Remove "Yes" votes from the "Can the ID be improved" DQA when the Community Taxon improves

This has been discussed in several places, but not very thoroughly. Basically, the problem is that when someone votes “Yes” in the aforementioned DQA question, the observation stays perpetually Needs ID until someone counters it with a “No” vote. The simple (from my perspective, anyway) solution would be to automatically remove the vote when a new ID is added. It could only occur when the new ID improves the Community Taxon, which probably makes the most sense, or for all IDs.

Edit: it seems that it would make the most sense to only remove votes when the Community Taxon improves, so I have changed the title to specify that.

To anyone who doesn’t know what the DQA is, it’s the Data Quality Assessment and it looks like this. The question I’m referring to is at the bottom:

I approved this, but will note that something very similar was suggested in this topic: https://forum.inaturalist.org/t/reset-can-the-community-taxon-still-be-confirmed-or-improved-after-taxon-swap/9266

2 Likes

I would specifically be in favor of a change to Community Taxon. If I add a Maverick ID and check the ‘Yes’ box, then someone else agreeing with my ID would move the C.T. up at least a level.

3 Likes

I don’t want to it be that way, if I mark it to get a ssp. id and all next ones are still at species id, why would I need to “yes” vote be gone? Or next id is not from user I believe is knowledgable, it would require revote it. It’d be more rational to get notifications about such observations rather that automatically undo your action.

4 Likes

Since the DQA here is “can the Community Taxon still be confirmed or improved” (my emphasis), I would suggest that votes only need to be removed by the system if and when the Community Taxon changes. And then, both yes and no votes should be removed, since all of those votes were based on the Community Taxon at the time the votes were cast, and not on the newly changed Community Taxon. I would vote for something that functions this way.

EDIT: although, I did express this concern about such functionality a while back…

3 Likes

Yeah, I think this is a valid concern. It could also happen if there’s a legitimate back and forth between IDers and the CI gets changed one or more times. It could revert to the original ID that the DQA was for (and then not have it). I think this would be pretty uncommon, but worth considering.

1 Like

then we can be prompted to remove our vote - if we agree to.

1 Like

I would not vote for this if it removed the “no” votes. I typically only use “no” if it is a record without sufficient evidence to move it to a finer taxon. Automatically removing that vote when someone adds an “improving” ID defeats what seems like the main functionality for the “no” option.

4 Likes

I definitely see your point here. As certain as one might be at the time, though, it is still one vote based on the information available at the time. If an expert comes along and sees something that I missed the first time, I think it would be appropriate for me to re-evaluate my no vote in that light, plus any other information that might have been added to the observation in the interim. I can always re-vote no if needed.

That said, the pending overhaul of the notifications system may well address the issues in this request, in particular

3 Likes

Yep, I have no problem re-evaluating the observation if someone else chimes in. I just don’t like the idea of the “no” vote automatically being removed with no notification. Sounds like the new updates would take care of the notification issue, although I would still rather the onus be on me to re-evaluate the record upon feedback, rather than having this happen automatically. I think it would be quite uncommon for me to remove a “no” based on others’ comments/feedback, as I generally try to only use it for cases where I am certain there is not enough information to refine the identification (and not just when I don’t have sufficient knowledge to do so).

3 Likes

I would argue that that whole question should be removed.

Last night I was among people who were discussing the need to split certain species of fungi into multiple species due to DNA evidence, with material from 2 specimens, both of which would previously have been treated as a single species, being genetically too different from other each other, to continue to treat both specimens, as a single species. For some of these species no one had yet determined ways to physically distinguish the 2 species. After becoming sufficiently familiar with specimens with each DNA code, someone might come along and find ways that they physically differ, allowing the 2 new species to be distinguished without a new DNA analysis for a new specimen.

Much the same is true when someone writes species descriptions, writes a key to distinguish any 2 species, writes a book, or webpages for 2 species. Beyond those ways offered in the descriptions, key, book, or webpage, those writing them, had not offered, or had not yet found, further ways to distinguish the 2 species. It is always a possibility that someone, who becomes especially familiar with the 2 species, to have learned to distinguish them, without observing those distinguishing features offered in the key, or any description of the 2 species, or any published reference material on the 2 species. I don’t know how one person can say it would be impossible for second person to distinguish a species, because the first person hasn’t found any of the distinguishing features in that observation, that the first person knows from the literature.

