Change wording used by the system when downgrading an observation to an higher level taxa

Thanks everyone for the feedback.

Branch Disagreements
Regarding ‘branch disagreements’, upupa-epops’s use case Let’s say someone posted a very blurry picture of a ladybug... is the one we had in mind. While we agree it should be rare and backed up by the community and comments, the idea is that there should be some mechanism for the community to role back the taxon that an observation is ‘filed under’ if the observer is not responsive and if the community consensus is that while it could be Seven-spotted Ladybug its really impossible to tell and we should err on the side of caution.

Regarding @Pisum and other’s concern that we’re lowering the bar with branch disagreements, eg: note that with the proposed new prompt, you’ll no longer need to know that there’s no way anyone could confirm / deny, you’ll just need to think that there’s no way anyone could confirm or deny. that’s a slightly lower bar for disagreement. those who ignored the letter of the current prompt to disagree would now have the new prompt wording on their side. the idea was actually to raise the bar (ie make it rarer for people to ‘branch disagree’). The concern with the existing ‘evidence’ based language is that if @charlie observed a plant and ID’d it to species because he’s an ‘expert’ but his photo didn’t capture all the characters necessary to identify it to species, someone else shouldn’t roll it back to genus merely because the ‘evidence was not sufficient’. The ‘I don’t think we can be certain…’ language was meant to capture the spirit that the community doesn’t think anyone is intentionally/knowingly advocating for that species which would eliminate the charlie situation (we can be reasonably certain charlie knows what he’s doing) but would cover for example, someone blindly clicking on a computer vision suggestion (in most of these cases the community can ascertain that the precision associated with such an ID wasn’t intentional).

Again, the idea is that the ‘branch disagreement’ lever should be very rarely pulled, so we agree with concerns about this lever being over-used. But we think the use case articulated by upupa-epops is legitimate, especially if the identification in question makes the observation sufficiently unusual/unlikely that it decreases the credibility of the site (even though it ‘could’ be that taxon) and the observer is unresponsive. But if folks think this is too niche a use case for a tool that could be broadly misused, thats fair.

Leading Disagreements:
Regarding ‘leading disagreements’, @Pisum and @tchakamaura brought up a legitimate issue with our plans we didn’t think about. First some terminology. In the following example User B’s ‘Identification taxon’ is Epilachna borealis.


Because the Identification Taxon is a sibling to and not an ancestor of the Community Taxon (C. septempuctata), its very easy to recognize the not only the ancestry to the Identification Taxon (green) which User B is certain the observation is, but also ‘Disagreement Branch’ (orange) which User B is certain the Taxon isn’t. It extends from the Community Taxon back to the common ancestor with User B’s Identification Taxon.

But in the case of ‘ancestor disagreements’, e.g.


Psium’s is correct that sometimes we might want to set the Identification Taxon in one place (e.g. ‘I’m certain this is Family Coccinellidae’) but the root of the disagreement branch somewhere else (e.g. ‘I’m certain this is not in the genus Coccinella, but it could be some other genus in the tribe…’)

Psium is correct that under the planned changes, this can’t be articulated. Choosing ‘Leading Disagreement’ will set the Identification Taxon at Family Coccinellidae but set the ‘disagreement branch’ at C. septempuctata which isn’t whats intended.
Similarly, choosing ‘Branch Disagreement’ (which isn’t the could-be-but-can’t-be-certain use case we’re intending by that choice) also wouldn’t do what’s intended as it would define the disagreement branch as everything below the Identification Taxon

Wrapped up in this is that since identifiers can only trigger disagreements by suggesting a taxon that is an ancestor of the Community Taxon (not the Observation Taxon), if User B did a ‘leading disagreement’ the Community Taxon would be at Family Coccinellidae but the Observation Taxon would be at Genus Coccinella and there’d be no way for future identifiers to trigger a disagreement with Genus Coccinella.

Do we need functionality that allows the flexibility for an identifier to:
a) disagree with taxa other than the Community Taxon
b) explicitly choose not just their Identification Taxon but also the root of the Disagreement Branch?

2 Likes

I’m not sure it’s my place to say we need these, but I have certainly wanted them and would use them.

1 Like

Yes, I would love to be able to pick at what level I disagree. Allowing to disagree with something other than the community taxon would be useful in situation like this: https://www.inaturalist.org/observations/17244684

Here, I am not certain of the species but know the genus, but since someone has already made an ID to species it feels like I am disagreeing when I add my ID even though I am not. And the result is I often skip adding IDs in such situations, even though it means these observations get stuck in State of Matter. If there was an option to say I explicitly disagreed with Pediastrum duplex, then by not using it, it would be obvious that I wasn’t disagreeing with it and all would be good. And yes, I can just add a note stating that, but it’s not as quite as good as a built-in solution.

