if there is anything that should remove ‘reputation’ it is lack of response to @'s or dissenting IDs. It’s fine to hold your ground if you think you are right but you should at least be able to verify why. Maybe there’s a way to track nonresponsive or now-absent ID’ers?
2 Likes
For app users it can be hard to see mentions though.
2 Likes
@charlie - Though I can think of various examples where this would work, at least one key European diptera identifier doesn’t respond to tags - it’s meaningless in his case - I think he just isn’t a fan of the existing notifications system… but he’s one of the only experts we have in an exceptionally complex family.
Thanks @alex - appreciate the response.
I like this idea.
Guess this has already been mentioned on some thread somewhere but why not just let people choose the weight of their ID somehow? Most newcomers wouldn’t even expect their IDs to hold the same weight as those with experience or expertise and might prefer it if they were able to make mistakes without fear of reprisal. Whilst many experienced identifiers might also prefer some of their own IDs not to carry the same weight as others - it’s pretty common to add a question mark or mention something is a tentative ID to place an observation as a possible this or that but not want it to be too committed, especially if out of geographic area of knowledge or stepping outside of the taxa one knows. Sometimes even when I am fairly certain of something I hold back from adding a species level ID, as I know it’ll likely just get blindly agreed to by the observer rather than being looked over by someone else which I’d prefer.
In UK on iRecord, the verifiers have a weighted version like this - some records are accepted as “plausible”, whilst others are accepted as “correct”. The “correct” ones go to NBN (UK equivalent of GBIF), the “plausible” ones don’t.
4 Likes
my guess is at least to some extent that this stuff is just really hard to code and there hasn’t been time to develop that sort of thing, but i could be wrong. I wish there were a way to suggest an ID with less weight, like hey check this species out i think this looks right but i am not sure
4 Likes
i mean this is probably a controversial opinion but maybe we shouldn’t have people using the app to ID until it’s at a point where they can easily see mentions.
2 Likes
It really seems to me some experts are not fans of the system at all, so many of them are hanging around randomly leaving comments instead of ids and not reacting to any notifications.
3 Likes
Yes, they are different. I’m not saying that the founders of iNat are nefariously plotting to make a clickbait site, but that the organization’s vision determines what decisions are made. The metrics iNat can assess are countable items like posts about nature and not something as esoteric as engagement with nature, and that difference matters. How we define an object affects how we measure it which affects how we push for progress with it, etc.
Nevertheless, like you I do find iNat useful when used properly. I’m working on a manuscript exploring how much we can learn about a rare species from sites like iNat if we engage and manage our focus dataset well. iNat has also helped me learn to recognize species I don’t have specimens of, pick up on new characters for identification, and build relationships with other interested individuals. These are worthwhile benefits.
And I actually agree with your other reply, too. I don’t think “experts” need to be PhDs. Even within my professional field when I visit a museum collection there are robber fly experts with PhDs whose IDs I don’t really trust because they’re too sloppy and their true expertise was in ecology. But that isn’t to say true expertise doesn’t exist and we couldn’t find some way of identifying it.
The main bits about iNat’s attitude toward expertise that irk me are that it actually breaks its own ideal already by valuing some types of authority over others (e.g., we give curators super powers to set up the taxonomy, we allow users to opt out of community ID, etc) but refuses to consider that there might be ways to further improve the site by trying out some of the various proposals floated here over the years. Yes, it might be hard to code, but that doesn’t mean we shouldn’t try. iNat’s attitude toward primary literature is also frustrating, because it works for well-known charismatic megafauna like birds, but in the world of insects there are no reliable third-party databases and many genera don’t even have keys let alone guides.
People will still fight over whatever system we set up. That’s human nature. Conflict in and of itself isn’t bad, though! That’s the only way we make progress.
I agree that we need growth (people with expertise) to tackle growth (in observations). But until society agrees that biodiversity matters and will pay for this, it’s hard to get the people who already have the knowledge to spend their precious time to write new tutorials / keys, and hard to get the public to even care about accuracy for tiny insects with no practical use. The Dankowicz brothers have done a great job at trying to wade into this with their fly guides, but even those have limits and challenges.
6 Likes
that’s true, but it doesn’t add much value to the site most of the time
2 Likes
i actually thought they weren’t strict about the primary literature any more, or else some of the plant curators are breaking the rules, because plant taxonomy has been hacked apart to the point many groups of plants are barely possible to track on iNat, and i’d rather we stick with older standards. That is going off topic a bit but also links into the issues with IDs, based on how subspecies are being used and such
We (and by “we” I mean the more technically inclined members of our team plus some other collaborators, not me) actually did look pretty hard into whether we could algorithmically determine’s how “good” an identifer is on iNat (you can read a bit more here) but as @charlie says, it’s fiendlishly complicated and would create different incentives which would create different (maybe worse, maybe not as bad) issues. We’ve also done some dives into how reliable IDs on iNat are and will continue to explore that.
