Well, true, I should rephrase, I meant the basic functionality (making observations, IDing) is not that difficult to understand. A lot of people start in the app, I imagine, and what you can do there is pretty limited, so they don’t even have real access to most features from there.
1 Like
I think it’s very natural to learn what’s needed for a good observation experientially and iteratively, and receiving advice often hits emotionally in a way that’s counter to the learning process. I’ve made plenty of weak observations while thinking, “I wish I’d gotten a better picture” but the process of going through it is teaching me what I will need.
I try to stick to very pragmatic commentary that’s along the lines of “we need to see [feature] to distinguish [this] and [that].” because I think that’s least likely to pick a fight and most likely to help.
I do think that if you’re frustrated by novice users, it might well be feasible (and reasonable) to filter them out in the right identification mode view, or focus your id work on a category that’s less prone to being hit by the CNC flood.
6 Likes
I’ve wasted a lot of my time and others on this subject thinking I’m doing something for the betterment of ……However, I didn’t get anything special from there other than some cyclic/repetitive views and several insulting replies (putting things in my mouth that I didn’t say). Those discussions have done me and others no good (other than refraining me from posting my observations and IDing other’s. It’s not at all an important issue to mention though).
It would be much better for me to remain in the Forum and read the valuable opinions/posts of various wise people on other topics.
PS: A person who is a new entrant in a group doesn’t mean he/she is a novice new born Child. He/She may be an aged person who has spent a lot of life in many experiences.
5 Likes
Why would discussions about observation quality on the forum stop you from uploading observations or identifying stuff on iNat itself?
If you are talking about the discussion I think you are talking about (I won’t link to it so as to not call anybody out or anything), then I must disagree with this. I have read through the replies and they all look very respectful and polite to me. The reason for the cyclic and repetitive views is that both sides didn’t change their views, which is probably just the nature of internet arguments in general. That doesn’t mean that it is not a valuable discussion, as frustrating as it might be to not convince the other side of your own views.
1 Like
I was just goin through fungi of germany …
Yes, people should do as they please. If they want to upload an observation that hasn’t got a chance of identification, be my guest …
But as an identifier, I am extremely frustrated. I am guessing that 99%+ of the fungi are not IDable due to a lack of closeups, different angles, pics of the underside, infos on the substrate ect.
It is just endlessly scrolling through single pic observations in the hope of finding a tree fungus that I could actually try to identify. So I stopped again soon (as always).
I don’t know what to do about it, how to “educate” observers, how to reduce the mountain of unidentified observations. I am not proposing anything, I am not badmouthing laypeople, I don’t think that there should be any form of gatekeeping.
But I find this situation extremely frustrating from the viewpoint of someone who tries to ID stuff. Just sayin’.
8 Likes
What if iNaturalist made a simple animation or video as a guide/tutorial for new members as soon as they create an account? It could even have an option for new members to pick organisms that are of more interest to them, and iNaturalist provide them with a specific (but simple!) animation, to guide them through good practices, importance of photos of the more informative characters, etc., relevant to the broad group of organisms selected.
2 Likes
Again - I think new people imagine that ‘paid iNat staff who are biologists’ do the IDs. Or that IDs are computer automation with no human input. They don’t realise that identifiers are volunteers, more or less skilled, with much or not much time. Newbies don’t realise that they need to pull their weight with identifying. And iNat needs sustainable support for a tiny group of overwhelmed identifiers.
130 people make 25% of iNat IDs. Think on that.
9 Likes
Just to address one point about Seek quickly - the Seek CV model is a small (and, yes, older) version of the iNat CV that fits on a mobile device. Part of the point of Seek is that it can run offline, so it will never be able to have a CV model as good as the current iNat one (due to size/complexity which has increased dramatically!). It will also only have the more common species as a result. That’s not to say that an updated Seek model might not be an improvement, but it’s never going to be on par with the main iNat CV.
4 Likes
Low quality observations are good!
