'Needs ID' pile, and identifications

yeah I’m trying to be…I guess I’d say fairly conservative in using it but there are absolutely a lot of photos you can’t ID to species from. Heck, I’ve got a few of them up myself, and don’t plan to remove them (useful for notes for me). But I don’t really consider it harassment to check the box that says it can’t be improved either.


It’s really interesting to read all the strategies that people take for identifying iNat observations and whittling away at the Needs ID pile. I do think there’s a lot that can be accomplished by people working to ID all the new stuff (or maybe old stuff) within their local area, or all the plants, moths, etc. within that area.

There’s also some value in people trawling through the unknown observations and moving them towards the right Kingdom or Phylum, although that’s an area where I’d like iNat to be automatically nudging new users to add their own IDs:

You recently added 14 observations that have no suggested IDs. If you add a high-level ID like “Flowering Plants” or “Birds” your observations have a much better chance of being seen by knowledgeable identifiers. Click here for a tutorial.

I’ve tried to use both the above approaches, but the one that seems to be most effective and also most rewarding for me is as follows.

  1. Start by learning the basic ID characters for some particular small group of related taxa, e.g. a genus with 2–5 species, or just part of a larger genus. Let’s call it a “clade”. Understand the range/distribution of each and how they’re distinguished. Get access to a few useful resources (keys, photographic guides, papers in which the species were described or revised).

  2. Use Identify to search for observations at the genus level for a small portion of the range. Work through the observations, say county by county. For each county, check on which taxa are known to occur there and in adjoining counties. With that info, you can fairly quickly confirm, refine or revise IDs for a bunch of existing observations because the range of possibilities is quite limited. You know what specific characters to look for to distinguish the 3 possible species in this county, for example. Generally, I include all data quality grades (Needs ID, RG and Casual) in my searches because I want to know that I’ve checked every observation of the taxon and make sure I’m aware if my IDs are diverging from those of other experienced people.

Pro tip: If I’m going to exclude reviewed observations in my search, then I’ll often start with the last page of results and work backward. That way, you know that when you’re done with page 13, you can bring up page 12 and see a full page of 30 new observations without skipping any. Working the other way, if you ID 10 of 30 observations on page 1, then when you move to page 2 you’re ignoring 10 unreviewed observations that have slipped past you onto the “new” page 1.

  1. Once I’ve built up knowledge and confidence with observations already ID’ed to genus level, I’ll search across all or part of the clade’s range working up one taxon level at a time. So I might set Identify to show me all observations in the parent family, but with the lowest rank set to “Family”. That way I can see which of these family-level observations I can confidently move down to genus or species level. If I think there’s a good chance that observations might have been IDed in a sister genus, I might search across the whole family with the lowest rank set to genus. That’s going to pull in a whole lot more irrelevant observations, but might find me a few more that need fixing.

Pro tip: Once I’m at the point where I’m working on finding occasional “known” taxa within a big bucket of stuff I mostly don’t know, I add the setting &per_page=80 to the Identify URL. I find it more efficient to visually scan through 15 pages of 80 observations than through 40 pages of 30 each.

  1. From there, I’ll keep my search set to the same location range, but back up the taxon rank one level at a time, adjusting the lower bound as I go. As I mostly work with monocot plants, it doesn’t take long before I get to the point that I’m trawling through all Liliopsida, all Angiospermae, all Tracheophyta and all Plantae within the distribution of my target taxa. Mostly, I try to resist the distraction of ID’ing stuff I encounter outside the target taxa, as the work required to check the right references etc. isn’t efficient if these are different references each time. But if you can quickly kick something into Cactaceae or Rodentia, why not?

  2. Next up is “Unknown” for the same geographic area. This is one I definitely try to process from oldest to newest, as a fair portion of “Unknown” and missing photo observations are still being tweaked by the observer.

  3. If I’m feeling really thorough, I’ll adjust the search URL from &taxon_id=6345789 to &ident_taxon_id=6345789&without_taxon_id=6345789. That means that instead of searching for observations that have a particular community ID already, I’m now searching for observations that have a certain identification but don’t have a matching community ID. For example I’ll get a list of everything that someone thinks is a cactus and someone else doesn’t. At this point, the knowledge I have gained should make it pretty easy to add relevant IDs for the real cacti and push a few of the lookalikes in more appropriate directions.

  4. Last stop will likely be to check out the “Similar Species” that iNat suggests on each taxon page within my self-assigned clade. If iNat reports that lots of people confuse these organisms with Agapanthus and Alliums then I can review those taxa using &per_page=80 to find any obvious mis-IDs.

At this point (or earlier if I lose patience), I’ve IDed the clade as fully as I can and I’ll make myself a note to come back and check for newly added observations in the future. This is the point where I’m ready to broaden my clade of interest, such as to a sister genus, and that takes me back to step 1.

I know my approach allows me to get from zero knowledge to competent identifier in a fairly reliable way, but is it something that would work for other potential identifiers, or is it just too specific to me?


I do it that way too!


I avoid this problem by not using the page buttons. Instead, I use the “View More” button, which basically reloads your search filters and gives you the new page 1. It’s right above the page buttons, next to “Mark All As Reviewed”. (Of course, you could also just reload the page the normal way, but the button is more fun.)


The View More button does work well when you’ve marked all observations on the page as reviewed.

But there are times (e.g. when sifting through higher taxa) when I have IDed only 3 observations on a page of 80, and I have not tried to ID most of them (“not the things I’m looking for”). In those cases, I don’t want to “Mark All as Reviewed”, because these observations might be ones I need to review for some later search. If I use “View More” in that situation I’m going to see 77 of the same observations plus 3 new ones.


