The DQA “Based on the evidence, can the Community Taxon still be confirmed or improved?” / “No, it’s as good as it can be” is able to push an observation to limbo when it should not. This DQA should be disabled in some cases.
An observation of a Viola (Plantae) got these IDs and is now “Casual”, identified as “Life”:
Do you think “No, it’s as good as it can be” is consistent with a Pre-Maverick observation (one more ID “Viola” and the ID “Viola sororia” becomes a Maverick) identified as Life with only IDs at rank species and genus?
I don’t think this is a case for disabling the DQA (some things can’t get further than “Life”). It’s a case for countering the DQA vote (and perhaps tagging some other identifiers to help you with that).
We should avoid extra useless work, as far as possible.
We should avoid pushing an observation to limbo, as far as possible.
This observation should not become “Casual” (= limbo) because it does not deserve it and because this can be avoided by automatically disabling the DQA.
And identifiers adding an homonym ID need a special extra warning (like a confirmation popup).
Many people don’t understand the point of the checkbox “It’s as good as it can be”, and based on this misunderstanding they will often check this box when they do not intend to say that the Community ID cannot be improved and the observation should go to RG (if CID below family level) or to Casual (if CID at family level or above).
Consider two scenarios:
A very blurry photo of a passerine bird that’s too indistinct to narrow down to family. Maybe the observer identified it as a Hermit Thrush, the next identifier added a disagreeing ID as Passeriformes, and maybe there have been other Passeriformes IDs. Selecting “It’s as good as it can be” provides a way to move this to Casual with a CID of Passeriformes. This seems like a genuine case for using “It’s as good as it can be” on a observation with a high-level CID.
Jean Philippe’s Viola vs Viola Skipper example, where at least one identifier misunderstood the subject of the observation. (This could also just be genuine disagreement of the “plant vs. fungus” or “moth pupa vs. gall” variety.) In this case conflicting low-level IDs from widely separated branches of the tree of life have resulted in a current high-level CID.
So, do we have a heuristic that will distinguish these two scenarios? I would say that the difference is that in scenario 1, the CID is supported by multiple IDs at that same high level. In scenario 2, the CID reflects the fact that the ongoing ID process hasn’t yet reached equilibrium. Can we find some logic that distinguishes these?
That isn’t misunderstanding the subject. The identifier clearly intended to identify it as a plant (les papillons n’ont pas de fleurs), but typed “Viola” and picked the butterfly by mistake. Misunderstanding the subject would be, for example, identifying Asteraceae instead of Misumena in a picture of a spider on a flower head.
I may have made a wrong assumption from the image that was shared, but the DQA would seem inappropriate in either scenario (when the observer mistypes the taxon name; or when an identifier misunderstands the intended focus). In both scenarios, the observation would have a high-level CID because of a currently unresolved disagreement between low-level IDs and offering the DQA option “It’s as good as it can be” seems unhelpful.
Neither of these scenarios seem to apply here. The first identifier (the observer?) identified this as a plant. As @phma pointed out, the second identifier also understood that this was a plant, but misclicked on the insect taxon (thanks to the scientific community allowing the same generic name to be assigned to plants and animals). Based on the comment added by the second identifer, I presume that person also clicked the DQA, intending it mean “this plant ID cannot be improved beyond genus.” A third identifier also selected the insect, probably by clicking the Agree button while failing to notice the kingdom disagreement.
If neither of these identifiers can be persuaded to withdraw their IDs, we now need four more identifiers to select the plant genus to make that the CID. At that point, if the DQA box is still checked, it would make that genus go to RG. Alternately, one vote against the DQA would remove it from the Casual rank, immediately.
Fine. My reconstruction of the ID sequence was clearly wrong. But actually, that just reinforces @jeanphilippeb’s original point. In none of these scenarios does it make sense to offer the DQA box to identifiers. Nothing is improved by voting for or against DQA for observations of this sort with conflicting low-level IDs (no matter which of the sequences of conflicting low-level IDs happened in a particular observation). The DQA option is only helpful when the CID reflects multiple identifiers adding IDs above species.
So what would be good logic?
My suggestion is to hide the DQA option unless the CID is supported by more than two thirds of identifications at that same level. So one Viola plant ID and two Viola Skipper IDs would have a CID of Life and no DQA option. Three Passeriformes IDs and one Hermit Thrush ID would have a CID of Passerifomes and allow users to vote “No, it’s as good as it can be”. (Two Passeriformes IDs alone would also allow “No, it’s as good as it can be”.)
That logic would seem to allow people to move observations to RG/Casual when needed, and avoid offering a confusing option in circumstances where the effect is likely to be misunderstood.
but. Identifiers have to remember to check if the problem lies with a (wrong) DQA. Meanwhile the obs languishes in limbo. It the observer was a newbie they abandon iNat in disgust - I tried it, no reaction, waste of time!
