Kindness, or photo quality?

This is a photo from a school project.

https://www.inaturalist.org/observations/33428429

There are several other photos from this place, so I am absolutely certain the curved green thing is indeed Trifolium repens. Otherwise, I could never tell. Students need ID’s because they’re encouraging. So, should I put a name on this?

2 Likes

If you’re sure what it is, then I think you should ID it.

6 Likes

I just ID’d a bird, not because the bird was recognizable, but because I recognized myself holding the bird in the photo (and I remembered what bird it was from about an hour earlier). So I completely agree with @alexis_orion, if you’re positive of the ID, then go ahead.

Also, the answer to the topic question should always be kindness. Too often it isn’t.

14 Likes

I say kindness, unless I’m absolutely sure the student is approaching the class project in bad faith. This may be their first exposure to understanding and observing what’s in the world around them.

If they post photos of blurry organisms that we can’t tell by general shape, I find giving a few tips or asking a question helps. If they never answer, it’s no big deal.

8 Likes

I would probably agree with the ID via the comments area. Maybe a moderator would weigh in, as I say that because I don’t know what and how an ID to a fuzzy photo affects the computer AI.

2 Likes

One fuzzy photo is not going to impact the AI ability to identify a species, which as I write has over 21,000 observations. If anything having a range of quality (as well as features etc) helps as it more accurately reflects what is submitted.

In terms of if you should ID it, I personally cant tell if that is clover or someone spilled their spinach salad on the ground. It still comes back to this question to me, if shown the photo, told when and where it was taken, and given no other information, what would your ID be?

You should feel no sense of obligation to identify something for someone doing an assignment etc.

6 Likes

usually, if i id based on other observations, i’ll simply note what the other relevant observations are in the comment.

5 Likes

In this case (and many similar cases), the observer quite likely has no idea what they are observing. Rather, they are using iNat as an identification tool and simply agreeing to the top suggestion from the Computer Vision system. So another agreement is really just reinforcing the Computer Vision suggestion. That may or may not be a good thing.

2 Likes

I usually link them, with a statement like “based on a classmate’s photo.” Not super useful for app users, but still some explaintion of how I could possibly know.

6 Likes

In this case, I think the observer is agreeing with a classmate’s identification. I doubt the computer vision could do anything useful with this.

Linking to some of the other photos is a good idea. I don’t have time for it right now, but I may, later.

3 Likes

The icon on the observer’s identification indicates that it was a Computer Vision suggestion (as described here: https://www.inaturalist.org/pages/help#cv-select).

I have actually seen many photos worse than this example get a CV identification - in fact I would say that a significant weakness of the CV system is that it will always provide a suggestion.

5 Likes

Never knew that.

3 Likes

I’m not entirely convinced that an ID based on the fact someone else submitted that species is the safest approach. Should an ID not be based on the evidence in that observation?

4 Likes

if i see that two people submitted observations at about the same time in about the same location, and they label it as the same species, but one has a blurry photo while another has a very clear photo, i have no problem saying that the blurry photo observation is the same species as the clear photo observation, as long as the blurry photo isn’t obviously something very different. if someone wants to suggest a different id, i think that’s fine.

5 Likes

At least 5 students in this class submitted photos of Trifolium repens from the lawn, on the same day at the same place. Three of them labeled it as T. repens. Trifolium repens is by far the most common clover that looks like this at this season. I think it’s a safe bet that this is what it is. I’ve added links to two much better T. repens photos from this set.

Since there are 5 replicates (not duplicate; from different people), this observation is of little importance to iNaturalist, but it may matter to the person who submitted it.

6 Likes

I think there’s a middle ground here, with two possible solutions. One is to not officially submit an ID, and just make a comment about it, and the other one is to officially submit an ID, but to either mark the observation as not having evidence of the organism, or to mark it as the Community Taxon not being able to be confirmed or improved. That way the student gets an ID, but it doesn’t mess with the AI, with people going through observations, nor with research. I don’t think you have to lie to students and tell them that they’re making hugely valuable contributions to science when they’re not. They’re mostly fully aware that they’re just learning, so that might prompt them to improve their observations.

4 Likes

actually, it doesn’t always provide a suggestion. i’ve had a few observations of well-camouflaged organisms that the computer vision didn’t provide suggestions for. i assume that any of the suggestions computer vision makes must fall above a certain confidence level.

1 Like