Okay, that is a good point. In that case I don’t have any more suggestions.

1 Like

It’s just different techniques for different requirements and approaches I think. Sounds like “View More” works pretty well for the way you approach IDs, and that’s great.


I just mark all as reviewed and hit “refresh.”


I just mark all as reviewed and hit “refresh.”

That works if I’m trying to identify every observation on the page to the best of my ability. If there are some I plain cannot identify, then marking them “Reviewed” is the right response.

But if I’m trawling through a page of 80 observations IDed as Class Liliopsida in Central America and giving IDs to just the 3 Tigridieae observations it would be a mistake for me to mark the other 77 as “Reviewed”. A month later when I come looking to ID Sisyrinchieae, those 77 observations would be excluded from my search.

Basically, it’s most useful to me to have “Reviewed” mean something like “I determined an ID or decided I’m never going to be able to ID this thing.”


Yes! That’s the way I use it, too. I consider the “Reviewed” button to be the “I never want to see this again” button. Sometimes I will want to see again many of the observations I’ve seen.


Yes, I use it that way as well. Although I admit it does does make it more likely I will overlook an individual observation.


Hmmm… I used to try to ID unknowns or local wildflowers, but started feeling abashed about some mistakes I made (when I overreached my knowledge or got too attracted to AI suggestions). I don’t recall sensing anyone setting out to make me feel abashed, but some conscientious ID people would follow up and post very educational, very detailed info about IDing in that arena that were over my head. Plus, there was often a lot of repetitive copy/pasta to do to try explain to new users how iNat is meant to work*. So, I kinda pulled in my horns. I did not mean to stop IDing, but I took a break and the time just stretched on and on.

(*afaik, there’s no visible progress -plans or status reports- for the “improved on-boarding” that could, maybe, reduce the Unknown load. )


Come back. What I move from Unknown to Aves or Lepidoptera often gets IDed to species while I am ploughing thru the next thousand.


Absolutely, yes. Recruiting additional identifiers – primarily from within the existing community, with some additional external outreach for experts in certain problematic taxa – is by far the most efficient way to help with the Needs ID pile.

I have been running IDathons and identification workshops when I’m not sick. Unfortunately, I am frequently sick. Yay, chronic illness!
However, I would be thrilled to provide guidance and some practical assistance to any other person who would like to run such events. Even a one-day event can make a big difference!

I have given similar feedback to iNat staff and they seemed receptive… But they have a lot on their plate already!


This is why a mix of experts and amateurs is necessary. The amateurs can sort out much of the bulk, leaving a relatively organized pile of Needs ID, with preliminary guesses given, and the easiest taxa already verified. Then an expert can come in, ID the hard stuff, and then use the “cannot be improved” button as much as is needed.

You said it even better. Sorry for the repetition, I’m going through these posts chronologically :/



Onboarding is a high priority for the iNat team, but after basic server stability I think

Another thing I’m working on – making resources that are beginner-friendly easier to find!


Absolutely. It takes experience to build skill, and skill and/or resources to make experience effective.
Some of the experts here are coming from decades of being “the big fish in the pond” and are used to being the final word. They can be brusque or even plain rude. It’s not OK, and you should do your level best to disregard them and their bad attitude.
Fortunately there are far more patient and generous experts on iNat, many of whom have posted on this very thread. If you can buddy up with a mentor in your area of interest, your learning will progress much faster, you will have a safety net for mistakes, and you’ll have a designated person of whom to ask questions.

And for you experts… If you notice someone in your area of interest who seems to be truly interested in your taxa/region, why not take the initiative to apprentice them? It’s the best possible way to review your core skills, and in the long run it will ease your ID “burden”.


I would estimate that at least half of all new observations are of species very common to their area. Between local observations and Unknowns, it is very very difficult to run out of easy IDs. And every easy ID will give you more experience towards your ability to recognize what isn’t common. And it will also make you better understand the variations within taxa as well.

Do not despair! I regularly ID observations from weeks, months, and years in the past. All you need is for someone sufficiently interested to come by.
And think about museum collections – items can stay there for decades until a scientist uses them again, but that does not detract from their great value :)

Did you get any responses? I feel there are solutions, but the first step towards solving a problem is raising awareness. Do your fellow spider identifiers also recognize the need? And are they willing to try to remedy it?


See also this brilliant wiki on keeping identifying interesting.

There is a bottleneck between Order and Family for most taxa. The generalists have done what they can and the specialists are looking at genus or species. The more people can fill in this specific gap, the better! I know @marina_gorbunova fills this role for Pterygota, for instance.

It’s not impossible. Often it’s only a deadlock because the initial ID was made by someone who has since left. One more vote can pull the ID back to the correct category more often than you think, and then the specialists will jump in.

Let me join those who have encouraged you. Anything you can do is appreciated, and we understand that mistakes are part of learning. Everyone makes mistakes :)


A top-down approach like this is a good starting point. If you find yourself stuck on technical language or by a lack of gestalt, you can always try from another angle or freely skip up or down the ranks. Ironically, however, there are not that many newbie-friendly yet thorough guides for ID at a high level. It’s “too easy” for the experts so it can be neglected. However, in some cases, great beginners’ guides do exist. Usually this is the case when a taxon is popular – such as for birds or wildflowers.

Another good approach, as many species have specific limited ranges. It is actually more useful for more specific ranks and rarer species, however. Invasives in particular love to break the rules about where they are “supposed” to be. It’s their whole thing.

Through specialized and scholarly sources. Normal Google won’t bring up much. Ask identifiers in your area what sources they use ;)