guarantee that it will in fact even get corrected promptly.

(Again, please note: there is a difference between uncritically using CV suggestions and using them as a starting point to make a judicious best guess. As with “agrees”, it is usually quite obvious whether a user put any thought into their ID – even if they happen to be wrong, people tend to make different kinds of mistakes than the CV.)

What I tell these people is that I will spend as much effort on my ID as they appear to have done – in other words, if users who have enough experience on iNat to know better are clearly simply accepting the CV suggestion without even taking a few seconds to think about whether it makes sense, I will correct it to the best ID that I can provide off the top of my head, but I do not see any reason to spend extra time researching it.

Both of these practices may provide short-term gratification for the user (getting a specific ID or that coveted “RG” label for their observation), but long-term they hurt everyone – not just in the form of incorrect RG observations. Because careless IDs take up IDer time and energy, and IDers are likely to be reluctant to suggest tentative IDs if they know that likely as not the observer will simply agree with them.


I’m totally guilty of doing this, just to get that extra species to my list. I began stopping myself from doing it recently. I know this data gets used in research and my faulty data might contaminate it.


you could put that on the IdentiFriday thread. We need a better way to liaise between busy taxon specialists and identifiers who want to help.

Another reason why iNat needs a strategy to train up second tier identifiers to do the grunt work and free up taxon specialists. Apart from swimming in a sea of Unknowns for Africa, I work thru these 2 each day - doesn’t need a taxon specialist mostly, but does need ID help (mine are Africa but tweak location to suit you)

Kingdom Disagreement trapped at Life

Plant conflicts trapped at Plantae. There are probably animal conflicts too, but I ignore those - for all the green stuff we have just that single iconic taxon.

this whole discussion makes me think that there must be a better way to capture ID guidance from experts and make it more visible to the average iNat user. I’m partial to the idea of being able to fav comments & collecting the best and most useful ones under a new tab on the Taxon page. That would probably have to be curator-mediated though, and they are already overworked.


I’m dreaming of a wiki-based tab on each taxon page for ID tipps (features, lookalikes etc.)


in this case I would rather link experts and ask if the ID can be taken further - observations which have previously been ignored or avoided might be of rare and/ or difficult identifiable species groups, hence it might take longer until a possible wrong ID (due to the application of Cunningham’s law) is being scrutinized/ corrected.

I can see which are, languishing because they don’t show field marks - those … dicots.
And then the ones where taxon specialists leave a comment - can’t ID to sp without … (it is getting the ignored, but with useful photo info, to the right taxon specialist, which is the sticking point)
I much enjoy the obs where I say cautiously Orthoptera - then chemp takes it to first obs of a living specimen.

1 Like

Microscopy, yes; that’s morphology, just on a small scale. But DNA work? When a species can be defined and described (diagnosis, in the botanical sense) solely on the basis of a sequence, I will consider that as a valid argument. Until then, no.

Only ID what you’re sure of? Well, yes, as a general rule, but if we all did that, I fear the IDing process would grind slowly to a halt. There’s so little certainty out there. As @dianastuder says:

I confess that sometimes I do risk a certain level of unsureness to get a stubborn observation moving. BUT I always add a “perhaps” or a “could be” in the comments to indicate my doubts. I would love to see a way to add a tentative ID as a “cf” (other than a difficult-to-find observation field), but I’m guessing that would be difficult to implement.


As active identifiers we cover that uncertainty by responding promptly to notifications. For those obs my ID is both tentative, and temporary till the next ID rolls in for or against.


My first response to “A kind reminder to only identify what you are sure of!” was, “Well, duh!” but then . . . .

I give the name that I’m confident of, but sometimes not completely certain. And sometimes I’m wrong, whether I was “sure” or simply “confident.” Sometimes I think it’s probably that but I know I’m not sure, and then I add “maybe” or “tentative ID” or question marks to head off people who might unthinkingly agree. Sometimes, mainly in disagreements where more ID’s are needed to get to RG at the correct alternative, I’ll add an agreement with someone who I consider an expert. If I’ve consulted an expert about the observation, I’ll post that person’s ID (if he’s not on iNaturalist) or agree if he posts a ID.

And sometimes I do give an ID I’m not at all sure of (but might be true) in order to get the IDing moving. For example, recently I added the ID Bibio to an insect that looked a lot like observations I’ve posted and seen ID’d as Bibio, though it wasn’t identical to mine. I felt that was more helpful than labeling it “winged insect.”

So, I do consider IDing only what I’m sure of to be the right approach – but I won’t always do the right thing, and I don’t think that’s always wrong.