I see. You’re saying that it acts as a vote against the original species if more people later on agree it is species A.
I don’t fully agree; why would my vote saying that the picture can’t narrow something to species be negated by more people agreeing? My vote in this example is a vote against every other option, because I know that there is no way to tell if its species A or species b without information. I can’t tell its either because I acknowledge that it could BE either.
EDIT: if I’m still not interpreting you correctly, I’m not sure I can. because like, I can’t give you multiple genuses of fungi that are basically impossible to narrow done species for with just pictures. Like, 100% impossible you cannot tell without information like tree species, flavor, smell, even down to differences in spores. You cannot tell based off of macroscopic visual data.
There is no yes vote in those cases. You can’t exclude any options, so you are forced to pick the lowest tier that encompasses all the options
@lothlin OK, I can see how this can be a problem. (In the above-mentioned example, there really isn’t such an absolute impossibility of a species ID based on macroscopic visual data, in my opinion. And it deals with only 2 candidate species.)
So the issue becomes more complicated. Still, a categorical statement like “this is not the chosen species” when it very well could be (we just can’t know) should not be the preferred choice, because it is simply untrue and is not really what you intended to express. I guess there should be additional options that clarify your intentions more precisely. In a better world, it should be enough to comment and add a “non-disputing” ID (option 1). I’ve actually had a few cases where the observers then themselves downgraded their ID to a higher taxon. But that’s far from the majority of cases.
In addition to re-wording the current options so that they actually express what they mean, we need an option for cases where it is generally accepted by multiple experts that the provided info does not allow for a species ID. I’d still be in favor of requiring a majority of IDs (by two votes as is the case for outright disagreements) in order to “officially” display the downgraded taxon.
The other major example is when we’re dealing with an undescribed species - again, its happening a lot with fungi because so many of our species in north america were thought to be the same as their European counterparts, but DNA testing is showing that they’re separate species - but until someone publishes a paper on it, everyone basically has to ID to genus and make a note about the which undescribed species it is.
My approach to this is practical rather than philosophical. I usually ID birds, and sometimes someone upload a photo with (probably) a bird in it, which occupies at least 3, maybe up to 10, pixels. In this theoretical example, they ID it as an Alder Flycatcher, plotted in San Francisco. That species is very rare in the western US, so I ID it as “birds” and choose 2, because it’s almost certainly not an Alder Flycatcher.
I don’t interpret the “disagreement” as being that the bird in question is an Alder Flycatcher, but rather I’m disagreeing that the photo is definitive evidence of an Alder Flycatcher. I guess I use the “beyond a reasonable shadow of a doubt” threshold, because I will agree with an ID on a fuzzy photo when it’s unlikely that any similar species are expected in the local area, but that’s another topic completely.
First ID is wrong.
You come second, and hard disagree.
CID now counts that as one wrong + one ancestor disagreement, and demands more than two thirds right.
The effect is to require FIVE competent identifiers, where you might have been happy / lucky to find two. Without your hard disagreement CID only requires 3 against that wrong one - and that third is also not necessarily easy or possible to find.
You might not like the wording, but I think most people using either kind of disagree knowing what they are doing. Hard disagree can be used to mean “it is impossible that this is A”, but also as “there are several species this could be and there is no evidence this is A”. The phrasing of that question seems kind of clear to me in that way.
In your discussion of the logic behind it, I think you fail to include the layer aspect of taxonomy. Setting back the community ID level from say species to genus doesn’t mean excluding that one species in particular, it excludes any kind of species level ID and that can be the point. To come back to your logic lingo, if the state of science means that a single view picture of an organism in a given place means that it can only be identified as “species A OR species B”, identifying in the first place as species A contradicts that, and setting back the community ID to a taxon that includes both possible species restores it to an “accurate ID”.
The first ID may not be wrong in some cases. so iNat will have 2 possible IDs for an observation. If there is no hard disagree, the identity will remain the same I think. so it is ok for anyone to identify an organism at any level of its classification according to his or her ability. In the event of a hard-disagree, yes it will require more identifiers, so I think the system is intact. A very contentious observation may mean the particular observation is probably ambiguous or in the process of being examined and so not accorded its research grade status.
Something that has two different replies/meanings is hardly clear, I’d say
And the OP is (in my opinion) right about the unexpected/undesired consequence of a click: it results in a mention akin to “[species] is excluded by [this user]” (= “[this user] disagrees [species] is in this observation”) - even in those few cases where it was not what [this user] intended to convey)
So… there should be three different buttons for three different viewpoints: “this is definitely not [species] but one among many others”, “this is [genus] and maybe it is [species] but I can’t decide”, “this is [genus] and maybe it is [species] but in the current state of science no one in the whole world will be able to tell from such limited data”
There is a notion that some species of organism cannot be identified with just a picture. For fungus, bacteria, yes I can’t imagine. For butterflies, sometimes people can tell. We assumed that it is ‘impossible’. We can never know who are the other iNaturalist users out there. There could be up and coming young scientists whose skill are very good or has the backing from their DNA lab department. or perhaps a native to a land who has extensive knowledge of their fauna and can differentiate creatures in a way others find impossible.
I have learnt to check that the ID I added, does what I expected it to.
That the CID lands, where I expect it to.
If not - to look at where the algorithm is working as intended, but not as expected.
And perhaps to rethink what ID I add.
Or to hunt down another @mention to get the CID to finally work as expected.
Or to delete my ID, since it makes no difference.
