I love it if someone says they find “my” taxon intriguing, and would be delighted to go back and forth with them about specific examples. Discussing and checking references on one ID for an hour is way more interesting than plugging through a hundred in the same time that nobody seems to care about.
4 Likes
@jhammock one area that comes to mind is leafminers. Observations added to the Leafminers of North America project are definitely watched by @ceiseman. Maybe that’s a bigger group than you want to try to tackle, but there’s probably a bunch of unidentified leaf miner observations out there.
Or if you want a more commonly observed and smaller group, Lonicera is probably manageable and definitely immediately useful for the City Nature Challenge ;-) Should be lots of instant gratification once you learn the few species around here (and you can get some of these wrong ones off the list too!). (Note for others: Jen’s in DC near me so we also know each other in real life)
1 Like
we have morrowii and tartarica and they are nearly impossible to tell apart from photos and hybridize so that is a pain
Similarly in my area most are actually maacki x tatarica (L. x bella), which, like you note, needs a good shot of the ovary anatomy in flower… probably most IDed as either species are hybrids. Some typical tatarica too.
Whelp, looks like I should rescind my Lonicera suggestion and be more careful with my IDs! Sigh.
I have not read all of the comments here, so my apologies if these points have been addressed already. So, first, I think we don’t need “identifiers”, we need identifications that are likely to be correct. Those are different things. Getting identifications that are likely to be correct fundamentally relies on making iNaturalist a fruitful place for people with taxonomic expertise to spend their (limited) time. As a person with taxonomic expertise (I don’t want to toot my own horn but I do want to be clear that I am not being inaccurate in this assessment–I have a PhD in plant systematics; plant taxonomy and identification has been a central component of my work for the eight years since that PhD; I have published multiple papers on plant taxonomy, including multiple new species), there are a few things about iNaturalist that I think are likely to make it less appealing to people in this category:
Taxonomic inflexibility. People who work in taxonomy are, as a general rule, knowledgable enough about the subject that following an “authoritative source” like POWO is not satisfying. These are the people whose research is crudely summarized in a resource like POWO. “Oh, we can’t call it that because we are following POWO,” is, frankly, insulting to someone in that position. I’ve brought this issue up repeatedly in the old google group and it seems like most people who participate in these fora don’t see this as a pressing concern. I believe this is because most people who participate in these fora are not in the user group we are talking about here. If you are not a taxonomist, taxonomic freedom is probably not a pressing issue for you. So I feel like this is a real issue that is consistently underrated because we are, basically, asking non-taxonomists how to get taxonomists involved. And then it comes up again a couple months later, and we again ask non-taxonomists how to get taxonomists involved–and don’t really close the loop that this isn’t how to answer the question, it didn’t work last time and it is probably not going to work this time.
Related to the above, the iNaturalist taxonomic database is really contrary to how actual taxonomy works in a lot of ways, and thus counter-intuitive to many taxonomists. It has the feel of what would happen if you asked a bunch of biodiversity-interested coders to recreate the nomenclatural codes and they all thought, “Oh, that doesn’t work right, we can do it better.” So in many ways iNaturalist has kind of its own parallel nomenclatural system that shares a lot of superficial features with systems like ICNafp, but is in other ways deeply incompatible with it. Anyone who has put the time in to thoroughly learn the ICNafp or ICZN systems is probably not going to be all that enthusiastic about putting in a bunch of effort to learn a new parallel system that does kind of the same thing but on fundamentally different principles.
Difficulty getting to “interesting” observations efficiently. For someone who is a specialist on a particular taxon, this problem is greatly reduced–you just do a search by your taxon of interest. For someone who is a generalist, this is harder. This is why topics like student projects and cultivated observations have come up in iNaturalist fora repeatedly. Student projects aren’t “wrong” but they have a very low signal-to-noise ratio for most people with some expertise. The same goes for cultivated observations. If there are a large number of observations that you, personally, aren’t interested in, and it is difficult for you to filter those observations out and get to the ones you are interested in, iNaturalist becomes less rewarding.
