Hey everyone!
I’m trying to start a guide for the ID A Thon on the forum with everyone’s input! Please share ID tips, tricks, priorities, etc. to help both new and experienced identifiers learn new things for the ID-A-Thon
In my opinion, priority should be given to types of observations (in order from most to least important).
- Unknowns
- Disagreements/anomalies
- Needs ID observations at or above family level
- Needs ID observations between species complex and subfamily level
- Needs ID observations at species level
- Research Grade observations with 2 species-level identifications
- Research Grade observations with 3+ species-level identifications
- Casual observations
7 Likes
Not everyone is going to good at or have the inclination for the same times of IDing. While some people may find unknowns rewarding, others may find it more rewarding to focus on a particular species, or a particular broader taxon, or anything within a region, etc.
I don’t see any one status level as “more important” or “less important” to work on than others (except that in most cases it probably does not add as much value to add additional IDs to observations that are already RG if there are plenty of needs ID observations in that taxa).
The important thing should be that people are contributing in a way that reflects their knowledge and interests.
28 Likes
Interesting priorities! I used to think IDing Unknowns was really important, but now I’ve moved away from doing so many of those. Nowadays - and I reserve the right to change my priorities in the future - I mostly identify Needs ID observations at species level, particularly those I can move to Research Grade with just my ID. Doing that moves observations out of the Needs ID pile, moves them into export to GBIF (usually), and rewards the observer for being knowledgeable or using the CV effectively.
I never look at Research Grade observations, unless I’m called in to help with something. Ditto for Casual observations. On the other hand, I routinely mark non-wild organisms as captive/cultivated (thus moving them out of Needs ID) and I do a fair amount of marking genus-level observations “As good as it can be” if I think it can’t be IDed below genus and I can agree with at least one other identifier (thus moving them to Research Grade).
Occasionally, I look at disagreements or Needs ID observations above the genus level, but not all that often. I suppose there are plenty of Needs ID observations already at species or genus level in my region, for species I can ID, so I don’t need to poke around in other piles of observations.
So, my priorities would look like this:
- Needs ID observations at species level
- Needs ID observations at genus level
- Other kinds of Needs ID or Unknown observations
- Casual and Research Grade observations
Quite a different set of priorities! And yet each to their own - I wish you fun with Unknowns in the next month!
15 Likes
In general I engage in three types of identifying on iNat. For all of them the most important filters for me by far are Location and Rank (low end):
-
Rough placement: sifting through very high taxonomic levels Unknown-Class to spot anything I can bump a level or more. In this mode I’m not really too concerned with getting anything precise, just helping to outweigh bad IDs and nudge things along. Starting by filtering to your State or County equivalent can also help your likelihood of recognizing these observations a lot.
-
Fine-tuning: Trying to further refine to Family or better. In this I start lower, usually between Subclass-Superfamily and use my (limited) expertise of that taxon to move observations to family or better if possible. This is what I do the most of, and I find this tends to be the most helpful for building your own expertise because doing this type of identifying has a high likelihood an actual expert will then see that observation and move it to genus or better.
-
CV correction: After a while identifying (or observing), you start to get an idea of what the AI likes to over-recommend. I like to check these species or genera regularly and correct them because I know they get flooded with a lot of observations that don’t belong. This has led to me discovering a lot of look-alikes and that helps broaden my overall knowledge. If I notice an exceptionally large amount of these, I’ll get curious and start looking through RG observations of the over-recommended species/genus and often I’ll find mistaken RGs as well that should be corrected.
I think the most important of these is probably #3, because it helps to combat negative feedback loops, but they are all uniquely rewarding.
14 Likes
Robbie, I don’t mean to derail your thread about priorities, but what came into my mind was @norwichtim
Tim waded into the fire, and he survived, but he emerged in a different form. He was shaped by the journey.
One taxon can be a priority!
6 Likes
Anything that is helpful for identifiers for the ID a Thon should be in here!
7 Likes
Should I join? I already spend about half an hour (more if breakfast takes longer to cook) identifying unknowns and anything else I know better than it’s already identified, and occasionally search disagreements.
1 Like
I definitely would place the Unknowns as much higher priority that species level or genus level.
3 Likes
My priorities tip to where I get notifications - which reassure me that the ID was worth my time and effort. I do Unknowns which are local.
High priority for me (chosen locations Western Cape, then Africa) are the Newbies. Accounts created in the last week. That is our opportunity to hope for a response to our questions, and to share iNat’s netiquette and working as intended / expected.
Then the problem children. Kingdom Disagreement (trapped at Life). Broad IDs dumped in limbo where taxon specialists do not filter. Pre-Mavericks and Placeholders.
Still a residue from our Great Southern Bioblitz, and sigh CNC but that residue gets sifted down to difficult / impossible bran and chaff as the months and years roll by.
Identifiers who do Unknowns may never realise that Needs ID is almost as many obs ? I rely on taxon specialists’ filters to sweep thru what I direct their way.
If the Unknowns and broad IDs overwhelm you, and you have some taxon knowledge to apply, then the phylogenetic projects let you choose a slice to help
Try the various options and see which one works for you. I like a batch where I can give each obs its moment in the sun - but I don’t research various possibilities (much - I will consider A B or C). Fill the lake of data with one bucket of clean water at a time.
