iNat has said this will not happen, and I don’t think it should. While it may persuade some taxon experts to make (more) IDs, I think it will dissuade a lot of other people and will result in fewer IDs overall. Additionally, such a change would be an insult kinda to all the non-expert IDers who do good work already.
Also, if an expert could just de facto override any previous ID, there’d be no need for discussion, no need for comments, and far less helpful information that helps everyone learn.
As it is, RG observations show a high degree of accuracy, so I don’t think there is any need for any changes like this.
4 Likes
To help suggest an identification…to help improve the identification…for both the observer and for iNaturalist - is the reward unto itself. I keep a note on my desk to remind me of both confirmation bias and thinking about Type I and II errors when involved in the identification mode. I saw this above:
Questions: is this a huge problem? Or has this been quantified in any way?
2 Likes
Scott,
If you highlight the text with your cursor, Discourse will give you the option to “Copy Quote”, which you can then paste into your message:
3 Likes
i don’t think there should be rewards / badges for specific identifiers, but i don’t think it would be terrible to have metrics for taxa and places. i think it could help with group efforts to identify, and it could provide additional useful information.
for places, i think it would be nice to have a breakdown similar to what’s provided on this page: https://jumear.github.io/stirfry/iNat_obs_counts_by_iconic_taxa?place_id=18.
for taxa, i think it would be nice to show:
- RG to verifiable obs ratio
- verifiable to all obs ratio
- average number of IDs per observation, maybe broken down by overall and needs ID.
- % of obs with an ID by others
- % of obs with multiple IDs having a disagreeing ID
- average days between obs submit and first ID by others.
- identification category breakdown (ex. https://jumear.github.io/stirfry/iNat_id_summary_by_category?taxon_id=47735)
i think another way to encourage group efforts is to add mechanisms to make it easier to work with others. for example:
- improve the notifications system in general
- make it easier to find when others make disagreements (so that folks can follow up quickly on those)
- provide folks with a way to show their status (ex. “working on identifications, feel free to ask questions” vs “do not disturb”)
10 Likes
No, we want to encourage knowledgeable IDs. This is not the same thing as IDs by taxon experts.
As previous discussions have pointed out, the problem of how to decide who is an expert (in which taxa? where?) is non-trivial.
And there are other users who have been motivated to gain expertise as a result of IDing on iNat. I suspect many of these people (I am one, but far from the only one) would probably never have felt that there was space for them to learn and grow if there were a tiered system where the IDs of “experts” counted more than the IDs of everyone else. The fact that anyone can participate may be a weakness of democracy, but it is also one of its strengths – it means that expertise gets shared among more people; it is no longer the task of a limited exclusive group to manage an ever-growing task.
You say that people have left the platform because they think their expertise should count more than that of regular users. There is, however, another problem, and that is people giving up in despair or burning out because there are too many observations and too few skilled IDers to look at them.
Democracy does not guarantee that people make good decisions, of course. One key requirement for it to work well is education – which again requires that experts do not insist that their knowledge be a domain reserved to the select few.
The vast majority of observations on iNat do not need extensive specialized training. Even in difficult taxa (e.g. many arthropods) there are usually at least a few species that can be learned with only a little bit of effort and attention. There would be no advantage to prioritizing the IDs of experts for such taxa.
It really depends on the taxon. There are lots of taxa where the error rate is close to zero. There are others where a fairly substantial percentage of the observations are mis-ID’d. If there are no taxon specialists who have gone through to try to clean it up, and if there are misconceptions about the taxon – for example, people assuming that there is only one species like it because it is the only one represented on iNat or the only one mentioned in popular field guides – it is not unusual that these observations will get confirming IDs and be RG for a very long time before getting corrected. For these taxa, some of which may have tens of thousands of observations, it is indeed a problem.
13 Likes
thank you.
1 Like
Questions: is this a huge problem? Or has this been quantified in any way?
[/quote]
A few months ago, iNaturalist ran a big experiment to try to determine the rate of correct identification. Conclusion: 95% correct overall, but varies widely with taxon. Sometimes the rate of correct identification is 100% or nearly so. Sometimes it’s pathetic. Any researcher planning to use iNaturalist data needs to sample observations in order to figure out what the rate is for the taxa of interest.
10 Likes
Just a semantics problem, by “taxon expert” I mean someone that has the knowledge to correctly identify a taxon.
I don’t think this would be the case. There still will be observations not identified by experts where democratic identification would hold. In the case where an expert was identifying things, you could study theirs. I don’t know why it would discourage me to have someones ID hold more weight than mine if they were truly more knowledgeable. In part, I learned my cicada identification from studying an experts IDs. He was one that left the platform for the reasons discussed above.
1 Like
This has been said for as long as I have been here. What can be done about it?
