I am currently across the country visiting family and have been doing some iNatting on the side when I can. I’m not very familiar with the local wildlife and as such have been largely relying on the CV to suggest initial IDs for my observations. Often times the CV will suggest a genus/broader taxon followed by a species in that genus, or two species from the same genus. In scenarios like that, I tend to select the genus/broader taxon as my opening ID for 2 reasons
Had I selected species A from let’s say a subfamily and someone IDs it as species B, then the observation is stuck at subfamily instead of non-RG species.
Whereas if I select the broader taxon it is stuck at non-RG species. My logic here is that someone may specialize in IDing let’s say a genus within this broader taxon, and this way they are able to see my observation instead of potentially missing it as it will be marked as subfamily had I initally ID’d to the incorrect species.
It takes away the ability to just click “agree” to the observation, forcing someone to take a closer look at my observation.
I’ve had a few instances of people who “ID” in my specialty taxon just spam-clicking “Agree” to observations and I want to minimize the risk of my observations falling victim to that.
My question for y’all is this:
Would you rather me initially ID to species and risk being wrong, or play it safe and ID to genus/family when suggested by the CV to?
I mainly ask this because I know how tedious it can get sometimes to have to type in an ID or use the compare tab to pick one, and just clicking “agree” is much easier. I’d imagine this is especially true for those who go through a lot more observations than me e.g. plant & bird specialists.
2nd Question:
Do y’all use the CV tab to quick select a species when IDing?
I personally just type in the species name or use the “compare” tab but I’d like to get a sense of if others do the same. I personally am always a little skeptical when I see someone used CV to ID my observation as none of the other fish IDers use it, but I could see it being more common in bigger more complex taxons.
I just learned about the existence of the forum and now realize about this issue. Before I didnt know there was a problem with many wrong ids. so I have sinned.
So far what I do, since I only dedicate myself to identify things in my backyard, is to use the robot a lot, and I do so because with majority of things, I dont get any single ID. like small moths, fungi, common birds, and stuffs like that… they get over looked.
I also have experienced people tagging my observations with wrong things… I believe is because they are having fun looking other people posts, and enjoy it, not necesarily know it. the people who does that in my experience, usually pick on very general observations like those who say “bird” “plant” and so goes on. they dont do it on more specific names observations.
So if some extranger is gonna wrong identify my observation I prefer to do it myself, sometimes that kind of accouds with thousand of IDs come and correct me, and I just correct my own post.
I can change my mind in the future, since I just realized about this issue, but until then… is easier to correct my own wrong id that wait for another person to correct theirs.
Regarding your first question, personally I identify to species if I think that’s what it is. I am a little more likely to identify boldly (i.e. when I am not 100% sure) with my own observations than with other people’s. I think it’s perfectly fine to do this so long as you at least attempt to pay attention so that you can go back and correct your mistakes.
Regarding your second question, I regularly use the CV to select IDs I could make without it. I think plenty of people do this, to the extent that it’s really meaningless if the observation says you used the CV.
My reasons are:
I cannot be bothered to sit up properly so that my keyboard and my mouse are equally accessible at all times. It is sometimes easier to just use the mouse.
I can’t always remember which common names iNaturalist uses, and I can identify more species than I can remember the names of off the top of my head. (“I know this, this is that one thing…”)
I like to double-check the maps and photo references for some species. While I’m on that tab, I might as well just just select the taxon from there.
I really don’t care if my ID has the little sparkles (for computer-assisted ID) or not.
Unless I am misreading what you said, I don’t think you have done anything wrong here? I think it is 100% ok to use the robot to identify your observations initally, as that is mainly what it is for.
Can I ask you to talk more about " also have experienced people tagging my observations with wrong things… I believe is because they are having fun looking other people posts, and enjoy it, not necesarily know it."?
Do you mean they are disagreeing with the ID you/the robot initially put in? Or are they putting obviously incorrect IDs on purpose (e.g. identifying a bird as a lizard)?
In my experience, I sometimes upload an observation, and the robot sometimes doesnt know what animal/plant/bug it is, so it gives like a recommendations that dont look like it?
So what I do is I search not based on zone, because many times the thing im reporting hasnt been reported before… but sometimes that even doesnt work?
So… if I leave for example an observation as “bird”, because I couldnt find a similar one on the robot list…
…sometimes, someone will come, and pick on from the robot list, when it clearly doesnt look like it… the robot is wrong, yet the person clicked on what the robot recommended…
why they do that? IDK… they are having fun enjoying INAT their own way? I dont see anything wrong, but I think is better to correct your own mistake when is later correctly identified, than to ask someone else to correct it.