Most people who upload low quality images are fairly new at iNaturalist, and maybe observations in general. Many of them grow tired of it after a while and leave the platform, but some get better and a few become experts. We should welcome the bad quality and cultivate those users.
iNaturalist is uniquely suited to handle anybody’s knowledge or ability, from the worst to the expert. One day maybe even I will progress from the ranks of the worst…
10 Likes
Coming across a group of these observations is definitely vexing, for sure, but outside of the City Nature Challenge, when poor quality observations are common, anecdotally I haven’t seen any major uptick in the percentage “poor quality” observations. As iNat has grown, there’s definitely a greater overall number, but many still seem reasonable.
IMO, posting an observation without any ID to it at all is not something to blame the user for. The app itself doesn’t really make it clear that the observer needs to add an ID, and if the observer is out of cell range they literally can’t add an identification. It’s a design issue, and one that will be addressed in the upcoming app in a few ways. It won’t prevent all cases of it, but it should greatly reduce that from happening.
As for poor quality photos, I think that needs to be defined more specifically. If you mean a photo that can be identified past family (and thus be eligible for Research Grade), I think many photos probably meet that criterion. Unfortunately smartphone cameras, which the vast majority of people will be using, are not designed for photographing plants, insects, or birds. Plants don’t scurry away but it’s often really difficult to get proper in-focus shots of them - something I’ve struggled with recently and I’m pretty experienced with photography and generally know which parts of the plant to photograph.
I’d give that feedback, in a civil and constructive manner, of course, to the CNC organizers and maybe to the organizers of specific CNC projects that are particularly problematic. Explain the burden it puts on the community. I think personal stories like that would go a long way. Posting here won’t move the needle any.
Definitely something that I think would be great, but complex to implement.
I agree onboarding should be better, but I’m not sure forum posts are a particularly useful way to gauge the learning curve. A lot of forums and subreddits deal with the same questions over and over again because people don’t search for discussions that have already addressed the question, or they just want to talk to others. iNat also hasn’t had good and findable help resources, and we’re working on that.
iNat is pretty complex, and so are taxonomy and the like. These are just things that take time and experience to learn. I don’t know one would want to throw a lot at people at first, because it will be overwhelming, but I agree there’s currently very little for new users to latch onto.
For what it’s worth, I made a new how-to you can share, for people how want to help refining IDs to observations. I’ll also note that there’s an entire Tutorials section here on the Forum that anyone could make a tutorial for and link to. If you want to recruit people and help them grow as identifiers, it’s something you could use.
You can have text at the ready that explains to people what kind of photos they should take and use a text expander to easily add it as a comment. It won’t fix the existing observation, but it should help people improve their future observations.
9 Likes
So many replies, so many points of view. I wish I were able to address all the issues one by one, but it would take much more than the time I have available right now. So thanks to everyone who’s taken part in the debate till now and forgive me if I comment on just two of the last replies…
This really gets to the core of the issue. I suppose I tend to see the “worst” side of things as I spend most of my IDing time going through Unknowns in my geographic areas of interest and my feeling is that the vast majority of those contributing “sloppy” (the perfect word) observations just don’t realise they’re part of something much, much greater than a smartphone app. I’m ever more convinced that there needs to be some way to communicate to new users that iNaturalist is a community and their actions/non-actions therefore have a direct effect on all other members of that community. There are real people involved and the precious time and effort they will be investing deserves to be respected. At the very least, that means taking as much care over their own observations as others will take trying to ID them. But how can this be effectively communicated?
Yes of course, if that were actually what was happening, I’d be absolutely in agreement. But with such an overwhelming quantity of observations and so few IDers, who is going to find the time to cultivate those potential experts, to help them learn and get better? Or even to realise that such a thing is possible?
Every observation needs the attention of at least one other IDer to move it out of the Needs ID pool, probably more. And paradoxically, the worse the observation, the longer it usually remains in the Needs ID pool, requiring the attention of ever more IDers before (perhaps) someone takes the courage to declare it impossible to ID. My worry is that the vast stream of “snap and post” observations (a smartphone makes this just SO easy) risks clogging the ID process to the point where some observers and IDers more sensitive to quality may become too frustrated and decide the platform is no longer the right place for them. And that would be such a pity.
4 Likes
I looked up iNaturalist again and looked at the introduction and sign-up process. There are instructions how to use the software and reference to Community Guidelines which are the rules for participants.