2 Likes

re: branch disagreements, if

then if you reword the new choice 2 as
“Yes, we can’t be certain beyond [ID node]” instead of
“Yes, I don’t think we can be certain beyond [ID node]”
then i think that choice will be equivalent to the current prompt.
and i think i would be clearer to note the disagreement as something along the lines of “[Disagreer] disagrees because it’s not possible to make an ID finer than [ID node]” or “… because we can’t be certain beyond [ID node]”.

… and if you remove branch disagreement altogether, then, of course, that really raises the bar.

re: leading disagreements
i think you’re making it too complicated. all you need to do is make observation taxon = community taxon. in your blog, looking at the Leading Disagreement diagram, however the community taxon is calculated to get Insecta after B’s ID, do the same thing for the observation taxon. that’s it. the rest of the diagram looks fine.

if there’s some technical reason the observation taxon needs to differ from the community taxon for leading disagreements, then i think instead of doing whatever you proposed to do, do this instead:
offer a branch disagreement that gets automatically withdrawn after a finer (or equivalent) disagreement to the original ID is made. that’s it. (in some ways, something like that would also address my latest note about the usefulness of the proposed leading disagreement functionality.)

if you’re thinking of opening up a whole can of worms by letting people select disagreement levels for a potential leading disagreements implementation, then i think the better can of worms to open up is to allow people to pick the specific existing (distinct) IDs they disagree with. no need to learn taxonomy with this kind of approach. just look at what others have suggested and knock down the (distinct) one(s) you disagree with (after making your own ID, of course).

@loarie Yes this covers it. Allowing ppl to choose the place of disagreement is what the pisum example needs to be accurate, though it’s not necessary in that the user could research which other kind of legume it is, but that would be hard. I like the separation of disagreement and agreement as it allows the user to be judicious and refrain support for taxa they’re not sure of and still get the result of disgareing with the tip.
I think this will cover a use case I mentioned in another thread,where it is not possible to add an agreeing id (to the new cid and without disgareement with some specific IDs compatible with th cid which are already present) if your I’d will be one of those to tip an observation from one branch of the tree of life to another.
https://forum.inaturalist.org/t/offer-the-ability-to-choose-whether-to-hard-disagree-or-not-to-the-next-finest-level-that-the-current-identifications-support-when-one-disagrees-with-the-current-community-id/3921/7

I think the separation will accomplish what @pisum wants, in allowing people to choose what they agree and disgaree with. The prompt might include an interactive phylogeny, including all current IDs, as is provided in the example txt here, and this will remove the concern where people have to know the taxonomy already. The current status quo on obs and IDs is to know the taxonomy and I’d, or get the computer / computer vision to provide it by messing with the compare and suggestions tabs and filters.

1 Like

@pisum’s comments about leading disagreements make a lot of sense:

i think you’re making it too complicated. all you need to do is make observation taxon = community taxon.

We changed how the Observation Taxon works so that in most cases it matches the Community Taxon, the exceptions are in an update to the blog post that I copied below.

While the idea of making it so people can arbitrarily set the branch they are disagreeing with and also disagree with any identification is interesting and certainly something we could do down the road, it does introduce a lot of complexity. I think pisum is right that with this tweak to the Observation Taxon, our existing simpler plans for dealing with ancestor disagreement problems will be a good first step. And we should hold off on the ‘can of worms’ for now

In contrast, the Observation Taxon will always match the Community Taxon unless:
a) there is just a single identification, then the Observation Taxon will be defined by that identification
b) the observer opts out of the Community Taxon, then the Observation Taxon will be defined by the observers identification
c) there are no disagreements and there is an identifications of descendants of the Community Taxon, then the Observation Taxon will be defined by the finest such identification (because the community likes that a single non-controversial identification being able to ‘move the ball forward’)*

*if that finest identification is of infra-species rank (eg subspecies), the Observation Taxon won’t roll forward to that rank from the Community Taxon if that identification was added later (because the community doesn’t like what would be Research Grade observations at species rank being rolled forward to Needs ID observations at infra-species rank). However, if the Observation Taxon was initially set at infra-species rank from a single identification, a non-disagreeing identification of an ancestor won’t roll the Observation Taxon back to the Community Taxon.

