Nor do any number of other types of records add value (observations without media, observations where the photo is so out-of-focus that nothing can be discerned, pets in costumes, etc.). But they are considered permissible use of iNat. Humans are among the lifeforms on this planet. iNat has chosen to include them in their taxonomy, even if it has also been decided that they cannot be research grade (again, like many other types of records). This suggests that in principle, humans are not considered completely outside the scope of iNat and its mission of connecting people with nature.
People may choose to upload images of themselves or of others with consent. Should these photos also be removed merely because they show humans?
Photos do not have to show the person themselves in order to be problematic – for example, if they contain sensitive personal information like ID cards (e.g. sometimes schoolchildren have been imprudently instructed to do this in order to show that the photo is their own).
Observations with photos of humans include both observations where the person is the subject, and observations where they are not (e.g., observations of other organisms where a human is also present).
Merely removing all observations with an ID of “human” or all observations showing humans would not solve these other sorts of situations.
A flag “the subject is a human” for observations of humans that have been ID’d as something else would also not be a suitable response, because it does not address the offensive intention of the ID. It would also mean that there would be a risk that users start flagging observations of other organisms in which humans are present (insect on a person, person holding a caught fish or a plant) because they don’t think that images should include people or even parts of people (e.g. I have seen users object to the inclusion of hands or fingers in photos and tell the observer that this should be cropped out).
Using an AI to find images where humans are present and blurring faces or removing such images would also be highly problematic and would require review of every such record to check for false positives or photos with humans where humans are not the subject. Chances are it would not catch all images (many AI models have ethnic biases), which sends an equally problematic message to the people whose images are not being identified as human.
Hands are perfectly acceptable and recommended by official iNat resources for help taking clear photos of small subjects. I would reply or send them a message if you see someone asking hands to be cropped out. (And if you don’t feel comfortable doing so, staff or curators can help.)
My point was that there is already a certain amount of sentiment – correct or not – against inclusion of people (or even bits of people) in observation photos, even when humans are not the subject and even when there is no personally identifying material. This means that any solution to dealing with inappropriate photos of humans that is formulated in general terms such as removing all observations of humans or flagging photos to be hidden because the subject is a human is likely to be perceived by some as a signal that humans should never be included in photos, even in the form of fingers or hands, leading to overuse of flagging for images that are, in fact, perfectly fine.
…and would also be subject to false positives because of the “Ignore the elephant seal” phenomenon. Some of the more “seriously into iNat” folks may have gone through their old photo albums, found pictures where an identifiable organism is incidentally present (such as the tree they were standing in front of when the picture was taken), and uploaded them as observations of that organism.
I think most of us would agree that ‘human’ observations are generally on the less valuable side of things (although often acceptable, and sometimes valuable). I don’t know how easy it is to implement automatic face-blurring, and I agree that false positives are a problem, but this could be avoided if auto face-blurring (or some other auto-solution) were only applied to observations in the genus Homo. It also means that if a human has been photographed and uploaded as ‘baboon’ or whatever, even non-curators can take action to protect the individual simply by identifying as Homo sapiens, so that the auto solution will be applied. Of course, flags should also be made and addressed for abusive comments / ID’s etc. but at least there would be something that doesn’t need to wait for a curator response.
I don’t think that humans should be obscured from other photos, or that people should be required to crop their faces out. That’s just making the site harder to use.
I do think that it is an issue that warrants some further systematic safeguard.
As a new user I’m confused why people are so invested in having photos of identifiable human faces on the site, that doesn’t seem like the low-drama choice to make. Is there value in photos of identifiable human faces that I’m not seeing?
There isn’t value for iNaturalist in having photos of human faces (with rare exceptions). The problem is that young students assigned to use iNaturalist for class often post photos of their buddies or other people, often with insulting identifications. The question here is, that being true, how do we deal with the photos to minimize potential harm, considering lack of time by staff and others to deal with this?
An unfortunate trend seen lately is high schools in the US have seen a significant uptick in human observations. I’ve had to flag at least one human observation identified as “Apes”, “Primates”, “Monkeys and Apes”, etc… each day this week.
Teachers should also have an Inat account and make their students put their observations in a project(around half of human faces come from high/secondary schools worldwide). That may reduce the joke IDs and observations.
“To help manage privacy concerns, we automatically detect photos containing people and apply a Person tag. This tag prevents these images from appearing in most public views.”
You don’t have any guidelines for consent in pictures, but you SHOULD when it’s minors being photographed without consent, in SCHOOLS, WITH LOCATION. Anyone aware how creepy that is or just pretending not to care… you guys should have seperate terms for human observations. Because pictures of unknowing children should not be uploaded to then be made fun of or found, so weird.
Should observations with consensual photos of humans be deleted? Yes, this is not facebook, this is a site for citizen science identification of (primarily wild) species.
There are other classes of inappropriate content. Yes, but that is not an argument in favor of keeping THIS class of inappropriate content.
It is not inappropriate in itself that there are humans in a photo, it is inappropriate that the subject of the photo is a human. A human holding the organism to be identified is entirely appropriate and within the scope of this site.
The flag would be evaluated by a moderator, outcome determined at the moderator’s discretion. This is a suitable response.
I do not propose AI moderators or mandatory blurring of humans. I do not see the point.
I’ve flagged around 60 observations of humans identified separately from Homo/Homo sapiens. A few of those flags from a week ago are still open, and that is 60 flags.
I reviewed around 2500 observations. Potentially 5000 more dehumanizing IDs exist, which is a huge burden on curators, staff, and the flaggers alike.
If this is the case – and if the probem is prank IDs of classmates, which are likely to be nearly all from high/secondary schools – then this calls for a focused approach, not a blanket, sitewide intervention.
The difficulty as I understand it is that the person adding the prank ID is likely to be the same person nonconsensually uploading the picture in the first place; so the suggestion to restrict new accounts’ ability to add ID other users’ observations would not work well for that. Another difficulty is that these problem observations seem to originate from class projects in which the instructor already did not follow the best practices in the Educators’ Guide; which means that adding guidelines to the Educators’ Guide is unlikely to avail much. So, can we come up with focused ideas to deal with the specific problem?
Another problem is that identifiers and curators are already stretched really thin dealing with the new observations being posted on iNaturalist. We would like an approach that can be implemented with minimal attention from us. An approach that can also be applied to all the observations that already exist.
I suspect adding missed automated blurring, correcting wrong automated blurring or wrong automoderator deletions is likely to take more time and effort than glancing at submissions and deciding if the subject of the photo is a human.