7 Likes
woaaah this is hella useful!
one small doubt- in all three projects there wasn’t a single mantodea related clade , im curious to know how or why :0 (was looking for it because I also ID mantises)
@jeanphilippeb no Mantodea ?
some Mantodea from among the Hexapoda @aditya_puthenpura ?
@earthknight, I’m curious why you think that, because I find it fascinating how different people use iNat differently. Since a lot of Unknowns are made by people who are new to iNat, do you think identifying their Unknowns keeps them interested in iNat long enough to figure it out? Does it bother you there are good data sitting there in Unknowns, unrecognized and unappreciated? Do you simply enjoy the exercise of figuring out for yourself what something is? Or is there some other reason you prioritize Unknowns?
3 Likes
@yayemaster also asked for tips for the ID-a-thon, so here’s my big one:
Filter, filter, filter!
No matter what your priorities are for identifying, exploiting iNat’s filtering abilities really helps make one’s identifying more efficient. If you open iNat from the Identify tab, like this: https://www.inaturalist.org/observations/identify you get every single Needs ID observation around the world, more than 108 million right now (yikes!). You could scroll through hundreds of those before you hit upon something you feel confident identifying.
So, I think the first step is to filter by location. I live in the small state of Massachusetts in the northeastern United States, so I usually start by filtering for the location “New England,” which is the six small states around me, like this: https://www.inaturalist.org/observations/identify?place_id=52339 I did that by typing New England into the Place box near the top of the Identify page.
Usually, I start my day by just scrolling through 5 or 10 pages (so, about 150 to 300 observations) with this New England filter, because I can identify at least a few of those observations.
Then I get serious about filtering. If you click on the Filter button near the top of the Identify tab, you see ways to filter by iconic taxa (birds, plants, fungi, etc.), by date observed, by whether the observations has sounds, and so on. So if you’re only interested in identifying mammals, you can click that iconic taxon button and you get all the mammal observations in New England that need an ID: https://www.inaturalist.org/observations/identify?iconic_taxa=Mammalia&place_id=52339
That’s still some 30,000 observations. I would filter that even further. If you go back to the pop-up window that shows up when you click the Filter button, down towards the bottom lefthand corner of the window, you will see a button labeled More Filters. Click that, and you get many more ways to filter. For example, you could filter New England mammal observations for observations that have been annotated for the presence of bone: https://www.inaturalist.org/observations/identify?iconic_taxa=Mammalia&place_id=52339&term_id=22&term_value_id=27 That’s only about 600 observations. If you’re someone who knows skulls (not me!), this sort of filtering will bring you right down to where you can be very efficient in reviewing and IDing observations. (And that’s why annotating observations can be really helpful.)
If you’ve never used these specialized filters before, it can be confusing to figure them out. I’m happy to help you with crafting a useful filter for your interests. Just send me a message on iNat (I’m lynnharper there), or ask in the thread, or send me a message through this Forum.
I hope this helps!
14 Likes
No project “Unknown / Mantodea”.
Observations with Mantodea CV suggestions go to the project “Unknown / Hexapoda”.
The 1000 projects “Unknown / …” result from this process (maybe there were too few Mantodea observations in the set of observations without ID, compared to other taxa):
[…] I grabbed more than 500,000 observations without identification. For each observation, I analyzed the 10 identification suggestions proposed by the computer vision and generated a “best ID”. Then, I made an analysis of the distribution of all these “best IDs”. Then, I designed and run an algorithm for choosing 1000 taxa so that all the “best IDs” could be spread over these 1000 taxa, with the intention to create 1000 projects for grouping all the observations […]
1 Like
I think moving Unknowns to some level of classification helps the iNat database, as well as users and other people doing IDs, than identifying at the species level.
There are people who specialize in all branches of life, but when they search for the things they’re interested in making IDs for or using in research or whatever anything marked as Unknown does’t show up in the search. Moving them from that to even something as simple as ‘dicots’ or ‘bony fish’, or ‘fungi’, or whatever means that they are more easily found and that the iNat database is far more usable as a whole than it is if an Acer sp gets moved to Acer rubrum.
Often for research and inventory purpose you’re not necessarily interested in the exact species breakdown (although that is certainly important in many instances), but in the greater ratios of, say, reptiles to amphibians to birds to mammals, or similar ratios.
And when it comes to finding new species, that can happen at all classes, but it’s often at those higher Unknown, or Family, level initial classifications that these are initially listed as the people first recording them aren’t exactly sure where they fit, particularly in a citizen science context where many of the observers are not specialists.
5 Likes
That could be the case, even I rarely encounter mantises while IDing unknowns. Could be a trend everywhere. thanks!
1 Like
but if you are interested in Acer, whether rubrum or not, would you pick thru dicots for them ? I would put far more value on rubrum than another ahem dicot. There are 1.5 million going back to 1914! https://www.inaturalist.org/observations/270214214
I have still 20 pages of GSB for South Africa at plant families - mostly ignored for the past month. And for the foreseeable future ?
1 Like
I’d say if you have a taxonomic specialism, deal with those first, especially the ones not at RG species level. That is a better use of a specialist’s time than working through unknowns that could be anything from a tree to an arthropod.
13 Likes
yeaaa, if you’re good at a certain order of insects then you can divert attention into IDing all of em to lower taxa. if you’re relatively new (me) clear out unknowns, its an awesome learning experience and often rewarding and maybe you’ll find the taxon you wanna specialise on someday
3 Likes