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

in your blog example, are the titles of the Leading Disagreements and Branch Disagreements diagrams correct? (if they’re wrong, i think that explains some of my confusion.)

They are as far as I can tell. Can you point out whats confusing you?

here’s what i’m looking at:

three things that confuse me:

  1. the highlighted text talks about branch disagreement, but the diagram is labeled leading disagreement
  2. what’s the difference between community taxon and observation taxon?
  3. if the leading disagreement diagram is correct, if at the time B makes the Insecta ID, the only other ID is A’s 7-spotted, how does the system determine that the observation taxon is Lady Beetle (family) as opposed to Coccinella (genus) or Beetles (order)?

ah - good catch. I had the Figure 15 where Figure 16 should have been and vice-versa. I just fixed them. Thanks for catching that and apologies for the confusion

ok. that addresses confusing thing #1, but what about #2 and #3?

re: #2, I added this to the first section of the blog post to clarify:

In contrast, the Observation Taxon is the finest ranked taxon with at least one agreement that has no disagreements. The Observation Taxon will match the Community Taxon if: (a) there are no taxa with agreements and no disagreements finer than the Community Taxon, or (b) the taxa with agreements an no disagreements finer than the Community Taxon is of rank subspecies. If the observer opts out of the Community Taxon, the Observation Taxon will always be the Observer's identification.

re: #3, in the cartoon taxonomy there is no genus or order, just family. But the observation taxon is the finest taxon with at least one agreement and no disagreements, so if there were other nodes like genus in the taxonomy finer than Lady Beetle Family with no disagreements the observation taxon would be set to that

sorry for the confusion

i don’t like the proposed behavior re: #2 and #3. suppose our example taxonomy does include genus and order. in your Leading Disagreements example then, i would assume the Observation Taxon would be Coccinella after B’s ID. what if B knows it’s not Coccinella? how does B disagree with that genus? does he have to disagree twice? (once to get it to genus, and then again to disagree with the genus?)

i think it would make more sense for the observation taxon after B’s ID with Leading DIsagreement to be Insecta.

If I am reading it correctly, just disagreeing to family with the new option would achieve what you are asking.

no, in my interpretation / modification of the example, B was sure it was not 7-spotted, but thought it could be another lady beetle or another insect. if he IDs family and disagrees at that level, that doesn’t take into account the other insects he thinks it could be.

here’s a more real-world example that i deal with frequently. i monitor the genus Pisum (peas) worldwide. a lot of people around the world will classify any member of family Fabaceae (legumes) as Pisum or even any plant that vaguely resembles a legume as Pisum sativum. Fabaceae has subfamilies, tribes, and genera defined in iNaturalist’s taxonomy. there are only two species within Pisum, and they are fairly distinctive. a lot of the time, i’ll be able to tell the plant is not genus Pisum, but i won’t be able to tell you if it’s even Fabaceae or not because i don’t know every plant in the world. in the current system, most of the time in such a case, i’ll just leave a comment saying it’s not Pisum. i might make a suggestion based on similar looking plants in the area, but i won’t make an ID because i don’t want to do a “branch disagreement” up to Dicots (class Magnoliopsida). (if i do a branch disagreement up to Dicot, i’ll just have to come back and remove it after someone makes a better ID.)

the proposed “leading disagreement” option offers a potential solution, but according to @loarie’s explanation of how they plan to implement it, if the existing ID is species Pisum sativum and i ID as Magnoliopsida with Leading Disagreement, the effect of that will be that the Observation Taxon goes to genus Pisum, even though i know very well it’s not Pisum. so effectively, the solution is ineffective at best and makes things even more confusing at worst.

if they implement “leading disagreement” in the way that i say makes more sense to me, if i ID as Magnoliopsida with Leading Disagreement, then after my ID, the Observation Taxon will go to Magnoliopsida. in either case, if you come later and ID as genus Vicia, then i believe the Observation Taxon will go to tribe Fabeae, which is fine. but in the period between my ID and yours, i would rather the Observation taxon show as Magnoliopsida rather than Pisum.

or consider that Observation Taxon is Pisum after my ID, and you disagree with it, too. You ID as Magnoliopsida because you’re not sure what it is either. what does the system present to you to disagree with? does it ask you if you disagree with species Pisum sativum? does it ask you to disagree with a non-existent genus Pisum ID, or does it not offer a chance to disagree because your ID is the same as mine (even though the Observation Taxon is at genus)?