Doubly sad if the newbie is the one who innocently dumped it straight into Casual and no one ever looked at it again.
The feature request (from 2019!) might be valuable. But still…
The problem in this case is not the DQA, but the second and third identifiers misclicking on a taxon name that is duplicated across kingdoms. The best solution for this by another identifier would be to post a comment addressed to those two identifiers, asking if they had misclicked and requesting that they withdraw or edit their IDs.
If we want to spare identifiers of this task by automating something into iNaturalist, the best solution would be a pop-up warning when someone is entering an ID in cases where these generic names are duplicated. Something like “Do you mean Viola the plant or Viola the animal?”
A similar warning could also appear when common names are the same for different species.
Actually I think they’re both problems. The DQA should never have been clicked on something that was completely identifiable (I gather), and the reason for it having such a high-level ID is irrelevant. My first reaction to the initial suggestion was ‘but what about all those utterly unidentifiable photos!?’ - but a little thought showed that two IDs in two different kingdoms suggests the photo can’t really be unidentifiable. So while I’d still be a little reluctant to remove a valid way of making something Casual, I suspect that a kingdom disagreement at Life would never be a valid use.
Honestly, what I’d really like (but understand is unlikely to happen because of the potential for misusage) is a DQA option of ‘terrible photo’ which doesn’t require two IDs. I’ve certainly seen plenty of photos that could well be what they claim - but are bad enough photos I’d never be happy to agree. At the same time, I’m reluctant to shove them back to Dicots and DQA ‘as good as they can be’ because they might be right. As a result, they stay in Needs ID to annoy future IDers.
But back to the other question of duplicate (or similar) names, I’d love to see such warnings implemented. I’d also love a ‘kingdom icon’ or similar for each item in the ID selection dropdown - that should help avoid this problem, but would also be helpful more generally.
In this Viola example - I would like iNat to SHOW us the word Lepidoptera. It is not useful to display an unfamiliar (or it would not have been clicked) Genus as an unattached orphan.
iNat does not present this as flower or butterfly, it is just - oh, hello Viola!
It seems to me that the underlying issue in the case at hand is not misuse of the DQA, but IDer error – two people not realizing that the ID they entered was not the ID they intended. The DQA was intended to apply to an ID of Viola (the plant genus), and as such, would have been a reasonable use of this button (the first IDer commented that a species ID is not realistic without flowers). Changing how the DQA works might prevent the observation from accidentally being made casual in a case like this, but it would not really correct the issue; unless one or both of the IDers are made aware of their mistake, it is likely that the observation would be stuck at “life” for a long time even if it were “needs ID” instead of casual.
That said, I do think that the way this DQA is currently implemented has some potential for abuse because under certain circumstances it allows users to essentially override the community ID process – I could imagine, say, a scenario in which a user dislikes seeing photos that they feel are poor quality and therefore disagrees at a high level and uses the DQA to make it casual on principle, even though it is likely that the photo allows for a more specific ID.
But other DQA items are also subject to abuse (people using “no evidence of organism” to make drawings casual because they don’t think these are valid evidence, or “captive” for hitchhikers because they don’t think non-established species should be on range maps), so to some extent I think this is an issue of getting users to agree on community norms rather than a problem that could be prevented merely by changing how the DQA buttons work.
As has been brought up on other occasions, a bigger problem is the way that an observation can become RG for all practical purposes at species level based on a single species ID plus the use of “ID cannot be improved”.
I have sometimes wondered whether there is some one-sidedness in cases where an IDer pushes back an ID to genus and clicks “ID cannot be improved” because key features are not visible; while this is a pragmatic solution that is useful when trying to clean up a taxon, it does feel slightly heavy-handed – I am disagreeing with one or more species IDs and unilaterally declaring that the observation cannot be ID’d more specifically. Obviously other users can still override my disagreement by adding additional species IDs, but if the idea is that two independent assessments are required for an observation to become RG at species level, I suppose one might argue that two assessments of “a species ID is not possible” would be desirable before making an observation RG at a higher level.
I think if this were to happen, people would need to be encouraged to use that option more often - because there are plenty of cases where an observation really can’t go beyond a certain level, and to deliberately increase the number of observations in Needs ID which can’t be identified further seems a backward step.
Perhaps the ‘no it’s as good as it can be’ box should be disabled when the community ID is broader than all of the given IDs. That would be simple. It wouldn’t remove all possible examples of this type (someone might have filtered it to a broad ID early on in the process, for example), but it would help.
I wasn’t proposing that the button be changed to require 2 votes, just reflecting on the asymmetry inherent to the way it works.
(If I disagree with a species ID by pushing it back to genus, the community ID – consisting of my ID plus the ID I disagreed with – will be at genus, but it doesn’t represent a consensus in the same way that two genus IDs would, because while the person who entered the species ID also agrees that it is in the parent genus by implication, they may not agree with me that a more specific ID is not possible.)
