Human obs used for cyberbullying

[insectobserver123]

@wildskyflower

True, street view AI specifically may not have been the best example because it is optimized to almost never miss readily identifiable human faces

This is not how it actually works though, it seems to know that a person only has 1 face, every person whose butt is blurred has a non-blurred face, and people near street signs also don’t have blurred faces, the sign is blurred instead

If we blur faces it should be limited to obs of a human, not all obs with a human, as obs with a human and an organism may blur the organism as well, my camera’s face recognition locks onto certain parts of ants’ anatomy, for example

@pmeisenheimer

Yes it is. Explicitly. It has been proposed that images in which the subjects appear to be expressing an unwillingness to be photographed should be removed.

Only on inat, I am not advocating this policy be adopted on other platforms, nor am I advocating any other censorship anywhere, as long as inat is not the platform you are sharing these pics on you are not impacted by this

@wildskyflower

I just thought of an example of an observation where I think a piece of the ‘engagement with nature’ aspect of inat would have been lost without the photos that have human faces:

https://www.inaturalist.org/blog/22612-holy-mola-the-oral-history-of-an-inat-identification
https://www.inaturalist.org/observations/20522357

I am advocating no changes to photos like these, I would hope any policy implemented does not affect observations targeting an organism where the humans are incidental or for scale

This obs would not be censored by the policy I proposed, becasue the humans are not the target of the observation, are not patients in a medical facility, and there are no bad comments about them in the notes. It also would not be censored under tiwane’s proposal becasue the community taxon should be Mola tecta, not Homo sapiens

@tiwane

My proposal was to hide photos of observations with a Community ID of human, which means at least two users have IDed it as human. So one person wouldn’t have the power to hide a photo. And it could say something like “photos attached to human observations are hidden from view”.

As long as it is the photo and not the whole observation that gets hidden, I’m good with this. I just don’t want a situation where something gets 2 human IDs and then no one change their ID if they made a mistake

@cthawley

Maybe I am missing something, but if the photo is hidden, how do other users identify it as non-human? It seems like once it is hidden, the Community ID is going to remain as “Human” due to evidence to the contrary being unavailable.

If I post something as genus carpenter ants with notes saying “can’t tell if the thorax is red enough to be a new york carpenter ant or if it is just an eastern black carpenter ant”, and some trolls IDs it as human, other users can tell that I am someone who posts ants a lot and am not a troll, and can tell that the human ID is from a troll, so they can add generic “ant” IDs to make the photo visible and then we can get back to looking at the image to decide what kind of ant it is

@cthawley

basically proposes making moderation decisions based on a user’s reputation or other users’ opinion of the affected user, which I am pretty strongly opposed to. Being a known user is no guarantee that a user has acted appropriately - Many established users occasionally (or even frequently) make mistakes or violate guidelines. And a system like this would seem to be inherently biased against new users without a track record or users who aren’t known to whoever is doing the proposed moderation. New users often need education, not instant repercussions based only on the fact that they are new.

I am strongly opposed to that too, one of the first forum threads I was in I joined to object to a proposed reputation system. What I am saying is that if trolls falsely ID as human and the observer disagrees, other users can look at the observer’s observation history, determine that they normally make real observations, and add generic non-human IDs to make the photo come back as I described with the carpenter ants (and if the photo is human after all, these identifiers withdraw their IDs and report the observer for trolling lying about the human). It is true that a new user is less likely to find people to disagree with the troll human IDs than an experienced user, and this is an argument against auto hiding humans, but this problem is not improved by preventing anyone from disagreeing with human IDs, and nowhere have I proposed harsher treatment of new users who make mistakes

[pmeisenheimer]

@swampster

The latter part of the OP’s sentence that you did not include in your quote made it clear that they were not talking about photos where humans were not the focus of the observation.

The complete sentence reads:

@insectobserver123

This is exactly what I am proposing, I would argue that this flag should be used to hide all photos deliberately targeting an unaware or not clearly consenting person (as opposed to photos of a tree in a public place that happen to include bystanders)

The example serves to clarify but in no way alters the meaning the parts of the sentence I quoted. It also does nothing to address my objections to the proposed action.

@swampster

Almost all if not actually all the conversation regarding selected (vs. hiding all “human” observations) full photo removal (i.e., not partial removal such as facial blurring) has been regarding observations where humans are the target of the observation and the observation is irrelevant at best and malicious at worst.

Which in no way minimizes the issues I have with what is proposed.

In the very first post, the OP wrote:

@insectobserver123

To clarify, I am not arguing against human obs generally, only that there needs to be a way to flag and remove certain potentially malicious ones

I’m not sure exactly what the best policy on this would be, but I would like to create a discussion about the issue, I think photos with certain of the characteristics listed at the bottom of this post, or certain combinations of these characteristics, should be covered by the policy I am suggesting, but I am not saying exactly what the policy should be, as my goal with this post is to to create a discussion of how this should be approached, rather than a feature request for an exact policy

I have made clear in a couple of posts that I support aspects of this, the call for clear policy especially. I have also taken the OP at their word that they are looking for a discussion. That’s a good thing, as long good-faith disagreements are accepted.