1 Like

I think more specifically, “yes” and “no” would be best to become removed when CT improves and is species or subspecies (only).

Say there’s a family ID and an experienced identifier marks “no,” and then an inexperienced identifier IDs genus but is uncertain or incorrect, and a second user agrees. In that case it would be best if the original “no” remained. (or in some scenario like this in case it doesn’t fit, where there’s an uncertain or incorrect “improving” CT).

The idea to remove yes and no when CT improves and is species or subspecies (only) seems to get around these limitations. Another thing to point out is we’re generally suggesting (in any of these versions) CT (IDs) should affect DQA (subtractively), vs. DQA only affects CT in one direction currently. Which also raises the question of whether CT should affect DQA additively in any circumstances (automatically marking a “yes” or “no”).

1 Like

Your examples talk about cases where no one yet knows how to visually distinguish species but may later. But “yes” and “no,” like making species IDs, can’t always have perfect certainty, especially of what may be known later, so possible uncertainty or later revision doesn’t contradict the justification to use them. I think of IDs and DQA like an adaptive process which can undergo revision in response to changes or feedback over time. Including a user may change their own mind or refine ID without anyone else’s feedback. But also, many users read/respond to all notifications, and people can comment about DQA being wrong if they want to get user attention to change it. Finally, users can vote an opposing DQA answer to steer it in the opposite direction or cancel an existing one.

This also raises an idea, that I’d prefer changing DQA votes also create notifications, although doing so may create too many so it would be best only for disagreeing DQA votes.

1 Like

Only in my first example, of species newly split by DNA evidence, did no one know how to visually distinguish 2 species. Then I talk about how descriptions, keys, books, and webpages, always have had limits to what distinguishing features the author either published, or even knew about. Those descriptions, keys, books, and webpages have never been the final word on how to distinguish one species from another. The more you get to know any given species, the more you are likely to accurately recognize it without seeing the distinguishing features published in the reference materials. You may have learned about, or found, additional distinguishing features. You might even be able to regularly, and accurately distinguish a species without being able to describe just what you are seeing that allows you to distinguish it. I regularly see experts in a given taxon group accurately identify species, and I can’t see how they did it from what they saw, even if I had carefully memorized all of the distinguishing features for that species in the reference materials.

You also talk about what “is known”, as if it there is one universal brain that knows, or doesn’t know things. There have always been things things that are known by a limited number of people, that may not be published, and thus might be said to be “not known by science”.

When someone says that an observation can’t be distinguished any better than say genus level from the evidence in the observation, they are really saying that they don’t know how to distinguish it to say species level from the evidence in the observation. They can’t know that no one else can’t distinguish it to species level. Someone else may have found a way to recognize the species, that the person questioning their ability to get to species didn’t know about, which could potentially include distinguishing features that are not published, but that the person identifying to species learned about, or figured out, or in the case of gestalt, they somehow accurately recognized the species, but couldn’t explain what allowed them to go to species.

3 Likes

Most people don’t use DQA this way, or shouldn’t. It’s more usefully (and maybe even supposed to be) used when someone thinks ID can or can’t be refined by anyone, not only themselves. If someone marks Yes, that itself indicates they think someone else can refine it though they can’t, otherwise they would.

I’ve actually been in agreement that users sometimes mark DQA incorrectly, or only small groups of individuals might know how to refine ID, or general state of knowledge or new methods may be developed later on which allow ID to be refined. The only difference is I (like most I think) still find DQQ useful. It’s an uncommon argument to suggest removing DQA, so it’s not surprising others would disagree. I also did suggest a helpful modification, for iNat to automatically remove “yes” and “no” when Community ID improves and is species or subspecies (only).

In supporting using DQA, by analogy to making IDs the possibility of uncertainty or imperfection doesn’t outweigh ways in which it’s useful and justified overall. Like for IDs, users can also refine DQA over time in response to new info., or vote against each others’ DQA votes. This does a lot to address error or uncertainty concerns you raise.

3 Likes