2 Likes

Unless I am reading something wrong, I don’t think this is correct. In this case, your leading disagreement would bump the community ID to Magnoliopsida (provided there are not more than 2 agreements at species level, of course.)

1 Like

yes, the Community Taxon goes to Magnoliopsida, but the Observation Taxon goes to Pisum according to my reading of @loarie’s diagram and comments.

I don’t think Observation Taxon really matters, though (I’ve always been confused as to its purpose/function). Community Taxon is what shows up in the data and in searches, as far as I am aware.

no, observation taxon is what shows up on the observation header. community taxon is what shows up in the little community ID box. you can see this when someone rejects the community ID.

1 Like

Sorry, I must have gotten them mixed up. In that case - I agree with you and don’t see why the taxon would bump to an intermediate level (family) when no identifier has suggested it explicitly.

1 Like

i think you guys are actually talking about 2 different use cases. i think @loarie is talking more about a case where someone can definitively assert that an earlier ID is not correct, whereas @kiwifergus is talking about a case where the evidence is not definitive. here’s an example of what i think @kiwifergus might be talking about:

suppose A (from @loarie’s blog post) makes an observation that she IDs as a 7-spotted lady beetle, and the evidence is an extreme close-up photo of a lady beetle’s elytron. B looks at the photo, and all he sees is a mostly red rectangle with a couple of black spots in it. maybe there’s black leg at the edge. maybe there’s part of the pronotum at the edge. based on that, all B can say is that it’s likely a lady beetle. it could be a 7-spotted, but maybe it’s an Asian, or maybe it’s something else.

in the current system, B IDs this as family Lady Beetle, and he’s presented with this prompt:

Is the evidence provided enough to confirm this is 7-spotted?
Choice 1 (green): I don't know but I am sure this is a Lady Beetle
Choice 2 (orange): No, but it's a member of Lady Beetles

B thinks about the question “Is the evidence enough…?” and decides that it’s not enough evidence for him, but maybe there’s someone else out there who can look at the colors or the shape of the spots and actually confirm this is 7-spotted. so he interprets “Is the evidence enough” as “Is the evidence enough for anyone (not just me)”, and he picks Choice 1. This leaves the Community ID as 7-spotted.

but suppose B doesn’t think about or doesn’t think that another person out there could ID this to the species level. in that case, he would be comfortable picking Choice 2, which pushes the Community ID to Lady Beetles. or maybe B thinks there might be someone (not A) who could ID this to species level but doesn’t want the community ID to reflect species until that expert weighs in. in that case, B makes Choice 2 not because he thinks more evidence is needed for an expert to make an ID but because he doesn’t like the resulting community ID that Choice 1 would provide.

in other words, there are 4 thought processes (TPs):

  • TP1: I can’t ID to species, but there might be an expert out there who can ID to species --> Choice 1 (no disagreement)
  • TP2: I can’t ID to species, and I didn’t consider other possible experts --> Choice 2 (branch disagreement)
  • TP3: I can’t ID to species, and there are no others more expert --> Choice 2 (branch disagreement)
  • TP4: I just don’t want this to reflect a species ID --> Choice 2 (branch disagreement)

after the proposed changes, B will be presented with 3 choices instead of 2:

Are you disagreeing this is 7-spotted?
Choice 1 (orange): Yes, I'm certain it's not 7-spotted
Choice 2 (orange): Yes, I don't think we can be certain beyond Lady Beetles
Choice 3 (green): No, I'm not disagreeing

but notice that the proposed prompt becomes “Are you disagreeing…?” instead of “Is there enough evidence…?”. and notice the wording in the proposed Choice 2 is “I don’t think we can be certain…” (as opposed to “We can’t be certain…”). the effect is that you no longer need to know that there’s not enough evidence. now you just need to think that there’s not enough evidence. in the current system, TP4 requires that you ignore the system wording to produce a desired result. but the proposed (wording) change effectively eliminates the need for TP4 in favor of TP2/3. or looking at it another way, whereas TP4 involved making a rogue choice, TP2/3 are sanctioned choices.

so although you eliminate TP4 after the proposed changes, B’s remaining thought processes remain more or less the same, and the effect of each choice is the same:

  • TP1: I can’t ID to species, but there might be an expert out there who can ID to species – Choice 3 (no disagreement)
  • TP2: I can’t ID to species, and I didn’t consider other possible experts --> Choice 2 (branch disagreement)
  • TP3: I can’t ID to species, and there are no others more expert --> Choice 2 (branch disagreement)

in other words, all you’ve done in this kind of situation is encourage disagreement by officially sanctioning what once a rogue disagreement. (if that is not the intended / desired result, then maybe you can address that with different wording in the proposed prompts.)

that said, i don’t disagree that certain implementations of “leading disagreement” could be an improvement upon “branch disagreement” in other situations (see my notes in this thread about Pisum). (in fact, i don’t understand the need for branch disagreement at all other than backward compatibility. i would have preferred a model where you have to explicitly disagree with each distinct previously IDed taxon. see my earlier notes.)