We’ve also changed the CV suggestions so that by default only taxa which have been “seen nearby” are shown, which should cut down on California taxa being suggested for observations in Malaysia and the like, and @abhas has some pretty great designs and ideas for onboarding which we’re working on. I think the latter is extremely important and will hopefuly nip some of these issues in the bud.
I have to take some of theblame for this narrative since I’ve quoted it without nuance or context quite often. If you look at the What is it? page you’ll see:
our primary goal in operating iNaturalist is to connect people to nature, and by that we mean getting people to feel that the non-human world has personal significance, and is worth protecting. We have a pretty nerdy way of doing that, of course, but we really believe that recording information about nature in a social context is a tremendous way to understand the awesome depth and breadth of life on Earth.
Our secondary goal is to generate scientifically valuable biodiversity data from these personal encounters. We believe iNat can achieve both of these goals simultaneously - in fact that they reinforce one another - but when we get pulled in conflicting directions, we measure success by our primary goal. If we connect people to nature without contributing to any specific scientific outcomes or quantifiable conservation results, then we’re still doing our job, but if we just contribute to science without helping people care about the natural world, we’ll be on the wrong track.
So yes, the primary goal is to connect people to nature, but we believe that helping people connect with nature and generating scientifically valuable data reinforce each other, so they’re somewhat inextricable and I can tell you all of us on staff think the scientific aspect of iNat is extremely important. But it’s inevitable that when making decisions we may have to prioritize one over the other and in those cases the connecting people to nature goal generally takes precendence.
18 Likes
I would suggest that everyone can already do this, by
- choosing whether or not to leave an ID at all, and
- choosing the taxonomic level that matches their confidence level.
And this can be tailored differently by the individual for each observation. The challenge, of course, is to give folks better guidance on how to make these choices thoughtfully. (Including a clear understanding that every “agree” leaves a new ID, and is not just a “reward” for the other identifier.) So glad to hear about the
5 Likes
I am routinely surprised at some of the cynicism towards iNat even from within its own forums. As someone who is very new to the platform and is primarily an observer (apart from trying to help out with identifying what I’m familiar with in the local area) I feel very positive about iNaturalist, but I think it is a function of what you expect from it personally and as an overall outcome.
For learning, searching for species, and trying to record my personal observations it is unparalleled. Many of the features available are excellent compared to anything vaguely similar, and as a result I feel like my learning has accelerated rapidly. The AI is very useful for giving a starting point for figuring out groups even if you can only suggest a higher level taxa, and personally I think suggestions are great for a starting point (which I will promptly withdraw if it disagrees with others and I have no other prior). From my limited perspective it also seems like the majority of users do interact sensibly, are not identifying the majority of things haphazardly, and respond positively to discussion over ID suggestions. Overall I think the experience for the individual willing to stick it out is great.
In terms of the use of the data I am a little cynical as to its use directly from the platform but I also think this is not really a problem. Obviously it would be great if exported records were perfectly correct, but I think the analogy to a field notebook is much more sensible. I think iNat has the opportunity to take a position as a leading submission portal, to which some identification prior can be attached. From a UK perspective respective recording schemes should then vet used records as they already do (this current iRecord/iNat tension is ridiculous it can clearly become symbiotic). Additionally, there are extremely strict limits on what you can do with this data anyway, as observations are not collected systematically in any sense. But if you consider ‘Research Grade’ to just be a prior annotation for a record (which is obviously the limit for many groups that require expertise and equipment beyond casual observations anyway such as microscopy for fungi or insect genitalia) then I don’t think this is inherently a problem.
Mistakes are inevitable, and the scope of the endeavor is huge, but I think through a combination of community development as well as realistic expectations as to the limit of the data it can (and maybe already is?) settle into an important niche on an international scale. At the very least if this platform provides inspiration, education, and tools for future naturalists that states are unwilling or unable to provide for then that is a huge win.
17 Likes
i agree, while i have of course had my own feedback and such, and things i wish were different on iNat than they are (of course, because the site was not made solely for me and my weird nerdy quirks :D ) it blows anything else away and has by far revolutionized how i look at nature. Even if most other peoples’ data was no good, which isn’t the case, it is super useful to just look at my data and others using projects i create. It’s nothing short of a complete revolution in community ecology, for real.