Many of the specifics of this issue are addressed in policies developed elsewhere. I linked to one for Wikipedia Commons in an earlier post. There are others.

Anyway, I made a very specific comment about a very specific proposal. It was not about general questions related to joke posts or any of a wide variety of subcategories of conversation into which this topic has drifted over its course. It concerns the advisability of using subjective judgments about the mental or emotional state of people depicted in photographs as a criterion for hiding images. The base case mental state of interest is consent, I guess. Other considerations, particularly as concerns bullying, can be inferred from the ongoing discussion. This is a bad idea, in any context. I’m kind of surprised that this is even slightly controversial. There are people who claim to be able to determine, with certainty, what somebody is thinking by looking at a photo. They cannot.

There is a tendency, in these sorts of conversations, for something that I have come to think of as the Mrs. Lovejoy Effect to kick in. It’s called a bunch of other things, including What-About-the-Childrenism. It has the effect of guilting people into a false consensus. It’s the opposite of useful if the objective is a factually informed, rational discussion. I’m sure folks here would never go down that road but it’s good to keep an eye out for it in case it raises its head.

[wildskyflower]

@insectobserver123

This is not how it actually works though, it seems to know that a person only has 1 face, every person whose butt is blurred has a non-blurred face, and people near street signs also don’t have blurred faces, the sign is blurred instead

The paper that describes how streetview’s blurring software works is 14 years old; that is an eternity when it comes to computer vision technology. The systems available back in 2009 could barely tell the difference between a car and a refrigerator, and would have been utterly unusable for any inat purpose whatsoever. I have no idea if they have updated that system or not. However it seems like most complaints about failures to blur faces I can find are 10-15 years old.

[insectobserver123]

I was basing my conclusions about street view on my own observations of street view images within the past 5 years

[insectobserver123]

@pmeisenheimer

The example serves to clarify but in no way alters the meaning the parts of the sentence I quoted. It also does nothing to address my objections to the proposed action.

What I meant by the phrase “deliberately targeting” was photos taken for the sake of photographing a person, observations of a fish held by a person are “targeting” the fish, not the person, so they would not be covered by this policy I proposed unless there were derogatory comments about the person or the person was a patient somehow fishing inside a medical facility

[swampster]

@pmeisenheimer

There are people who claim to be able to determine, with certainty, what somebody is thinking by looking at a photo. They cannot.

Where are people claiming they can determin “with certainty”?

@insectobserver123

its hard to tell from where we stand what is harmless goofing off

@naturalist_nate

but as they appear on inat, you can’t tell if they are or not. and for that reason, I think they should be handled the same as cyberbullying issues.

I generally scroll past observations of humans when IDing unknowns. But after reading this topic I clicked on one I came across today. The very first one I clicked on was a close up of someone’s face that looked unhappy. The notes said “a very unusual creature”. It is possible this photo was a selfie and the observer was displaying self-deprecating humor -or- the photo was taken by someone else who was being a jerk -or- it’s an inside joke and both parties were aware and comfortable with the post. I claim no certainty, but again this post was irrelevant at best and malicious at worst and I would see no problem with removing it.

@pmeisenheimer

I support aspects of this, the call for clear policy especially.

Here I agree. Clear policies are a necessity. But I think those policies should allow for the removal of photos where malicious intent is not necessarily certain, but reasonably possible for any photos that are irrelevant to iNat’s purpose. Note: I want to emphasize that malicious intent must be reasonably possible for removal. Photos that are irrelevant but are clearly benign need not be hidden. And I understand “reasonably possible” is subjective, but so are many other judgement calls on iNat and this could be somewhat mitigated by providing examples in the guidelines.

[pmeisenheimer]

@swampster

Where are people claiming they can determin “with certainty”.

This was a reference to mediums, mentalists and other flim-flam artists, not anybody posting on this topic.

@swampster

malicious intent must be reasonably possible for removal. Photos that are irrelevant but are clearly benign need not be hidden. And I understand “reasonably possible” is subjective,

I think malicious intent matters but I have no doubt that a policy of this sort cannot be applied as a stand-alone without turning into a demoralizing swamp when it is turned over to volunteers to enforce. Extremely variable views of what’s reasonable will be inevitable. There needs to be a properly thought out and reasonably comprehensive policy on bullying and harassment that incorporates legal requirements as the minimum standard, explains what they are in plain language and builds on those standards in a careful, thoughtful way. This need not be as onerous as it probably sounds. There are existing templates. There needs to be as little subjectivity in the standards as possible, especially when penalties are being applied. There also needs to be a realistic limit on the expectations associated with the policy. Humans are imperfect animals who sometimes behave badly and iNat isn’t going to put an end to that. Ultimately, the only way to come close to eliminating bullying is by building a kinder, more collegial culture in the community.