I dont think they are trolling, neither incorrectly idetifying on purpose, they are just using the robot… and the people who do that, in my experience
usually do it with things broadly classified at first.
but if they are wrong, sometimes is hard to correct their mistake.
I’d say play it safe and don’t go to species if you’re not very confident. If you’re keeping up with your notifications, it isn’t a big deal to make some misidentifications, as you can always correct them when pointed out. The danger becomes if you’ve stopped using iNat 10 years from now and an observation you’ve misidentified now requires three people to correct it and “outvote” your misidentification. I go through a lot of old observations from 6-8 years ago, and many of the users aren’t active any more, making it very difficult to correct any mistakes that they made.
Yes, and in my experience, most people do. I know my North American moths pretty well, but if the CV gets the ID correct, I’m just going to click the name from the CV suggestions to save time. It doesn’t mean I don’t know the species and am “relying” on the CV, it’s just quicker than typing it it manually. In the insect groups I follow, I frequently see the CV icon on IDs suggested by the entomologist who literally named the species. I’d definitely caution against attributing anything one way or the other to an ID’s accuracy based on the presence of the CV icon.
Personally, I don’t tend to ID to species unless I’m pretty confident, largely because I’ve seen too many incorrect RG observations and would prefer to encourage two educated opinions rather than my ignorance and hopefully one educated (but possibly uneducated auto-agree) opinion. At the same time, if CV suggests a genus, I’ll tend to go for that rather than higher, because I watch my notifications and withdraw or correct any wrong IDs. I realise that it has the potential to be a problem if I did become inactive in future, but I think it’s still better anyway.
I use that tab, though I more often use the option of selecting from observations than CV. In general, though, I use whichever is more convenient in preference to typing names, given my tendency to misspell them. :-)
When adding IDs, I generally don’t feel annoyed by whatever choice the observer made when adding their ID. I assume everyone is doing their best and that we all have different strategies and resources for uploading. There are some edge cases that are annoying, but I don’t think this is what you are describing. If you are getting a large number of corrections to your initial IDs, consider reevaluating your approach. This is for your benefit as well.
If the iNat Next app is an option for you, it provides a confidence level for the AI/robot recommendations. You can also easily click on additional photos in the same observation to see if the recommendation stays consistent. When the confidence is high and consistent between different angles, I feel more comfortable adding an ID based on the recommendation for my own observations. Otherwise I add a broader ID.
iNat guidelines would be to use Bird or Fish - as I do - then follow the notifications from taxon specialists. Specialists need you to pick out their choice so they can work on it. A wrong sp ID, agreed with - now needs THREE taxon specialists. Including one who checks thru ALL the RG obs for the wrong ones.
As someone who IDs a taxon where the CV is wrong as often as not – please ID at a level that you are confident of. If you look at the CV suggestions and it isn’t obvious to you which one seems more likely to be correct, pick something more general. If the CV suggestions are very different (multiple genera/families) and you don’t know enough to figure out which one fits, pick a shared parent taxon. If you check the suggested species/genus and a large percentage of observations in your region are not RG, this probably means there are challenges with identification from photos and it would probably be wise to be somewhat conservative with your ID.
I would rather that people not guess without putting some thought into their ID, particularly when relying on the CV. It is more work to correct wrong IDs than it is to refine more specific ones: users often don’t withdraw IDs that are wrong (they don’t know they should or they miss the notification/don’t notice that the new ID conflicts with theirs); where there is a shortage of IDers, observations with wrong IDs might get missed and remain with an incorrect ID for months or years, and in the meantime are referred to by other users who don’t know that the ID is wrong, etc.
I will sometimes add a somewhat broader ID even if the CV suggestions seem plausible because it reflects my own current level of what I feel I understand enough to ID (i.e., I might leave the ID at “beetles” instead of a subfamily or genus because it is something where I don’t have a sense of how it fits into the more general beetle taxonomy; for a different specimen, I might be comfortable IDing it as a cerambycid or a staphylinid because I have a general sense of what beetles in these groups look like). This is my own personal ethic – when I add an ID I am saying “I think it is this” and I am not comfortable doing that if it is not an ID I feel like I could back up.
I would never judge anyone negatively for adding an honest broad ID that reflects their knowledge, even if the observation is of a taxon that might seem “obvious” to a more experienced user. In some cases I might point out that they can safely add something more specific (e.g. because there is only one species in the genus in the region), but this is meant as useful information, not criticism.
I do not find it annoying to type in an ID because someone added an initial broader ID. It takes all of a couple of seconds to type in the taxon name I want. This is only minimally slower than clicking Agree (assuming the initial CV suggestion was right in the first place – often it isn’t, so I need to type correct the name anyway) or waiting for the CV to provide its suggestions (which may or may not contain the taxon I need). In the Identify interface, the CV isn’t active on the main tab, so if one uses this workflow one has to switch tabs anyway and any efficiency savings are lost. I don’t think anyone should be typing “A” based on the thumbnails in the Identify interface without looking at the observation, because important things get missed this way (observer notes, additional photos, etc.).