1 Like

given the new wording, does the leading disagreement diagram from the blog post still make sense?

tbd_leaddisagree

1 Like

wow great catch - fixed. Thanks for all your help with this

2 Likes

you’re welcome. i think we’re in agreement now on how it makes sense for leading disagreement to work.

What happened to branch disagree?

re: branch disagreements, the only change from the original proposal in the blog post that we’re planning is changing the wording to

“Yes, we can’t be certain beyond [ID node]” instead of
“Yes, I don’t think we can be certain beyond [ID node]”

2 Likes

“Experts” are part of the community and have as much right as anyone to decide that the photo can’t be ID’d.

I never said they weren’t part of the community (or any less a part of it), nor did I say they had no right to decide that THEY can’t ID it.

I DO contest whether they have the right to say someone else can’t. Very subtle difference.

2 Likes

This post shows the different interpretations of this quite well:

https://forum.inaturalist.org/t/identification-etiquette-on-inaturalist-wiki/1503/68

@charlie points out that there are gestalt cues that an IDer can rely on, and @bouteloua does add that it should be done with caution (my interpretation of their exchange there)

My position is that we should be identifying purely to how we see ourselves, ie we should only be explicitly disagreeing if we can see evidence that it is not that species. If you think there is the possibility of that species, then it is a lie to say that it is definitely not that species, so you should not be explicitly disagreeing. What you should be doing is challenging the ID of the other observer through dialogue. Of course, this is just my position, which I believe fits with the concept and intent of community ID.

The problem is that a lot of observers/identifiers are absentee, and not around to participate in the dialogue and/or change their IDs when shown that it could be other than what they say. Further, many of these absentee identifiers are novice and relying on a CV suggestion that can be very worthy of challenging! If clearly a novice absentee or doesn’t respond to the challenge on the ID, then in that case I agree with Cassi that we should perhaps bump the ID with an explicit disagreement, but only if we ourselves will be around to withdraw that if further information is supplied that would change that position!

The main problem with this strategy, which has been mentioned elsewhere extensively, is that if everyone follows your protocol then there is no mechanism for the community to halt over-identifying observations (records reaching research grade when they can’t actually be confirmed to species/genus). The only way your strategy could be functional without allowing these over-identifications (which could also be thought of as false-positives) is if every user was careful to only add identifications when they are certain, responded to dialogue, and bumped their ID back up a taxonomic level in light of that discussion when warranted. We’re very far away from that being the case, and in my experience these over-identifications are not changed by the users in the vast majority of situations even with attempts at dialogue.

There probably are cases where these IDs are truly correct (that is, the organism was actually Species A despite insufficient evidence to confirm such in the media provided), but I would suggest that the majority of these over-identification cases are situations where the observer and identifiers are unaware of how to distinguish similar species and have just picked one. Allowing those latter records to remain research grade undermines the concept of community ID and reduces the overall quality of iNat data.

6 Likes

… which is where initiating dialogue can help address this fundamental flaw in how people are using the site. Simply “fixing” the observations CID and not addressing how it got that way is not stopping it from happening over and over again.

Teach them to fish… :)

3 Likes

I have no problem with initiating dialog (and often do) to try to fix that flaw in the behavior. But fixing the observation fixes the problem, whereas the dialog alone doesn’t in most cases so I opt to do both simultaneously.

If the observer/identifiers never respond, there’s no simple way to filter back through for those observations at a later date to fix the ID (and having to do that would take double the time). I’ll just keep fixing the IDs while simultaneously trying to initiate dialog, and am happy to correspond with the <5% who ever respond. For the other 95%, at least the data is improved even if there is no behavior change on the users’ part.

8 Likes

I think like many things there’s a gradient. Blurry photo of nondescript shrub in Virginia identified as a plant endemic to California deserts? Explicit disagree. Observation of a somewhat cryptic grass within geographic range by an established user without easy diagnostic features visible? Just a comment or a mark reviewed. And yes there are lots of grey areas. I tend not to explicitly disagree unless it cleans up the range maps. I also sometimes mark as needing another id when I comment

6 Likes

Coming to this thread late so I don’t know what, if anything has been amended since the start.
The present wording of the green and orange buttons is really confusing and must PLEASE be changed.

I made a mistake of my own recently picking the wrong button which left a bad ID in place, and now a well-meaning person has coarsened the IDs of some of my obs because he intended to agree with a higher taxa, not disagree with my ID.

@vynbos suggested clearer wording.

3 Likes

Agree, or at least maybe tell people what each one really does before they choose it: Explicit disagree or Partial disagree.

Possibly have an “undo” for a few minutes after in case you touched or clicked the wrong one accidentally.