Track and mark identifications made through "Agree?" buttons

#1

A fair number of observers treat the “Agree” button as if it meant “Thanks for adding an ID” instead of “I’ve independently come to the same conclusion.”

I propose that iNat start tracking which identifications are created by clicking the “Agree” button (if they aren’t already), and marking them on the user interface. For example, there is now a unique icon for identifications made with the assistance of the machine learning system. Another, different icon for users who simply click the Agree button seems like an obvious, easy improvement.

In particular, the IDs of a user who put a basic ID on their own observation, then immediately agreed with every new ID which other users added, should definitely be marked as suspect. Maybe require species-level IDs from 2 non-observer users on observations where the observer initially provided a higher-level ID before removing from the needs ID list? Brainstorming better done in a topic in General

0 Likes

Create Limits to the Power to Convert "Needs ID" obs to "Research Grade"
#2

I was guilty of precisely what you’ve described for a long while when I started using the site, not because of any malicious intent but because I misunderstood the function and only used the app for a long time, and I think your idea is a good one. Additionally, giving such users a head’s up or reminder on perhaps their first 25-50 obs (pop-up that asks if you actually can independently agree?) would curb that behavior for most well-meaning, new users.

Maybe?

1 Like

#3

well this links to the disappearing ID problem where I may start doing that again because otherwise, if the user who adds the ID leaves and deletes the IDs, i lose my ID.

2 Likes

#4

Here’s a recent example of an observation where blind agreement led to a RG observation with an incorrect ID (although it all turned out alright in the end): https://www.inaturalist.org/observations/19973888

Edit: you can see full timestamps by hovering over the “how long ago was this?” indicator (e.g. “1mo”) in the upper-right corner of the IDs.

1 Like

#5

@mira_l_b, would you describe what you used to think the function of the Agree button was? It’s sometimes difficult to see things from the point of view of someone who doesn’t know what I know - it all seems obvious when I already know how it’s supposed to work, so I’m not confident my guess about what’s going on is accurate. A first-person account might lead to better suggestions for how to handle this.

1 Like

#6

I’m happy to, @JeremyHussell… Much along the lines of what you suggested:

Although, in my case it wasn’t just_ that. It was partially an acknowledgement of the help of an IDer but I would always click on the taxa and go to its page and see if the ID made sense. Then, if I felt that it did or I couldn’t find a reason to disagree I would “agree.” It was nuanced a bit, since I understood that the “agree” function wasn’t supposed to be total blind agreement, so again, not just a, “thanks, that’s cool!” rather more like an, “okay, that seems reasonable enough…and thanks!” A little bit of a gray area, I guess. The problem with what I was doing, as I realized later after reading all sorts of comments in the google group and revisiting the help or starting guides pages, is that while it mostly was fine that I just agreed based on publicly available knowledge, as we all know there are tons of cases where a genus contains hundreds or even just dozens of species, some only able to be IDed through dissection, and I couldn’t justify those agreements, but didn’t know I couldn’t. The problem of not knowing what you don’t know. I now err on the side of extra cautious.

The other element, beyond just believing I was demonstrating some good-natured gratitude or something, was that being unfamiliar with the site my instincts told me it would be rude not to acknowledge some expert taking the time to ID my obs. In the beginning I didn’t realize anyone could ID anyone else’s obs and thought that all the IDs were from folks with higher competency than myself. (although I’m blown away by how much access I’ve had to specialists and their expertise!!!)

I mentioned that my use of inat was completely restricted to the app because it was meeting my need and I had no idea it was full of so much user potential (coming from using ebird as my point of reference which is much more one-sided and not at all interactive in the amazing way inat is) so another factor was not looking into who was making the IDs for me in the first place. Once I began to use the website, I re-read all the guidance and was embarrassed at my misunderstandings or rush to observe without proper self-education. I was then able to understand much more about how this all works. What seemed straightforward in the app became so much more complex (but in a good way from my view) once I visited the website. I still consider whether I should go back and withdraw agreements on earlier IDs but I’ve not done that yet, thinking that it would not only be a lot of work, but that nothing I’ve observed that I agreed on (possibly) erroneously is that significant and doing so would require a lot of tedious explanation to other users in comments. I may do this in the future, still considering it.