Egalitarianism. I’ve already hinted at this above… I think egalitarianism is highly desirable in many ways, but in past discussions ideas that people who have expertise should be catered to in any way are pretty quickly shot down on egalitarian grounds. “If they’re such prima donnas, why do we want them in our community?” Well, if you want accurate IDs, that requires expertise, and a “flat” social organization frankly does not do a good job of recognizing or rewarding expertise. I expect that my post here will get some of that same reaction. Look, folks, I get the egalitarianism schtick and in many other social contexts I would totally be on your side, but if the question is in fact how we get accurate IDs, strict egalitarianism is not the best system. You’ll be wondering why we don’t get enough people coming in and providing good IDs, suggestions like mine will be provided and shot down, and you’ll be wondering again in a couple months why we don’t get enough people coming in and providing good IDs. We need to close the loop, here. Recognizing / rewarding expertise and strict egalitarianism are not compatible concepts. We need to choose one. If we choose egalitarianism, we need to stop wondering why we have an expertise gap.
Undoing the work of the AI. I feel like this has been an increasing problem… the auto ID feature gives someone a species-level ID, they take it on faith, and then you’ve got a bunch of observations with wrong IDs, and working through Identify starts to feel like half of what you’re doing is undoing the AI. I’ve generally been steering away from this in the past, but I’m increasingly realizing I really should be going through all those observations with incorrect species-level IDs that I can’t provide a good species-level ID for and, well, putting the genus-level ID on there with the “no, really, the species-level ID is wrong” box checked. I find this one of the least fulfilling ways to spend time on iNaturalist, but if I believe in the idea of having accurate IDs on observations here–and I really do believe in that goal–I do need to put in my time to reverse all those damned auto-IDs. A genus-level ID that I can’t get to species harms no one; a species-level ID that is wrong is misinformation that needs to be corrected. And a lot of that misinformation is coming from those auto-IDs. I’m not saying this to bash the auto-ID feature, I think it’s really impressive that it does as well as it does and I hope it continues to improve–but if you’re producing a big stream of misinformation in the process, that’s a problem and it creates a workload that is particularly unsatisfying.
20 Likes
We have a lot of lessons learned in this space and would be happy to share. We also have a lot of proven experts who would be happy to contribute identifications (we’ve offered this before, but this was not approved of).
Try to put yourself in the mindset of an educator for those, it works for me. Every observation that you correct in that way is teaching the computer vision to be more careful when identifying. One day hopefully you can then focus on the difficult cases in your group of interest and the AI do the more trivial ones with high confidence.
5 Likes
A lot of good points here, many of them likely “beyond the scope” of what iNaturalist is and does. Of course it is frustrating to a professional taxonomist to tackle a knotty taxonomic or phylogenetic problem, publish the most informed conclusion and have it ignored by POWO and iNaturalist, or at least go through a lag time. But that is the reasonable compromise that keeps us from being whipsawed by controversy and allows a large, somewhat lumbering entity to interface with the public.
So what I take from your comments is that what we really need in identifiers is not so much laser-focused expertise as generic expertise, people like educators, wetland delineators, avid amateurs – not so much the generators of new knowledge as the users and disseminators of reasonably established “knowledge.” Personally I’d love to get every college professor who teaches a taxon course on board as an identifier. People who are keen on field biology and keen on helping other people learn.
8 Likes
To be clear the goal is to follow the defined references, not the requirement. Those are not the same thing. The site supports, and there are many examples of accepted deviations from both POWO and other defined references in the iNat database.
The goal is to have those deviations be both scientifically AND iNat community supported and not have the taxonomy subject to one person\s vision, and in particular to avoid the issue of constant swapping back and forth as passionate curators refuse to accept a change.
8 Likes
I’m not sure how true this is for the large amount (majority?) of cases where it is difficult or impossible to get an observation to species level. The computer vision currently can’t train off of non-species observations as far as I know. So if someone identifies something, using the computer vision, as an easy to identify species that clearly doesn’t match their observation, and I move it back to a higher level (genus or even higher often) and it stays there, then the computer vision won’t be able to learn from that (unless I am missing something).
Sorry–I haven’t had time to read through the whole thread, so if this has been mentioned, just ignore. I know the Cornell site offers courses in birding–which leads me to ask, are there other sites that offer courses? If so, is there a reference list here that someone could go to and consider? I realize that an online course isn’t going to make anyone an expert, but it would help ordinary persons (like me) improve their abilities a little to help with identifications. Plus, it’s just fun to learn more.
