For that reason I will vote mine alive, if someone votes them dead. The only exception being the ones found dead and set later.
2 Likes
Would you be OK with a “killed by =humans” field in that case, or what other strategy would you recommend for marking a “dead right now in the photo that someone is looking at” type observation?
I would prefer some other method. All of my pinned specimens were either found dead or I kept them as pets until they died. I put the data in my observation to the date collected and the location collected.
That would be fine with me, though I can not really understand the issue.
What if you are documenting a vulture eating roadkill, or flies on a dead animal? The organism the observation is for isn’t dead but it could still contain potentially disturbing images
8 Likes
I’ve been wondering this as well when I’m doing through the observations of vertebrates that have the word “dead” in the comments but lack annotation. A good number are of scavengers eating dead things (or of birds on dead trees, but I digress).
Also, welcome to the forum!
1 Like
This is a great question, and I think the solution depends on how/if this is implemented. I wonder if if the objective of obscuring all dead-annotated organisms is too narrow, and instead we flag only particularly disturbing observations. This way the feature is subjective, community-driven, and optional. Ultimately the goal should be to connect more people with nature, and I think that having some sort of disturbing content banner/mask/blur will encourage more engagement on iNat, and more sharing of that type of content.
5 Likes
Looking at it from that perspective, an Observation Field might be the way to go instead of the annotation route. Does anyone know if “objectionable content” type field(s) already exist that could be used?
2 Likes
Yay I think I just found one much better than “killed by” for relevant specimens: “Pinned/Mounted?” https://www.inaturalist.org/observation_fields/13139
4 Likes
That is much better!
among the other posts in this thread, i think i like the path that you’ve described the most (assuming they start down the path of censorship at all). i think it’s important to judge how a viewer might feel when viewing an image or observation rather than to judge the content of the image or observation alone. going further down that path, i think instead of “potentially objectionable” (this might cause you to be bothered), i would use a phrase more like “discretion advised” (another person thinks you might be bothered if you view this).
i think a model for how to go about deciding which observations / images get censored can be found in movie / television ratings systems and YouTube age restriction guidelines. you can define broad categories of things that we think other people might not want to witness – gore (ex. guts of a roadkilled dog), distress (ex. sounds of a frog being eaten by a snake), violence (ex. a photo of a chainsaw going through a tree), sex (ex. close-up crop of duck genitalia in action), language, etc. – and then rate based on those broad categories, perhaps on a gratuitous > some > none scale, and perhaps with an optional short description of why something was rated a particular way.
i think this has to be its own mechanism, not just something piggybacking on observation fields, tags, or even flags. (i think flags need to be reserved for things that violate terms of service.)
i think this, and the guidelines used, will be the among the toughest things to work out. eventually, you could employ some sort of AI to help initially rate observations, but i think even in that world, an appeal would always fall on a person or a group of people making a judgment. so you have to have very clear criteria that are hopefully broadly accepted by the broad user base, and you need to make sure folks are applying those criteria consistently. you might be able to have the community do an initial assessment, too, but i think there needs to always be a mechanism for appeal that would require a special group of folks making a final rating.
…
a few other thoughts:
- i think there are things other than images that folks might not want to encounter (sounds, text, etc.). i was originally thinking you might have a case for censoring individual photos, but i think it’s just simpler and more encompassing to censor an entire observation if any part of the observation is rated.
- i wonder how this will work technically, considering that iNaturalist data flows to so many other places? do you censor only in iNaturalist (web + apps) and send uncensored stuff to GBIF, etc. (assuming that they will have to censor on their own)? or do you hold back censored data? similarly, since folks can view the website without signing in at all, do you apply censorship by default (you must sign in to see this), or do you keep things uncensored by default? (i wonder if there are even legal / regulatory requirements that indicate that iNaturalist must restrict certain kinds of content? it would be much more complicated if you had to censor things differently based on the location of the observation, or the location of the observer, or the network affiliation.)
- i don’t think it’s appropriate to censor broadly based on taxa alone. it might be a useful separate feature to allow a specific user to specify their own list of taxa that they want to auto-censor though.
3 Likes
True, but so what? The organism was probably alive and soon dead on the same date, so neither category would be wrong. And we know very well that anyone wanting to use iNaturalist data is going to have to check it over.
6 Likes
I’d like to point out that what is actually shown when you mouse over the “Dead” annotation is this: Organism is dead or shows signs of imminent death. I would argue that the fact that the specimen is now dead and pinned means that at the time of observation it was in danger of imminent death.
6 Likes
yes, indeed! Although I do think they mean from non-anthropomorphic means or at least not intentionally… ie naturally on the way to that state. A picture of a wild deer just before you shoot it would still be marked alive, whereas a deer struck accidentally by a car and suffering fatal wounds would be marked dead. That’s how I interpret it…
3 Likes
Yeah, I know that’s what’s intended. I just wanted to bring up the point that in both situations the animal was alive at the time of observation, but dead soon after. So does it really matter what caused the imminent death? It depends on what the annotation is meant to represent. To expand on your example, what if your hunter friend brings home a deer he killed, and you take a picture and post it to iNat? Should the observation be marked “alive” and set at the time and place the deer was killed, like an observation of a pinned insect? Or should it have the location and date you saw the deer, marked “dead” and “Captive/cultivated”? Just some thoughts to consider.
4 Likes
Well, removing observations reduces the amount of data used and this reduces the precision with which you can estimate the phenology of a species for example. If you work on a large scale analysis looking at phenology of insects, there is no way that you can check each and every observation you use to determine if an insect is marked dead because it is pinned and if the date of the observation can be used. Of course, you still have to be aware of all the problems with the data and you need to be very careful in how you use and interpret it.
Another example could be herbarium specimens. Suppose you upload an observation of a pressed plant that was in bloom at the time of the collection. You would probably put the location and the date of the collection. It would make sense to also mark it as flowering, because it was at the time of the collection. You could mark it as not flowering too, because it is certainly not flowering anymore in your picture of the specimen, but then you loose all the information about the phenology! In fact, doing this would suggest in the data that the specimen was not flowering at the time of the observation, which is bad for the study of its phenology. The dead annotation should work in the same way I think.
But, this is a bit getting off topic ;-)
6 Likes
In the case of the deer, you could choose to use the location and the time when you saw your friend’s deer, but then you would have to mark the observation as casual and the observation would not be of any use to learn about the distribution of deer in space and time. It would be much more useful to use the location and the time when the deer was shot to learn something about deer distribution.
That seems perfect to me - simply a statement of fact leaving the conclusions to the viewer.
How about cases, where the first image is live and the second pinned? … or where the second one is “extreme close-up of genitalia” 
as in https://www.inaturalist.org/observations/69462345. Admittedly the first image on that one is pinned making the situation more simple, but it might as well not be.
Personally I don’t see that much problem with what happens AFTER the subject is dead. How it dies is another matter entirely.
I agree with you about how a herbarium specimen should be treated. I personally would mark an insect as alive on the date it was collected and killed because I think that is more useful. However, I don’t think the people who mark it dead are wrong.
iNaturalist data is wonderful in many ways but full of pitfalls. One need not check every record, but one must check every record that looks odd! iNaturalist saves you the time and effort of getting all those records but not the rest of the boring difficulty of checking.
6 Likes
Yes, but in a lot of cases it would be really difficult to detect oddness without looking specifically at each record. There is nothing odd about an insect marked as dead. However, before doing any analysis, you definitely need to know enough about iNat data to be aware that a given percentage of pinned insects are marked as dead even though they were alive at the time of observation.
3 Likes