I’ll leave my comments there because as I’ve owned up to elsewhere, I’m super verbose. Please let me know if anything requires more explanation or follow-up, as I’m happy to offer my perspective or suggestions if helpful.

Thanks, Mira

5 Likes

#7

Here’s what my experience as a new user was like in 2014: I always used the website. I knew some organisms fairly well, some less well, and some not at all. I was fortunate to have a few very helpful people following me, and I never had occasion (that I remember) to think that any of their IDs were incorrect, unless they started a debate among themselves. But I agreed to all the rest of their IDs that were offered. My agreements would have meant (1) that I had independently come to the same conclusion but hadn’t had time yet to put the ID on because they were so quick with theirs, or (2) that I assumed they were right and was happy to go with their ID. No one suggested that I should not be doing that, and I hadn’t seen anything on the iNaturalist.org site suggesting that it was not the right way. I figured it out by reading the google group later, which I didn’t know existed from the iNaturalist.org site, but finally found by googling my questions about how to do various things. Up to that point I had the impression that getting your obs to Research Grade asap was the goal. I’m not sure the message that it isn’t is clear enough, or prominent enough.

5 Likes

#8

I think nearly every active new user on iNaturalist does this, which suggests to me either that it is somehow a natural response or that the system somehow encourages it.
As others mentioned, I think it’s a combination of acknowledging the ID (and assuming the identifier is an expert), wanting to get RG, and looking at the species photo and thinking “that looks about right”. Yeah?

However it happens, I am confident that it has resulted in enough observations of difficult-to-ID species becoming Research Grade that they get the species into the machine learning suggestions, which people then start using to identify their own observations.

4 Likes

#9

I wonder if some “trust levels” would work? Such as no ability to agree with an ID on anyone’s observation (including one’s own) without some kind of history of being the first to ID something.

2 Likes

#10

That’s been discussed. I’m sure it will show up as a feature request here at some point.

1 Like

#11

another interesting “agree behaviour” is when there are a number of refinements, and someone comes in and agrees to each one one after the other. First they agree to family, then to genus, then to species. When you point out to them that you can only make one ID, and that by agreeing to the species, they are overiding their genus level agreement, and that they should only agree with the one ID that they do in fact fully agree with, they change that behaviour.

There is no weighting system in iNat. Conservative and knowledgable identifiers have the same weighting as the school student who likes to call his mate a baboon. This is easily rectified if we get to know who the quality identifiers are, and add our weight to their ID. In other words, I am not neccesarily agreeing with the ID, but agreeing that another identifier is very likely to know what they are talking about.

The other factor involved, is how active is the dentifier? Someone who has just joined is likely to add agreement out of a misunderstanding of what that means, and then to further compound the problem, they become inactive and are not around to change their ID following a discussion and further information coming to light.

I think identifications from new users are hugely problematic. We don’t want to stop them outright, because that will take away a draw card to getting them to be more active. I know it led me to learning a huge amount (and a greater appreciation of how little I know), which probably would not have happened if I had not been able to have an influence on an ID from the outset.

3 Likes

#12

This may be tangential, but…

I didn’t realize I could ID things to genus or above in the beginning, thinking I had to know exact species or nothing at all, which fed into uploading unknowns or relying on AI (which I also misunderstood thinking it was to be relied upon as an expert [I discussed this in one of my early gg posts]) and fed into my eagerness to agree so I could show I was participating.

3 Likes

#13

I feel like a small number of pop-up text boxes (as you said in your first reply on this thread) that point out important details and common issues, like many websites have upon signing up, would help minimize a bunch of these issues.