1 Like
Yes we have been talking about that for nearly 6 years! https://forum.inaturalist.org/t/recruiting-more-identifiers/2388
1 Like
I think that the problem is not that simple. It would be good if the task of identifying was more rewarding for those qualified to perform it (and that’s a spectrum, not a yes/no). If would also be good if the observations were uniformly presented in a form that made identification possible (and that is a pipe-dream, given that some organisms can be identified to species level only by dissection). So perhaps iNaturalist should be tailored to those areas where citizen science is feasible with a reasonably high degree of accuracy.
I would like to say that I have no objection to having my observations and ids vetted by highly qualified individuals such as museum curators or even private citizens well in touch with the technicalities. But identifying who these people are (and that too is a spectrum) immediately presents a new level of complexity…
2 Likes
No it isn’t just a matter of semantics. If you talk about experts, you need to decide how to determine whether someone is an expert – i.e., what makes an ID by a particular person for a particular taxon more reliable than an ID made by a different person.
It makes the reliability of an ID an attribute of the person making the ID, not an attribute of the knowledge that goes into the ID. It also suggests that expertise is static and not a process.
An expert can make an unknowledgeable ID if they are IDing outside their area of expertise.
A non-expert can make a knowledgeable ID if they put the relevant diligence and research into determining an ID, even if they previously had little knowledge of the taxon.
So: how would you measure “expertise”? Formal qualifications? Experience IDing a particular taxon? How far would this expertise extend? Would it cover only a specific region or would it extend globally? Would expertise in species A automatically mean that that person is also an expert in sister species B and parent genus X?
Because this is exclusionary. It says “you are not welcome” because you have not earned a right to have an opinion. It says, we have decided in advance that you cannot possibly ID this taxon as well as other people. It says that specialized expertise matters more than generalist knowledge.
Given that, based on comments in other forum threads, one major barrier to becoming active as an IDer is that users – particularly laypeople without formal training – feel they don’t know enough to contribute anything, having a tiered system of ID weight would only reinforce the impression that IDing is the domain of formally qualified experts and the rest of us have to stay quiet and be grateful for the gobbets of knowledge bestowed upon us from on high.
I hinted at some directions in my post. Improving the infrastructure to make IDing easier. Better support for collaboration and networking and sharing knowledge. Better onboarding and communication about the ID process, to encourage people to reflect on their own IDs and the IDs that they are provided (both CV suggestions and IDs by other users).
I think in some ways the CV is a weakness of iNat as well as one of its strengths – it encourages engagement and helps awaken an interest in nature through discovery (being able to photograph something and an immediate ID), but it tends to hinder the next step, which is thinking about how do I recognize it myself and why is it that species and not something else. I would love to see a CV implementation which would assign the user a somewhat less passive role – maybe walking them through some basic questions before providing a list of suggestions.
4 Likes
There’s an identify section? Oh god it’s so much easier to use as well 
How am I just finding out about this?
6 Likes
Two more thing for you to try.
If you are playing with filters in Explore. Bottom right is the word Identify - click that - and you have the MUCH easier interface to work with. (I think I only discovered that in recent months myself)
Identify has keyboard shortcuts. Click the keyboard icon there. X for Casual …
A for Agree, but careful with various IDs, which one you have agreed to.
To the original question.
I would like to be rewarded with better / more efficient use of iNat behind a soft barrier.
An active example - permission to change taxon pictures.
Looking at iNaturalist in the long term, having hard to ID observations just sit there isn’t that bad.It’s more important to have observations in the database. Cameras and computer vision keep improving. For example, iNaturalist is now a lot better at identifying ferns than it was four years ago.
Those mosses and lichens might be more identifiable in the future. Having the observations there to look at is valuable.
One way to improve the system of identifying observations would be to prod university professors to do better using iNaturalist as a teaching tool. The students in those classes could make better observations, and also start to identify some observations. Those students could be a key iNaturalist demographic going forward as the continue to work in natural sciences.
3 Likes
Thanks for the tips.
I feel like iNat should be sending an email out to anyone showing interest in identifying with this info.
For me the reward is learning. The website has been so useful in boosting my confidence enough for me to take on an Ecology degree and I was able to cite my iNat work in my application. It’s improved my identifying skills, my observation skills and things like photography and record keeping too. You have to be a bit proactive in seeking actual rewards but I’ve been able to secure a recorder’s grant for equipment thanks to iNat, and it’s helped me secure placements and give me some credibility with certain organisations.
17 Likes
Wow! You are an iNat success story.
I wish you a wonderful future ahead in Ecology.
2 Likes
As a long-time arthropod identifier, the main thing that discourages me at the moment is that I feel like I am actively working against the AI-suggested identifications all the time. Presumably the AI will improve over time and the problem is frequently that some observers are choosing a much more specific ID than is possible from the list of suggested ones. But a lot of what I do is downgrading ridiculously specific IDs from blurry photos of bugs that can barely be identified to order. I greatly wish that a stronger disclaimer could be placed on the suggested IDs from the AI, at least for some taxa.
16 Likes