Disruptive disagreement on identifications

@spiphany, Thank you for a very clear explanation of the problem. This should be fixed. I wonder why it has not been.

2 Likes

With an original ID of Anthophila, iNat considers a disagreeing ID of Hymenoptera to be a disagreement with Apoidea, Aculeata and Apocrita and by extension the child taxa within them. The only taxon within Hymenoptera with which it is not considered a disagreement is Symphyta (because it is not an ancestor of Anthophila).

Typically in a case like this, the person entering the disagreeing ID only really wants to explicitly disagree with Anthophila (if they knew what it was, they would not be entering a high-level ID), but this is not how the algorithm interprets it.

See the blog post here: https://www.inaturalist.org/blog/25514-clarifying-ancestor-disagreements

1 Like

Your day is my night. I’ll think it over tomorrow.

Now we have that line of text tucked under your ID
sedgequeen disagrees with Species A

But iNat should follow thru
sedgequeen disagrees with … each of the the subsequent IDs (which she has not even seen and may in fact vehemently AGREE with)
to remind us all that the Ancestor Disagreement lurks in the shadows wreaking havoc.

One right and one wrong cancel each other out.
If the third and fourth agree - they can go to RG. As they would have done, if the first pair were not there.

But my view is skewed by hunting down identifiers for the third, fourth. Good grief, fifth. Never mind 6. And 7! There just are NOT so many identifiers working on iNat.

1 Like

Reading through threads like this, I tend to think the problem is not iNat’s system, the problem is understanding and usage of INat’s system. Ancestor disagreement issues are subtle and not apparent to the naive user (which I was not that long ago!). That suggests that the solution is user education. Obviously not enough people are going to read the forum or the guidance docs, so any education has to inserted into the UI in a way that is unobtrusive. There could be a very brief blurb followed by a link to the documentation that explains more fully. Messages like this can be set up to appear the first time, but have a way to be turned off to not annoy more experienced identifiers.

Something like “You are about to enter an ancestor disagreement, an ID that asserts that this organism is NOT any of the following: [list of taxa] - are you sure? See [Ancestor disagreement].”

3 Likes

It isn’t necessarily just a problem with not understanding and using iNat’s system optimally.

Sometimes we need to disagree with the current ID of an observation and don’t have the knowledge to suggest an ID that is lower than order or whatever. Sometimes doing this results in ancestor disagreements.

The effects of this can be mitigated if users disagree at the lowest level possible and keep up with their notifications so they can withdraw in case of ancestor disagreements. But it is easy to miss notifications, particularly for people who make a lot of IDs, or one might be absent from iNat for a time due to other stuff going on in one’s life, etc.

The current system puts the burden on users to avoid hindering the process of arriving at a community consensus and – I would argue – doesn’t in most cases offer substantial advantages over a system that would allow one to disagree with a specific taxon only.

2 Likes

Sometimes I’ve just added a comment as to why I think the existing ID is wrong and what taxon I think it can be reliably IDed as (such as genus or family). That gives the observer and subsequent IDers something to work with, without causing disruption. They can choose to use my comment or ignore.

3 Likes

Problem is, I really don’t know much beyond Bees. So even if I could readily see that the individual is a wasp, I don’t know how closely related that wasp is to bees. So, yeah, I’m going to toss it up to Hymenoptera.

Because just because the organism is more closely related than what I put, if I don’t know, I’m going to put it as low as I know. If a few more people come along and move it in the right direction, I’ll withdraw my ID.

2 Likes

OK. It sounds like what might help is a combination of hard and soft disagreement. Meaning, an ID that is a vote against certain taxa, but not all taxa in the chain up to the chosen ID taxon.

I hard disagree it’s a bee because I know bees.
I say Insecta, but all the intermediate steps between Insecta and bee could be considered soft disagreements. If someone comes along and says it’s a fly, then it can go to RG because my disagreement with the order Diptera was only a soft disagreement rather than a hard disagreement.

The UI could present a UI that lets the user choose where to disagree and where to say “I don’t know”, given the taxon choices between the current ID and the chosen ID.

I wonder if it’s possible given the same database structure. Seems like it could be, if there’s already a record that distinguishes the two types of disagreements. If so, it just has to be settable with more intermediate options.

A key factor is to keep things simple for newbies and all users.

1 Like

I would be very happy if I could
either say it is not Species A
or it is a wasp

And for my ID to hold and count only there, where I put it. That Ancestor Disagreement is passive and cryptic, makes it harder for our good ID intentions to actually do what we intend.

I don’t think this is complicated, in fact I think this is what most users would assume the algorithm is doing.

2 Likes

Some contend that is is complicated for many users, and I have to agree. Or, at least cryptic. I read these threads and I just can’t get motivated to “ID”. Beyond pitfalls, it’s difficult to find information. It’s there but it’s too much work. And I already work 40+ hours/week. (It takes me months to circle back to one of my uploads that needs an update or response).

But, I appreciate that the rest of you who do what you do. And, FWIW, I continue reading all these threads, assuming that one day, by optical osmosis, I will have internalized the methodologies and details to do ID’s.

https://forum.inaturalist.org/t/puzzled-by-disagreement-consequences/43966/2

2 Likes

@jagerwin my comment was in reply to @spiphany who made an alternative suggestion for how it should work but wondered if it was complicated - but I think their suggestion is more intuitive than the current set-up - which is undoubtedly complicated :)

2 Likes

Well, by “complicated” I was actually thinking about the algorithm that determines how the the votes for or against a particular taxon are counted and what the community taxon is. I do think that my suggestion is more intuitive from the user perspective, but I’m finding it a bit harder to translate this into the mathematical calculations (how does the system know to display the ID as Magniolopsida at step 1 after the initial disagreement?).

However, I see that the proposed system I outlined is essentially the same as the alternative model described in the staff blog post (from 2019) I linked to above, so presumably it is feasible from a programming standpoint. It does make me wonder a bit why no change was ever implemented or if there was a deliberate decision at some point in favor of the current model, and if so, why.

2 Likes

Basically, identification is simple. Give the organism the name you think is right, at the lowest taxonomic level you can. There are all sorts of complications, yes, but you can’t go very wrong doing that, and some of the “wrong” is just people’s preferences, anyway. Please ID when you have the time and interest. And when you don’t, I hope you have fun with iNaturalist in other ways.

6 Likes

“I know that there are people with the experience in the right areas that will id it.” But in my experience this doesn’t happen without tagging someone or contacting the original disagreeing ID-er. Otherwise observation stays at a very high level.

True that some disagreements save time for the disagreeing specialist identifier. Later on…

Let me add a fresh example.
https://www.inaturalist.org/observations/20684433

Id by observer: Stenotaphrum secundatum gehört zu: Familie Poaceae Süßgräser

Disagreement: Reich Plantae Pflanzen

the evidence is not sufficient to identify this species.
the flower morphology prove that this is other type of species.

And I see that the family was correct. Now we are still at Plantae. Another identifier is required to get it to Poaceae. And then 4 more identifiers to get it to the correct genus/species.
The observer left us 4 years ago. The disagreeing identifer 3 years ago.

Do yo think we’ll ever arrive at a good id?

Yes. If you push, nudge nudge.
It is at Family and taxon specialists filter for that.

That’s certainly the problem I’ve seen trying to resolve the IDs on some records. If the participants don’t stick around, it can be hard to fix these situations and the observer probably doesn’t care anymore since they’ve gone elsewhere. So my time is better spent elsewhere as well.

2 Likes