I employ those strategies as well (as do other identifiers). Another strategy is to add a comment (not an ID) that states firmly (but not rudely) that the previous ID is not taxon X. That has the desirable effect of stopping the process in its tracks, so that the observation does not reach Research Grade.
There’s already an option for that, and its the ‘no, its as good as it can be’ checkbox under the question ‘can the community taxon be improved’ in data quality assessment
I disagree. I regularly disagree with ID’s, and choose option 2 to bump the ID up a level (or two) when the evidence does not allow for a positive ID to species. We can quibble over whether the options are worded correctly, but as far as I’m concerned, the result is what’s important. If I put a higher level ID on an observation and choose option 1, it has virtually no effect, and someone else with little or no expertise will likely come along and agree with the original ID, making it research grade.
No one should be expected to be all-knowing. We can only evaluate observations based on the evidence presented. Consider an observation for which there is no photo provided. It doesn’t matter what the observer tells us about what they saw, it’s still a “sight” record, and on iNat, it won’t ever be research grade (AFAIK). When a photo doesn’t adequately support ID to species, I consider it to be equivalent to a “sight” observation - because there isn’t photographic evidence to support the ID to species level. There’s nothing wrong with sight observations - I make a lot of them because I’m often too lazy to take a photo of every single butterfly I see (and some I see at a distance). I’ve seen experts make incorrect identifications on very good photos, where they have every opportunity to examine the photo closely and ponder it. Because of this, I’m a bit skeptical about what people claim they saw in the field, based on a momentary glimpse. If it’s something very simple like “it looked bright orange in flight”, I will give them the benefit of the doubt. If it’s some subtle detail that could easily be mis-interpreted then I fall back on the default of “we can’t know for sure because the evidence has not been presented”. I think a major problem is that folks get all wound up about observations not being ID’d to species and achieving research grade. We should all just relax and accept that not every observation is going to be ID’d to species level and/or achieve research grade.
As someone who does a lot of ID work, and who uses the resulting data for an external database, that is absolutely NOT and undesired or unexpected consequence. iNat is doing exactly what I expect and desire it to do. Yes, we could have a 3rd option, but I have trouble imagining what that button would actually do. What effect would it have on the ID?
Wow. In my experience, I’ve not found comments to reliably have that effect. There are lots of people who appear to merrily click “agree” on IDs (probably from the thumbnail view) without ever reading the sometimes lengthy discussions that are going on around IDs. I’ve seen many instances where I’ve bumped an ID up to genus, with a long explanation of why we can’t know that the species level ID is correct (or that it is explicitly incorrect, but the best we can do is still at a higher level) only to have someone else (often a novice user) come along and agree with the original ID.
If the organism is labeled Species A but we can’t tell from the photo if it is Species A or Species B, we need to get the observation out of the Species A group. Yes, it might be right but it also might be wrong. We want scientists who use this data to be able to have confidence that if the observation is species is labeled Species A, it is not Species B.
Option #2 gets the ambiguous observation out of the Species A group. That’s the goal. I think that’s good.
Option #1 lets the observation sail along as Species A, as if we know it is that. We don’t. I use option #1 for encouragement or when I know the organism is Species A but don’t know if it’s the subspecies the observer has selected, but I have reason to think the first identifier knows. Mostly, this is a bad option for cases where we know the observation cannot be ID’d to species from the photos provided.
Would changing the wording help? Maybe. The explanation might get very wordy and thus more confusing, but an actual improvement would be welcome. Improve it if you can.
Should we change how much power option #2 has to exclude later species ID’s. Probably. How much power should option #2 have? Something to debate. [Edit – option #2 makes RD harder to reach but not the way I said. See comment #24 above.]
Does it really exclude later species ID’s? I can’t cite an example off the top of my head, but I’ll bet there has been instances where I’ve bumped an ID up a level, and then someone else has come along and said “yeah, but what about X”, and I’ve reconsidered, and the ID went on to become research grade at the species level (either the original species, or some other closely related one). I’m also fairly sure I’ve seen someone reverse my bump by simply agreeing with the original ID (ie. there was no exclusion whatsoever). Maybe someone out there knows of a specific example where this problem has occurred, but it strikes me as an unlikely occurrence (if not purely hypothetical).
What I can definitely say for certain is that I often have to enlist help from multiple knowledgeable users to steer IDs in the right direction because occasional/novice users have entered incorrect IDs and appear to be oblivious to lengthy explanations of why their ID is incorrect. I find it remarkable that people with zero expertise are allowed to run wild and enter whatever IDs they like, yet there’s all kinds of handwringing about whether or not people with genuine expertise might bump an ID up a level. I can’t recall a case where a non-expert put a higher level ID on an ID and chose option 2. For my area/taxon this seems to be done only by a couple of experts, and we almost always provide an explanation when we do. Drive-by ID’ers putting incorrect IDs on observations is a much bigger problem. Am I suggesting that iNat should limit the power of casual users to enter IDs? No. I just deal with the problem as best I can without making a big fuss about it.
As I understand, this is the situation we have been asked to consider. I don’t know why someone bumping an Observation to genus using the red option is in any way detrimental since in the above scenario, the Observation should never have been identified to species in the first place. It was done erroneously.
Genus, Genus, can it be improved? No. Research Grade.
If at some point in time later it is determined that an aspect previously unknown allows such distinction by photos alone, that person is welcome to come sort through the historical Genus level observations to see if any have that aspect shown in photos.