1 Like

No…

Scott is describing the current implementation, whereas I am referring to a statement in that which suggests iNat policy (or at least sanctioned practise) is identifier B should be “correcting” identifier A by forcing the CID to what identifier B thinks it should be.

To put it in a nutshell, I think every identifier should be identifying based on their belief of what it is… NOT what they think the CID should be, which is what is often happening with the bump IDs. The bump should be a result of a correctly applied ID, rather than a desired objective obtained by “lying” about what you think it is…

2 Likes

Let’s say someone posted a very blurry picture of a ladybug from the eastern US 3 months ago. You can tell from the photo that the front end is darkish, and the back end is reddish-orange, and it has the shape of a ladybug, but the photo is too blurry to determine any pronotum or spot patterns. The observer has 10 observations, a generic username, and no profile photo or description. They identified it as 7-spotted using the computer vision, giving you no indication that they know anything about there being different species of ladybugs in the world.
To me leaving it as 7-spotted is incorrect. Of course it’s possible they do, but there’s no evidence that the user knows how to separate 7-spotted from any other species and from a blank slate it’s technically more likely to be Asian. I don’t see how it’s “lying” to say it’s a ladybug and shouldn’t be ID’d as 7-spotted? If it was a more experienced user I would probably more more hesitant to explicitly disagree.
I think it’s sort analagous to me calculating 1.57 grams on a chemistry question, but based on significant figure rules I have to answer 1.6 grams. It could well be 1.57, but my professor would give me half a mark off for saying 1.57 because it’s technically wrong.

2 Likes

in your example, in the context of the current system, i think the key is not whether the observer knows how to identify 7-spotted but whether it’s possible that anyone could positively confirm / deny 7-spotted based on the observer’s photo. if you think there is someone out there who could confirm or deny, then i think strictly going by the letter of current prompt, you should not disagree, and if you disagree, you’re sort of going rogue. however, if you know there’s no way anyone out there could confirm or deny 7-spotted based on the photo, then i think strictly going by the letter of the current prompt, you should disagree (branch disagreement).

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.

in the context of @loarie’s blog post, which is focused mostly on the addition of “leading disagreement” functionality, the key question here is whether they intended to also lower this bar for “branch disagreement” by wording the proposed prompt in this way?

1 Like

Yes, the magnoliopsida example has the poster wanting to change the ob id to magnoliopsida, not pisum. This counts as wanting to override the original identifier’s id. The cid is already correctly T magnoliopsida here. If the poster had gone first, then someone added an incorrect but more specific I’d, under old rules the ob taxon would still change as it has now, to pisum. There is no procedure for preventing the effects of incorrect IDs from others as compared to one’s own, which is what a change to knocking back the taxon Id to pisum amounts to here, given the lack of specific disagreement to the genus and above.

But, the options right now are to either disagree with the leaf only or disagree with everything under your chosen node,like insects, but none to disagree with an intermediate node, between the current I’d and the one you can give - here that would be an ability to disagree with pisum, not only pisum so. So branch and leading disagreements represent each extreme. What this requires of the user is to know the alternative branch of the tree to the current leading taxon, so they can put the more specific I’d themselves,e.g. pit which non-pisum branch of fabaceae it is on. They might not know this,and there night not always be one, in the case of paraphyletic groups where the presumed incorrect taxon is well nested and you want to indicate that anything else but that is acceptable.

@tchakamaura – just to clarify, the Pisum example is relevant only in the context of the proposed leading disagreement functionality (where the evidence is sufficient to positively rule out the existing ID). @kiwifergus’s objection seems not to be an objection to the proposed leading disagreement functionality but to a change to the proposed branch disagreement functionality (used when the evidence is not sufficient to rule out nor rule in). (or possibly he’s saying they should have gone in the other direction and removed branch disagreement altogether?) either way, these are 2 different issues.