What is missing is a description of intended use.
What is iNaturalist recommened for? Why should I use iNaturalist?
If it was recommended that people who want to know what a cultivated plant is, or what that animal in the zoo is, should use Google lens instead (and get additional useful information like where can you buy the plant or what the zoo opening times are) there would be less of cultivated and captive observations.
Recommend iNaturalist as free repository for photos and sound recordings of birds, wildflowers, fish and all wild organisms where a community assists you to identify your observations.
It would also be worth mentioning who is iNaturalist intended for.
You can remain anonymous and still access information or even better, sign up to contribute and engage, iNaturalist is a collaborative platform.
The key to better observations are the keys. An introduction to high level taxomony and links to keys could make wonders. Adding photos or descriptions of features the key asks for to determine taxa makes ID straightforward, whenever that is possible.
Last but not least, challenges are about volume of observations. The balance to quality would be restored if other challenges would focus on completeness of records and quality.
Some of the above could be addressed at low cost and without software changes.
4 Likes
You could almost re-use the code for required observation fields that are collected for projects… but it doesn’t trigger off the onclick of submit. If there’s a feature request to collect required obs fields on the onclick of submit… you could look at re-using that same code for the dialogue.
then you’d have a trusted group of people who create protected obs fields for dialogue questions. As long as they’re all multiple-choice the answer could be selected from a drop down and verified on the client side.
It still might be wayyyy more complicated than this… but maybe worth thinking about re-using user defined obs fields logic.
Here’s one thing that might help with lower quality records: if iNatters would include some information about themselves and their interests in their profile it would put a person behind the pseudonym. When I see rather poorly done records coming from a totally anonymous individual, I assume they are not serious users and are probably just passing through. Not worth the effort of tackling their records. But if they have something in their profile to suggest they might stick around and are interested, then I’m more inclined to look at their records and help where I can, even if the records are a bunch of badly-photo’d Unknowns.
8 Likes
Flipped thru to the CNC FAQ. There is - a lot of good advice there - but who will see it?
2 Likes
I came here from Seek. I am a new user, and originally installed Seek to ID what I saw around me, and learning more about the App created and linked an iNaturalist account to it. But the pictures it took on my phone were often poor, and even with my limited knowledge I realised that quite a few of the IDs were just wrong. So I un-linked it from iNat, and then stopped using it and post directly here now, and I find it much more rewarding.
Personally I would love an experienced mentor or series of education resources aimed at guiding newcomers to better observation entries. I’ve switched now from my phone to a camera for clear pictures, but still often don’t know what are the key characteristics are to capture in the image (subject allowing of course).
12 Likes
I envision, in the future, an AI that interacts with an observer as they try to submit a photo.
“Do you think this is an animal, a plant, or a mushroom?” [Fauna, Flora, or Fungi?]
[observer picks one]
“Would you like some tips on which things to photograph, in order to increase your chances of receiving an accurate identification?”
[Yes]
“Try to photograph the arrangement of the spider’s eyes. This is often used to tell different spiders apart.”
Etc.
This is probably technologically possible now, and is more of a funding issue for iNat than a technical challenge (just guessing here).
15 Likes
I find owning guidebooks to be helpful for figuring out what features are important for identifying species. (Rather than just relying in online resources.) For example, I have stacks of wildflower field guides, and by poring over them and reading the detailed descriptions I have learned that in order to tell similar species apart you often have to look carefully at not just the flower petals but also the sepals, the stem, the leaves, (including undersides, and upper and lower leaves), growth habit, size, etc. etc. So I try to take photos accordingly. Even so, I often find I don’t have quite enough info.
Also, when someone frequently IDs my observations of a certain taxon, and they appear to be experts or knowledgeable, I look at their own observations to see what kind of photos they take themselves.
1 Like
If you mean that the photo resolution wasn’t good, that’s unfortunately a trade-off for live ID suggestions (at least that’s my understanding, I’m not a software engineer). Analyzing full-resolution images at a satisfying rate is either not possible or would probably kill your battery.
1 Like
Most of your obs are RG.
For the others - you can see what obs go to RG for that genus - and see what field marks identifiers need.