6 Likes

#14

There is a third reason for a “blind” Agreement.
As a user from iSpot with a reputation system, when we moved our data from iSpot to iNat a lot of our identifications changes because of the absence of a reputation system on iNat.
So for instance, taking the genus Drimia that only three people on earth can really identify accurately (no field guides, difficult characters, etc.), on iSpot two of these people (the third did not join the site, but is now on iNaturalist) did most of the IDs and their Expert reputation sealed the ID.
This did not work when the data were transferred to iNaturalist, and so to ensure that the correct ID was the “Community ID”, we had a team agreeing to our expert’s IDs just to ensure that the most likely ID (by far) was the community ID.
Alas, it has stuck. When I see an ID by a taxonomic expert or by a high reputation amateur I simply agree. On iSpot I would never have done this because of my reputation. But on iNat I dont have a reputation so it does not matter that I agree without going to the monographs and keys. But if Jonathan Collwell IDs a Monkey Beetle or John Manning an Irid, I agree automatically because they are the world experts in those groups. If they are wrong, who on earth would know?

I agree that treating agreements differently from identifications is a cool ID. But then we need some additionally functionality. On iSpot that did agreements differently, one could not post an ID that was already on the observation: you had to agree to it. On iNat you can agree to a specific ID, to a community ID, or post an independent copy of an already existing ID (indeed, you can post a second copy of an ID that you made previously, which is weird, but handy if someone disagrees with you and you want to make a point - although I dont think that this was the intention).
Another difference from iSpot is that IDs and Agreements could not be altered. Once made, the only way to change them was to post a revised ID. That prevents people deleting mistakes (and any agreements made to it), and altering subsequent embarrassing explanations - which iNat allows. Users who deleted their accounts had their IDs reassigned as posted by Anonxyz (where xyz was a link to the user known only to site managers), and their reputation deprecated to novice.
I am not saying that iSpot was a good system, but include it here just for comparison and to explore options.

3 Likes

#16

I was about to vote for this Feature Request, until I read the responses, thought some more, and realized its not that simple.

First, I see nothing particularly wrong with using the “Agree” button. I use it all the time when I’m doing IDs - if I’m convinced that the leading ID is correct, it’s a lot faster than typing in an independent ID that says the same thing. So I’m not sure anything is accomplished by distinguishing it from using “Suggest an ID.” It’s often a useful shortcut.

In the specific case of an observer agreeing (perhaps blindly) with a lower-level ID, I agree that that’s often a problem, and essentially brings an observation to RG status with only one real ID. But there are also cases where an observer might agree for good reason, having actually thought about it and decided that another ID is correct. There is no way to distinguish these cases (without having a conversation with the observer), so penalizing the “Agree” button would penalize both thoughtful and thoughtless agreements.

The third issue in the proposal is the suggestion to

There are also legitimate reasons for an observer to initially provide a high-level ID (see this discussion: https://forum.inaturalist.org/t/why-do-some-serious-power-users-add-so-many-unknown-observations/282/29). In my own experience with the bulk-upload interface, I’ve found it easier to let things upload with default IDs (sometimes blank, sometimes erroneous), then go back and add add/correct the IDs. I don’t see any reason to require additonal IDs for this case, and I can’t see a way to distinguish this case from the problematic case, unless we force all IDers to add a comment explaining their reasoning.

4 Likes

#17

I also often use the higher level ID, when I want to get to the “Compare” button. Sometimes it is purely to because I have a blank (sorry, old age!) as to the species or genus name I am looking for, but I know the taxon at the fine level well.

3 Likes

#18

This might seem tangential to the topic, but I interpret the topic to be how do we avoid or find ‘unsafe’ research grade observations. One thing that would be useful is if we could search/filter by the number of identifications that a research grade observation has - if it has only two (or less ;-) identifications we might want to double check. (Personally I would find searching for zero identifications and/or comments would be useful for dealing to the ‘Unknown’ pile of observations too)

–Tony

3 Likes

#19

I do this fairly often as well. Sometimes I do it because it is faster than consulting other references to refresh my memory which species in a genus (e.g.) are found in a particular area.

0 Likes

#20

Since this request has been sitting untouched for a while, and I’m the only one who voted for it, I figured I should edit it for clarity, in part prompted by a post where I explained the difference between agreements and independent identifications in a way that people seem to like.

0 Likes

#21

Sorry if I missed it somewhere else in the discussion, but I’m not seeing how this is going to be very helpful if it comes with no information about the motivation behind the Agree. I use Agree the same way as

My ID rate would slow down a lot if I felt the need to avoid the “stigma” of using Agree.

1 Like