You’ve hit on one of the many reasons I do all my iNaturalist submissions via the website, not the mobile app. When I’m in a new place I take several pictures of a new subject I want to ID. Later that day when I upload them to inaturalist.org, I can get a CV suggestion for each of the photos individually. If they all agree, I combine the photos into one observation and use that ID with some confidence. If they don’t agree, I use more resources to research them before submitting an ID.
I know this is not what you meant, but combining photos into one observation can be problematic for identifying when the observation contains multiple subjects. This is especially true when the observation contains multiple species, but even different individuals from the same species should each be their own observation. Observers have sometimes said something like, I didn’t want to make more work for the identifiers by combining separate individuals. I’ve never found identifying multiple individuals from the same species by the same observer to be onerous and if anything separating allows each individual to be annotated properly while combining often precludes that.
You’re right about what I meant. I take several pictures of one subject from different aspects to see if the CV will give a robust consistent ID for them. I’m not combining different individuals.
In my opinion you should identify the observation to what YOU believe it to be.
If you can tell it’s a frog, then identify it as a “frog” (Anura). If you can tell it’s a “pond frog” (genus Rana) identify it as that, and if you’re one of the very few who can tell a moor frog from a common frog, then identify it to species.
The CV can narrow search and help you remember names, and the ‘compare’ tab can help you verify that the species you think it is occur in the area - but IMO one should not assume that the CV is correct, and not take wild guesses. It should be YOUR identification, not the CVs.
the best way to get new IDs is
(1) provide a starting category - to the best of your judgement and knowledge. for example, “Order Lepidoptera” will help butterfly identifiers see your observation.
(2) be patient. museum specimens can often wait years to get identified - on iNaturalist, where identifiers are not paid, the wait is comparatively much shorter. we have gotten very used to expecting immediate gratification, but all science benefits from patience. better to get the right answer slowly than the wrong answer immediately.
if an observation has no IDs (not even your own), adding a general ID (like “birds” on a pigeon is not wrong. in reality, it is helpful. if someone later adds the “pigeon” ID, this will narrow down the observation’s categorization.
all in all, it’s great that you go back to review your previous IDs!
I just want to thank you for bringing up these questions! I have learned two very useful things from this conversation. I never knew that IDs using the CV were tagged. I suppose I noticed the tag with the sparkles but it never occurred to me what it meant. And secondly, I learned about the Chrome extension. Thanks so much for this great conversation!
I have spent a lot of time IDing in the past six months and would agree that you should pick the broader identification when you are not sure of species. I do this because it makes me uncomfortable when I select a species that I’m not 100% confident of and then the original observer comes along and immediately agrees with me. Now this is RG, and I’m not sure it should be.
I also use the CV tab to make IDs even when I know what the organism is just because it’s easier and also to double check my memory.
As an identifier, I don’t think I can expect everyone to know what the CV is good at and what it’s bad at, so use your best judgment about whether to go to species or not. I’d prefer it if more observers took a bit of time to compare existing observations with their own photographs before picking an ID, so a good rule of thumb might be to pick genus by default when you don’t have time to do more.
I almost never use the “Agree” button when I’m identifying, so that’s a non-factor.
I also don’t use CV to select an ID when I’m identifying. Maybe if I were trying to quickly ID observations which the CV is already good at (e.g. birds), but because I mostly ID things which the CV is pretty bad at, I’m in the habit of typing the species names myself.
In terms of things that I wish observers did, cropping the first photo is the my biggest want. ID’ing to the highest level possible is nice, but I’d rather a nice cropped photo showing the target organism, since even if you upload a 1,000,000x1,000,000 pixel photo, it’s only shown as a 2000x2000 pixel photo to identifiers, no matter how much you try to zoom in.
I often use the “compare” function to quickly pick the ID that I want, since it’s faster than typing, but at least with birds, anything that I commonly ID can be brought up with a 4 letter code, so it doesn’t really matter if the observer enters it at species level or genus/family/order. Birds get ID’d very fast, as long as the picture is good.
life i do not know i id boldly to species level as suggested by cv, unless i absolutely have no idea even with artificial intelligence. in the field, on my cell phone.
then i go home and look at them once more, googling a bit and sometimes i change that id, to genus level, even to family level, or i think it is something else entirely.
but if i left it on some general level, like insects, the chance that someone would go and id it for me is, in my experience, wastly lower than when i id it wrong on a lower level. so there is that.
