False "research grade" observations

[bernhard_hiller]

I really enjoy using the computer vision feature. Because it comes also with fun. Yes, I do it for that fun:
E.g. I saw a mouse in my garden (in Germany). Computer vision still thinks it’s a Thomomys bottae (from the west side of north america): https://www.inaturalist.org/observations/42963952
Fortunately, the mouse did not jump upon me and bite me with its strong incisivi. I hope it won’t fell the tree next to it either…
With a rabbit in the vineyards, I saw suggestions of large african mammals.

And, I have fun with other people agreeing to such suggestions. E.g. a photograph of a millipede taken in Penang: “Texas Rattler”. That boy had good luck too, was not biten by the snake (the Penang school project was mentioned in a different thread on quality issues).

[bernhard_hiller]

@krancmm

Agree-bots .

When you open a taxon page, you’ll be shown a list of people who did most ids of that taxon.
Wanna be in that list?
Gaming the system is easy. Just search for observations of that taxon, and agree with the id.
Fear that you could be detected? Well, if you don’t agree with wrong ids, nobody can tell. So just take a closer look: who gave the id? One of the specialists for the taxon? Only then agree.

But …
https://www.inaturalist.org/observations/37191216
some frogs sitting in a flower pot, some where in south thailand.
Some one knowing frogs gives the id " Papurana grisea", another guy with a good reputation re frogs agress with that shortly after.
I am interested in the taxon, look it up. A frog from new guinea in thailand? Does not make sense. I ask the guys if they are sure. No answer.
Was that a trap intended by the first identifier? I do not know.

[Star3]

The thing is “CV said it” will only get you so far.

Knowingly & repeatedly posting joke IDs can and has resulted in account suspensions, and if someone determines you really should’ve known better than to blindly agree with an obviously wrong ID (especially if you’ve been already been warned, and/or you are an active & experienced user), then the fact that it was a CV suggestion may not matter.

Jokes are accepted in the descriptions and comments, not in the IDs.

[mertensia]

Situations like this are bizarre and frustrating. Sometimes I need to take an iNat break because of the lack of critical thinking- and I think the CV often enables that. Overall the CV is pretty amazing (and makes me able to put up my obs faster than I’d type them in) but it’s not always right. It should be more socially acceptable to say “I don’t know” than it is, and I think iNat should reinforce to only identify to the level that one can comfortably.

[charlie]

I’m not sure the algorithm is really the reason for these incorrect identifications. I’ve been on Inaturalist since 2011 and I feel like they were actually worst maybe 2013-2014 or something. I don’t think they are worse now “per capita” because of the ID algorithm.

[mertensia]

If it’s not the CV, then what makes someone id their plant as a New Zealand endemic?

[robertarcher397]

Because of similar common names. ID a tree in South Africa as Forest Cabbage Tree, a New Zealand endemic. The South African Cabbage trees are species of Cussonia. A Black thorn is not the same as Black-thorn. And once these ID are present on the wrong continent other people ID their plants as this.

[charlie]

Also there are times when the selected ID is absurd and it’s hard to imagine why it happened. Typos sometimes too.

[mertensia]

And if it has the icon that indicates they used the CV? Many of the examples linked in the thread do. I don’t know how you can say it’s not the CV when there’s a stamp on there indicating it was used.

[dianastuder]

But just because someone used CV, doesn’t mean that they relied only on that.
People often use that as a quick way to get to their chosen ID without typing out That plant

[fogartyf]

Yes - I often do this when batch loading on my PC. It tends to be more effiicient when uploading dozens of IDs, rather than typing the name out for each. At least with birds, it tends to include the correct species >90% of the time in the dropdown menu.

[sedgequeen]

I’ve put in bizarrely wrong identifications occasionally by clicking on the wrong name in the computer’s list. It’s a hand/eye coordination thing. Fortunately, people have pointed that out and I’ve fixed the problem each time, but it can happen.

[mertensia]

Yes, I use it as that. I’m impressed with how well it does but it’s not infallible. It has been “very sure” in cases where I know it definitely isn’t what it suggests.

In the case of a plant not in New Zealand identified as a plant endemic to New Zealand with a little CV certification badge, it’s purely the CV enabling bad ids. Not always an NZ endemic, but for example. This is what I’m talking about. Please do not tell me these misidentifications could not possibly happen because the CV.

I see this practice all the time. The problem compounds. Then it says “seen nearby” in the suggestion, leading to more misidentifications across the board.

[robotpie]

Damn, quite a shame indeed. In my early iNat days I must admit I was like that, blindly agreeing to things. It was a learning process indeed to be more critical of the IDs that others and myself give to observations. But we need to strike a balance between being stern to that user and also guiding them in the right direction. Of course at the end of the day they choose how to live their life so only they can take the matter into their own hands, because technically no iNat rules were broken.

[dianastuder]

all the time?!
This is a human user issue, not a CV issue.
It remains up to us to evaluate the suggestion, then accept or reject.

[mertensia]

Enabling is the key word here.

[sedgequeen]

American Plum (Prunus americana). Who wouldn’t agree that a plum tree in North America is American Plum? But those in the far western states mostly are other, similar species.