6 Likes
I don’t have stats, but I am sure “majority” doesn’t actually do things as good as can be, as any comunity minority does the biggest job, so minority of iders do most of ids and minority of observers make the most observations. But most of users on iNat do 1-10 observations and never come back or add random id to someone else’s observation and leave, they don’t spend enough time to “acclimatise”, nor they really want it as they check and go (which can be their problem or app/website problem). From earlier staff posts we saw that half of accounts are just blank ones.
We can’t just turn the blind eye on a problem, as we do now, we need experts and their effort, they also need a reason to stick on the site, if we just say “hey, it’s gonna be filtered anyway, so why care?” problem will stay forever while it is actually solvable.
Also should point out ost of venting is from ider perspective, seeing thousands of pages of something barely recognizable as life ided as species and confirmed by user’s friend or two. As an observer you don’t really see that unless you go check taxon ranges and all observations of it.
6 Likes
I was not saying there are not problems or the problems are not worth solving, just that I feel optimistic about the current state from my perspective :). I am quite interested in the contribution demographics, maybe I will have a play with the API to look at the distribution of ‘account experience’ within RG observers and identifiers. Unless anyone can point me to someone who has already looked at this of course!
4 Likes
my experience over this past decade on iNat is if someone tries it and doesn’t get into it, there’s no real point in prodding them. Most valuable content comes from power observers and power ID’ers and for others the site really is more of an outreach and education tool. Likewise many ‘duress users’ add content when assigned for a class, but they too leave and never curate their data, creating a headache for us when it’s a large or repeating class. For ‘experts’ who try iNat (or don’t try iNat) then disengage and gripe, they aren’t going to do any different no matter what is done, the only thing that changes their behavior is if others around them use it and talk about it. (to be honest i think most ‘experts’ who dislike iNat dislike it because there isn’t the power and privelige structure they have around them in universities, they are called out on mistakes and normal unwashed amateurs can challenge them and they don’t like that) I think it would be interesting to create a separate portal and different functionality for us power users, but classifying who is or isn’t a power user could be tricky too.
5 Likes
I disagree each expert that leave is a lost case anyway, if I weren’t observer I would see the mess that has no end and would think a lot of why doing it at all, people can not love order, but they surely love seeing progress in resolving issues, now it’s a war where you will loose no matter what.
2 Likes
It isn’t a disparagement of those users, just a power user is created by a certain range of criteria and i don’t think we can turn casual users into engaged power users. I have other theories that relate to one of my other threads but will leave it at that :)
There is an inverse relationship though between confidence and the taxonomic level of IDs users commit to. New users will be the most overly-confident in going to species level in complex taxa. Which is why this isn’t working.
Sure. But I’d argue that the optimal way to guide folks to choose thoughtfully as you say would always be something intuitive and inherent in the design - a shift in affordances. More granularity on the sort of ID people can leave, would be one approach to this.
At the moment, the interface gives users one big red button to push.
Surprise surprise, they push it.
7 Likes
Hello again, I thought people interested new accounts making identifications might be briefly interested in this. Not sure how you would look at the impact of users making lots of guesses though as this is hard to detect systematically?
Using Scotland as an example because the request was within the recommended practices for the API, I searched for identifications that matched the observations ID on RG observations, and a little user information as well. Counted up some stuff and I got this (activity count is the sum of observations and identifications, the number in brackets is the total number of identifiers in the category):
100% of all RG IDs in Scotland by users activity count > 0. (8317)
98.86% of all RG IDs in Scotland by users activity count > 10. (6796)
91.88% of all RG IDs in Scotland by users activity count > 100. (4656)
77.65% of all RG IDs in Scotland by users activity count > 1000. (2825)
54.68% of all RG IDs in Scotland by users activity count > 10000. (1060)
26.75% of all RG IDs in Scotland by users activity count > 100000. (139)
(wow some people have made many identifications!)
I only did this quickly but I hope I didn’t make any mistakes. Will have to check through, make some plots, and I might expand on it and make a separate post. Seems like most identifications that match the consensus of research grade observations are done by users with medium or high activity, depending on your definition. If this held true with some added robustness and extrapolation, I agree with sbushes that design decisions could reduce the effect of new accounts but also reduce the frustration experienced by dedicated identifiers.
For anyone interested or good at spotting mistakes my API request URL was ‘https://api.inaturalist.org/v1/identifications/identifiers?current_taxon=true¤t=true&place_id=7382&quality_grade=research&page=1&per_page=10000&order=desc&order_by=created_at’ per page.
3 Likes