'Needs ID' pile, and identifications

My chosen pile
sits around 5K pages.
But - since I clear the new ones as they come in, and chew away at the older backlog
It is NOT the SAME 5K !
Which is okay 5K is a nice round target, and one day, I will dip below it.

It is good to choose a target that you can see move.
I added this one, and see stuff from October which seems to have slipped thru the cracks. Moving steadily down. 166 pages …

You can tweaks URLs and filters to get a possible target for yours.


I think it’s one of iNat points you can upload any photo, but there could be more guidelines, anyway problem is in ratio, if there would be more iders than observers, no blurry photo would stop them from clearing the pile.


I agree, what I’m suggesting is that iNat take notice of needs ID pile and misidentifications (e.g. from CV) and fine-tune things whenever the amount/ratio gets too bad. I’m not disagreeing that it already needs help, just explaining that they have ways to counter this in theory, and maybe don’t view it as being as hopeless currently than some do. They may have a different idea of the threshold of the pile or misIDs that would be needed before taking more intensive counteracting responses.

I think it’s the result of positioning iNat as learning tool, it’s a compromise website takes to get more people into it, I think it’s working as there’re many correct RGs too, they’re just not noticed that much as Needs ID. That’s why I think taking such actions seems as not very plausible, even though they could actually work, as they have a potential of scaring people off. Maybe if we still will be alive when new guidelines will see the light of sun, it will change id situation for the better.


True, I think more could be helped if the site put more emphasis on training new (or current) users how to observe, ID, etc. better. I understand there are already training/learning docs, but I imagine something like a “trial IDing” video game that people must complete a certain number of levels in before advancing to the main site. Incentivizing clear and cropped photos would also help a lot. Because the problem isn’t just an influx of many photos, but an influx containing many difficult or impossible to ID photos, which only worsen CV guessing problems.


I like that idea. A video game before you start. I see good intentions and enthusiasm … sliding into a wrong species, that takes many votes to overturn.


Or bumping something all the way back to family because you disagree with the exact species. I see that every day in Asteraceae, and often, it could have been bumped back only to subfamily, or tribe, or sometimes even genus, and been at a finer ID than family.

You don’t even necessarily have to know about those infrafamilial levels. If you start typing “Thistle,” among the two choices will be “Thistles and Allies” and “Thistles and Burdocks.” Either one of these will work if you disagree with the exact species but at least agree that it is a thistle. They will work even if you don’t know about Subfamily Carduoideae or Subtribe Carduinae.

Similarly, if you see that the flower is basically dandelion-looking, you can start typing “Dandelion,” and one of your choices will be “Chicories, Dandelions, and Allies.” This will work even if you have never heard of Subfamily Cichorioideae.

Yet in these cases and many others, the person disagreeing with the species will revert it all the way back to Family Asteraceae – thus making it that much harder to get to a finer level than family.


“Oh, it’s not THAT beetle, it’s another one, so it goes back to Insecta” :)


I see people bumping up all the way to Kingdom, but I don’t think most understand all the levels they are disagreeing with. In my opinion, this is important information and needs to be spelled out in the observation itself (or at least naming a few of them–something liike “Identifier disagrees that this is a Seep Monkeyflower, or a Monkeyflower, or a Dicot, or a Flowering Plant, or a Vascular Plant”).


The problem, I suppose, is that it’s possible to disagree with different levels in different ways. Say there is an observation of a common species. If I come across that as someone who knows the common species, but is unfamiliar with the alternatives - I may know that it is not the common species but still be completely clueless about what it is. I may bump it back to family. What I mean by that might not be that I disagree with the genus, but that I disagree with the species, and the most I am sure of is the family. I can’t identify at genus just because I don’t know enough to disagree with it - my ID has to be positive based on my knowledge. To me it almost feels like a bug that I am assumed to be actively disagreeing with the intervening levels - and I agree that most people also don’t understand that this is happening.


But all that bumping it back to family does is require more identifiers. Sometimes that takes years. A simpler way would be to leave a comment about what exactly you think about it and then if someone else agrees with the first ID check the DQA box that the ID can be improved.

1 Like

That’s true. The way I’ve started dealing with that is to try and go back if someone then adds another ID, withdraw my disagreeing ID and replace it with an identical non-disagreeing ID. But that takes effort, and for some people (depending on their ID habits) that will quite reasonably be too much.

If something is a very recent observation I might take your point, because there’s a greater likelihood of inconveniencing someone who might come along soon; but when something’s been languishing with a wrong ID for many months or even years it seems better to bump it back to a level where it is (a) correct, and (b) might be seen by someone using different filters.

The other problem with not bumping back is that you are left hoping that specialists in the correct taxon will happen to also look through the incorrect taxon and find it, which they may simply never do!


To me I find the following important to consider;

  1. New users should find guidance to not worry about ID’ing to species level. I know early on I hated being stuck at family/kingdom, so I’d let the AI guess a genus/species (sorry!). Novices often probably aren’t here for the ‘science’ so they don’t understand taxonomy, they just want to know exactly what they’ve found. We need to teach that family/kingdom level is fine, be patient, someone will help refine it down the line. Or not, try not to be too disappointed and → provide info to generate interest…

  2. Don’t view this site as a short term project. ID’s might come in 10-15 years! The importance of just being involved is to capture data. You never know if that half-blurry photo you got of a bird you ID to family turns out 10 years later to be an unexpected appearance that spawns a research project :)


Yeah, I recently got blasted by someone because I tried explaining implicit disagreements to them after they did this and they were adamant that I was talking about the wrong platform. Even when I linked to the iNat help page describing ancestor disagreements he blew up at me. Huge, long rants. So I walked away.

Anyway, once “hard disagreements” became a possibility, I wish there was more ongoing onboarding for users to help them understand the implications, more than just a forum post or beginning-user training.


But a lot of these observations are from users who are no longer active. Are there really that many people who are excluded by bumping back to family? It seems way better than leaving it alone, to me.

1 Like

I met people who were using iNat for many years and they didn’t know about “new” system, and it was up for two years, no rants and a cool person, but shows how you can somehow miss such big change, people should be forced visiting forum or they have only a partial experience.

1 Like

I think one issue in play with regard to number of “Needs ID” observations is the way the algorithm restricts observation movement out of coarser levels.

I added this in a separate post, as it’s a distinct conversation to my mind, but essentially I think the algorithm should be tweaked to require less community energy to overcome incorrect initial IDs / autosuggests.

1 Like

I’m sure the way it works currently will stay the same.

Comments won’t work on 90%+ of users, I agree that we need to take care of our ids and withdraw disagreements, maybe there should be an option to find those that don’t allow community taxon change, as we can find maverics.


oh sorry, i meant if you know it is wrong, bumping back to family seems way better than leaving the observation with no disagreeing ID. I wasn’t very clear there.