I wish to report that about 65 dedicated and hard-working project members (plus an unknown number of other IDers who were not project members) can reduce the pile of plant Needs ID observations in New England and New York by 2.2% (or 34,766 observations) in 48 hours, in case anyone is looking for more identifying anecdata. By comparison, last weekend over the same time period - but without a concentrated effort - the Needs ID pile went down by 1,119.
Hey, identifiers, you should know about the new way iNat wants us to deal with multi-species observations - read the details on this blog post: https://www.inaturalist.org/blog/90263
Thanks for telling me; never would have known that little gem was in a blog post with a totally unrelated title.
Same, I randomly clicked on an active thread and saw a major Life improvement!
Now this pile will be getting smaller. It’s nice that all these photosets will finally have a proper resting place togther in Casual.
edit- just found overlapping project
I’ve been working on that project a little already. I’ll do more soon.
Someone should add a thread to the “News and Updates” category. I think this is a really important addition and I also would not have known about it otherwise.
If anyone wants another pile of them, this one isn’t great because it’s not in Identify, but it is very large.
edit: Loarie converted this into some Identify links over on the other thread (over 10,000 obs!)
News and Updates is a monthly roundup.
That’s very handy to know. I would never have thought to check the blog post for it either
Oh that make sense. my bad
I waded in to wrestle a bunch of them. Apropos the new DQA discussion, I may have been a tetch overzealous in a couple of cases, but I did my best to exercise restraint. It’s not as though we’re going to run out of links anytime soon.
Friday!
A question: Do any of you worry about the state of identifications on iNat ten or twenty years from now? Here, I’m not talking about the perpetual dearth of identifiers, nor am I concerned about the finer points of how to apply a DQA annotation. I’m not even worried about lumping and splitting and name changes and all that taxonomic hoop-jumping. I’m worried because I see growth in a class of observations that aren’t easily identifiable.
I can best clarify this by an example, I hope: Very Competent and Experienced iNaturalist Jane, in the course of making a botanical survey of a conservation property, makes an observation of a common shrub that’s not in flower or fruit, a species that has rather ordinary leaves. (Those of you in eastern part of North American can envision something like Northern Spicebush or American Fly-Honeysuckle here.) Since Jane is very experienced and competent, she identifies it correctly when she uploads the photo to iNat. Very Competent and Experienced identifier John comes along the next day and thinks to himself, “I bet Jane’s correct, but without flowers/fruit/etc., I can’t be sure what this is. So I’ll just move on to the next observation.” Years go by, ten identifiers look at that observation, but none feel confident in agreeing with Jane’s ID. Also, none of them feel sure her ID is wrong, so they don’t post an ID of, say, Dicot.
Example #2: Ordinary observer posts a photo of a conifer taken from a distance and IDs it as Eastern White Pine (a very common conifer around my region). Or they post a photo of just the bark. Diligent identifiers think, “I’d really rather see a photo of a cone or the number of needles in a bundle. I don’t feel comfortable agreeing with it as Eastern White Pine; I’ll just move on.” They also aren’t confident the observer is wrong in calling it an Eastern White Pine.
Essentially, these observations are unidentifiable at the species level (or even genus), but they are also “too good” to end up as Casual. What do we do with such observations?
I suppose this worry of mine comes from my wanting to complete the task of identification, which I recognize is a foolish desire (to put it mildly), but I also wonder if the sheer number of these unidentifiable observations is daunting to would-be identifiers.
For me personnally the question is: do we need to worry about completed IDs when we think about the general purpose of iNat (“iNaturalist provides a place to record and organize nature findings, meet other nature enthusiasts, and learn about the natural world” from the FAQ)?
So from my perspective as someone from both sides (many observations vs dipping my toe into the pool concerning IDs):
Yes, of course we all love to satisfy that urge to tidy (shiny RG! Advancing Science!Empty that Needs-ID queue!).
But does it really hurt that observation if it never gets that second confirmation to lift it to RG?
As someone coming from tech and data science, my faith is that on one side the AI and the methods themselves will grow and get better and better (ChatGPT for identifications!), freeing IDers up to (maybe) develop into experts in their small corner of nature.
But on the other hand, my faith is also into the community and those engaged IDers, which is proven everytime one of my lost fungi observations gets a further ID (even if they might never get to species, which is my fault since often I’m too lazy and pressed for time to get those needed key identifying features…) or somebody IDs an observation uploaded 5 or 10 years ago. IDing in iNat sometimes feels like an ant hill or bee stock to me: many many workers doing their thing - sometimes working in tandem and sometimes a lone worker coming your way (and observation).
The main thing is just having patience - and that has been rewarded often in my case!
I don’t worry too much about this because I prefer to focus on the observations that can be ID’d and how they’re useful. However . . . .
Yes, you’re right, the unidentifiable do accumulate. I see this when identifying – or trying to identify – observations from two or three years ago (in taxa/places I can expect to ID). The percent unidentifiable will only grow. It’s frustrating, though not too much of a discouragement to IDers who work in the last year or two.
Right now, the best we can do is mark “No, it can’t be improved.” I hate to do this in most cases because I think, “I can’t ID it but maybe someone else can.” With observations that have waited a long time, though, we should probably click that button more often.
The urge to tidy! Yep, that’s me (but not vacuuming).
You are right that it doesn’t hurt the observation if it never makes it to RG (or to Casual, for that matter). In fact, I bet some observers would prefer that their observations remain at Needs ID rather than reach RG at the genus level.
I have a lot of faith in the computer vision, too, but I also see many observations that fool the CV, even here where the CV is often correct. I think having human identifiers check on things is the best way to go.
I am often tempted to mark something as “can’t be improved” but I know very well that my ID skills are limited, so usually I don’t, except on my own observations. And maybe older observations of, say, a Sphagnum at a distance.
Yes it is. But I have convinced myself it is okay to move on - to an obs where I CAN ID.
That urge to make order out of chaos - led me to the Pre-Maverick Project.
And identifying African Unknowns - where I have a backlog about 800 (first one I see is 3 pictures of the same plant - spread across 3 obs - I see copypasta in my future)
I may be overly positive about the future, but I don’t worry to much about it. Ten years is a lot of time, and maybe a “new Jane” will start identifying stuff and be sure enough to agree to the “original Jane” ID. Or maybe someone expert enough will show up and know that there is a common Pine and a less common Pine in the area, that can’t be distinguished with bark alone ? Who knows ?
When searching through old obs, it is obvious that only the difficult ones are left. But being able to tidy up a bit what’s left is kinda rewarding, it’s like finding a hidden gem that no one found before you. I guess it’s a bit of a mindset, but you can only be “that” perfectionist when IDing, else it is really too frustrating.
Decided to work on Primrose-leaved Violet since it is blooming right now. Lots of new obs but looking at old ones too.
How long will that be in AI years? We don’t even know what the scale will be yet…