4 Likes
nah we need to figure out if there is a ‘section’ we can make for the two and the hybrid,because it doesn’t really matter which one you are seeing for most purposes anyway, they are both invasive. But putting them at Lonicera lumps in some neat native ones too. But i am drifting off topic.
Disagree with this and the overall great success of iNat (despite its imperfections) is the most emphatic proof that this is a bad idea. People have tried creating what you are suggesting and they have all pretty much failed. Unless you count super specific stuff like bugguide. In fact i think the push for non egalitarianism is a contaminant to science in general and has led to much nastier things such as the racism and sexism that has rooted deeply in academia as well.
I agree strongly that we need more academics, land managers policymakers, ‘professional’ naturalists, and experts of all sources involved in iNat. But as one person who is one of the above i am honestly really bugged by people implying that we would not remain here without special treatment. Life is not egalitarian sure, but it’s not too much to ask that our reputation and social capital come from how we interact within this community. We don’t need some external entity telling us who should get higher weighting for their IDs or whatever. If you are the world’s expert on blowflies, well, you will have better blowfly IDs than anyone else and your reputation will build. I don’t need some institution telling me to give oyu more blowfly points. Earn it like everyone else.
The issue with data quality is 95% or more due to the duress user issue. If we fixed that we’d see an explosion in data quality and more adoption of iNat by experts. I have no doubt.
10 Likes
I do not have any data on which of those cases is the majority: de-identifiying it into a higher category or identifying it as a different species. Since @aspidoscelis mentioned he has some professional training in identifying plants I assumed for him its the latter.
It is true however that if you put it into a higher category and if the photo would therefore be not used in the training of the AI the information that it is not this particular species to which it has been wrongly identified is lost. However, only as long as someone identifies it to correct species then it is back and corrects the AI.
2 Likes
One way of recruiting new IDers is for current IDers to craft and upload images that are marked up with easily digestible ID tips (text, arrows, etc.). Here’s one I made for Calopteron terminale, a secondary upload for this observation. So if I make an ID for a Calopteron spp. and the observer is curious about why, I can quickly provide either a link to the image or even embed the image in a comment. I know of at least one person who then went on to use that information to correct misidentified species in the genus.
I don’t have data, but I think conveying ID tips visually can convert interested users into IDers at a higher rate than just providing ID tips by text. It takes a few minutes to create such graphics but I highly recommend them. I also like making them, in part, because the process teaches me about the organism. And, if it matters, I’m not a Calopteron expert. I just decided one day that it might be fun to learn how to ID them. I think iNaturalist is full of similar folks … people who are genuinely interested in improving their ID skills but don’t know where to start. It’s just hard to pick up ID skills with the current system (the “why” of an ID is rarely communicated).
18 Likes
One type of email that might get my interest is a plea to ID a particular taxa that I might be interested in. E.g., it might be clear from recent activity that I’ve provided some IDs for species X. Your letter might ask whether I might be willing to start Following that taxa. Once a taxa is Followed a user is much more likely to proof IDs as they come in. The email might provide a URL that does that in a single click. You might also ask me to consider going through past IDs to clean up misidentifications (with perhaps some tips on how to do that). I’m sure there are folks on the forum who will be violently opposed to such outreach but I think you might, still, find some folks who would respond favorably to such suggestions. These pleas might also be displayed on the Dashboard, of course.
5 Likes
This has been my main motivation for identifying to this point, but I don’t know how to communicate why other people might want to do it for the same reason.
4 Likes
I’d like to second the idea for some sort of graphical bonus for those who spend time identifying. E.g., if a person’s ratio of IDs to Observations is > 1, the user icon would be modified in some way. Displaying the ratio might be too boring … but perhaps there’s a way. Currently the only way to know whether an ID provider is actually somebody who IDs A LOT (and presumably has the expertise) is to jump over to their profile and snoop. I do that constantly and it would be nice to know that just from user icon.
3 Likes
This is soooo important, I think. I wish the suggestion could be baked into the instructions somewhere. Apologies if it’s already there.
I’d also really like to see observations on the Identify page be sorted to benefit those who provide IDs. It pains me when I travel to a high-IDer’s page and see hundreds of unidentified observations (out of their are of expertise, obviously), many of them years old. I know that there’s a mandate to provide IDs for all the new users, but a little love for the IDers might go a long way to making active members stay active.
4 Likes