For what it’s worth, I don’t fully understand what motivates most (maybe all) of the posts of humans on iNat. I find it kind of sad that anybody would get a rush out of posting some of the stuff that gets posted. But then, I can’t remember the last time I took a selfie, other than to check my self-administered buzz cut at the back, so what do I know?

[DianaStuder]

Can we step back to iNat is about Engaging with Nature.

A flag similar to DQA - no evidence of organism - ‘observation of people, not relevant for iNat’.

• how to hide the photo, similar to a copyright flag?

Sunfish with incidental people, or our obs with equally washed up deep ocean giant squid with people for scale. That is what iNat is for.

And separately, if the ‘picture only of people’ is among other relevant photos, that people picture should be hidden from taxon pictures. I would blame that for iNat suggesting the ‘people with mushroom’ should then prompt IDing any people as Fungi. There is a hidden reason WHY CV suggests IDs that seem weird to merely human eyes.

New people come to iNat as more social media. But here we have flowers, or whatever. They dump random photos which include people. Mostly not relevant. We need policy / guidelines with flags and DQA to apply.

[mazer]

There are always going to be bad actors on a site.
Thank you for caring.
I have seen all sorts of questionable observations posted.
Recently one person decided to mess with everyone and make the most obscure nonsensical identifications she could, none of which had anything to do with the anything in the photo.
All we can do is continue to self police these observations. It takes the fun out of the equation, especially when it involves bullying.

[anon83178471]

@mazer

obscure nonsensical identifications she could, none of which had anything to do with the anything in the photo

If you didn’t, or if no one else did, please flag these when you see them. That makes it a lot easier for curators to find these types of situations and work on resolving them.

[sandboa]

I think the idea of removing photos for this is excellent and frankly I can’t see a valid counter argument. We should be able to remove media that is used as harassment. Frankly, I think our obligations to do so are both legal and ethical.

In regard to audio, it gets a bit tricker. Is someone’s voice recorded incidentally and without consent really harrassment/bullying? I guess it depends what they say. But at least in 11 US states and the EU (and who knows how many other countries), it could be considered a violation of two-party consent laws. It might be illegal to post such recordings on iNaturalist.

[insectobserver123]

I’m not referring to anything recorded incidentally unless there is some particular privacy concern with that observation

This is really about observations where a human is the primary object of the observation

[insectobserver123]

I will also propose that human obs have auto obscured location, as is done with endangered species

[doodoodoodoo]

I’d definitely be in favour of a flag that hid these images automatically. Could the flagging feature be used maliciously? Sure, but I’m guessing that’s true of things like copyright reports. It’s not a whole new, unprecedent feature being suggested. Maybe the staff or some trusted users could look through the flagged pictures and un-hide them if appropriate.

I wouldn’t even be opposed to observations of humans being banned or severely restricted. Yes, we’re animals and we shouldn’t consider ourselves to be separate from nature, yadda yadda. However, most observations on this site already imply the presence of humans at a given location (someone had to take those pictures or set up those trail cams) and I’m not sure what can be gained by the observations of humans. Scientists and scholars interested in studying human distribution around the world have much more detailed and comprehensive resources than iNat’s human observations, which are kind of sparse and are often jokey selfies and the like.

The ones where random students are identified as various animal species or insults should definitely be hidden or deleted. I’m sorry, but I don’t see what can be gained by making children and teenagers feel exposed or bullied. If I were in their shoes, I certainly wouldn’t become a power user of a site that enabled my bullies. :/

[insectobserver123]

Tiwane has said it would be technically difficult to implement this flag, but auto hiding photos when the community taxon is human would work

[doodoodoodoo]

Thanks, I must have missed the technical bit! Doesn’t the copyright flag work like that, though? Anyway, I’m fine with any solutions from a technical perspective, as long as the issue at hand is solved/addressed :)

[insectobserver123]

The problem is that curators don’t have the ability to remove photos, so if they resolve the flag the photo is visible again, meaning that if the photo is bad the only way to keep the photo hidden is to never resolve the flag, which means that every valid flag never gets resolved, and stays on the list of unresolved flags forever. I imagine this is confusing for curators and clutters up the list of unresolved flags, making it harder to find the real unresolved flags

[DianaStuder]

I tried using the copyright flag as a workaround.
That was immediately resolved - as it isn’t a copyright issue.
End of.

[bartwursten]

I totally agree. It would not only solve the problems presented in this post but to be honest I fail to see the value of records of humans. In my opinion it serves none of the purposes of (popular) scientific data gathering that INaturalist stands for.

Greetings Bart

[anon83178471]

IMO, you can always make another one and say “this is to hide the picture” or something like that. I usually leave a comment that just says “to hide” or “to hide, sent to staff”. Something like that so nobody touches it.