When Observations are Photos of Photos

I’ve seen an increase in the number of “observations” where a person takes a photo of a photo, computer screen, someone else’s mobile phone, etc. While the species involved are IDable, as a scientist I’m very concerned that this can lead to incorrect location information. Is this something others are encountering? Apologies if this has already been discussed, I couldn’t find any similar topics. Is there a standard reply we should give these observers? As an example, please see this observation: https://www.inaturalist.org/observations/27784695

1 Like

I’ve seen cases where someone took a smartphone pic of his buddy’s SLR camera screen, such as of a telephoto pic of a bird, presumably because the person lacked a telephoto camera. The assumption is that both saw the bird at same time/place, which might not always be the case. If it’s a photo of a Facebook page pic, that is a concern and requires some explanation by the submitter. I’d say any photo of a photo needs some explanation.

2 Likes

If you’re pretty sure it’s not reflective of a real observation, you could go under the Data Quality Assesment and vote “no” on Evidence of organism & Recent evidence of an organism, but I think you’d also want to leave a comment asking about it-- if they respond and say that it’s a photo of something a group they were in saw but they didn’t get a photo, or something like that, and the photo was from the group, you would want to retract the votes against it. That’s what I think, anyways.

3 Likes

Another consideration is copyright violation, especially if it’s obvious that the photo includes some portion of a website or a species that definitely doesn’t occur in the area of the observer. These can often be back-traced through a reverse image search, but it should be noted that this is more likely to be unreliable in cases of photos of photos. In cases of copyright infringement, the photo should be flagged as such (click the “i” icon on the photo in the observation, find “Flag This Photo”, and select “copyright infringement”).

2 Likes

I’ve been noticing a lot of those too.

If it’s an obvious screen shot of a website, you can mark it as a copyright violation. When it’s less clear, I usually check the location it’s marked at - half the time it’s located in a suburban wasteland, and obviously not taken there, in which case I mark location inaccurate and move on. But it’s not always clear-cut.

2 Likes

I have also seen a lot of cases like this, and wondered how best to treat them. The guidance quite clearly discourages using other people’s photos. However, for many of these cases, flagging for copyright infringement just seems too heavy-handed, since the originator appears to have willingly allowed their photo to be copied. The frequently used responses page also has a section related to this issue, but I’m not sure whether the suggested responses fully cover cases like this.

I think it may be better to just make the assumption that the date and location are very likely to be wrong, and vote accordingly in the DQA. It’s probably fair to do this even when the date/location seems plausible, given that the user hasn’t been explicit about the provenance of the photo. In itself, the observation may be valuable, but if the user didn’t make it, it probably shouldn’t be allowed to become research grade so easily. And I suspect that the user won’t really care very much about that, because very often they are just trying to get an ID for a friend who doesn’t have an iNat account.

1 Like

I’m not overly familiar with how the DQA settings work, but in taking @bazwal 's suggestion, can’t someone come in later and change the date/location back to acceptable? I do agree that most of these violate iNat’s terms in that they are not the observer’s photos, but some could be. They also could be observations of a taxon that is found in the area, but the ‘photo of a photo’ is taken some distance away, in which case it still represents inaccurate data. I feel like we either need a standard response encouraging folks to change or take down their “observation” or the post needs to be flagged for all the reasons being given in these comments. Here’s another one from yesterday: https://www.inaturalist.org/observations/30254968

1 Like

Yes - but the downvotes will remain associated with the observation until the voter chooses to remove them, so that does still serve to indicate that the observation has (or had) DQA issues. However, I agree that this is not sufficient. More needs to be done to explicitly encourage users not to use other people’s photos - and also to encourage the other people to join iNat instead of allowing their photos to be copied.

PS: there’s another thread that deals with essentially the same issues that you might be interested in.

Just to show how serious this issue is, the following 12 observations have this problem. These are just in the 10 days since the last response to this post, and this is in one family of beetles. I think this presents a serious data integrity issue for iNat.

https://www.inaturalist.org/observations/30450596
https://www.inaturalist.org/observations/30449590
https://www.inaturalist.org/observations/14885954
https://www.inaturalist.org/observations/30756504
https://www.inaturalist.org/observations/30779643
https://www.inaturalist.org/observations/13875488
https://www.inaturalist.org/observations/30952692
https://www.inaturalist.org/observations/29126031
https://www.inaturalist.org/observations/14332381
https://www.inaturalist.org/observations/31149647
https://www.inaturalist.org/observations/31198232
https://www.inaturalist.org/observations/31202272

I see a couple of themes in the group of examples that you provided - some of the what is this thing and how does it work (first 30 observations) that I have noticed myself. These I feel are the people that it is useful to welcome and point to the website. Learning new things is kind of messy. Time and friendly folks cleans up these messes.

There is also an example of an overwhelming enthusiasm of youth - numerous observations on iNat; activity in the same area on other sites - a situation where mentorship and modelling desired behaviour would probably pay big dividends.

when observers remain engaged with the community over time, there is always the opportunity for their collection of observations to continue to be improved by the observers themselves - comments added to make clear why the photograph was used instead of the organism in the wild; locations and dates corrected manually from field notes; observations that do not fit the requirements of the site removed.

This potential of improvement over time due to the observer’s efforts represents the ascendancy of iNaturalist over mass uploadings of geo-tagged dated images scraped from the web.

Data of all kinds is inherently dirty - there is no perfect collection method free from bias or error. Citizen science like this is a way to collect data for science - and make no mistake, I want my iNaturalist data to be useful to science. At the same time, I feel that thinking of iNaturalist as only a data collection service misses out on the equally important effect that observing and documenting has on the observers.

What if both of these things are important? How would we change our own behaviour to each other to recognize that?

6 Likes

Are you really sure that these are a serious issue? I glanced at a few and they looked like phone pics of a desktop monitor or camera screen. And the top example supports that idea since they later updated the photo. Just because people have a different workflow than you doesn’t mean they did anything wrong. The screen grab is good enough for identification. Maybe this is their way of cropping and resizing the image.

The more serious issue is marking an observation as inaccurate date simply because you don’t understand someone’s workflow (seemed to be the case for several of the examples). Is there something about the date or range that is outside the biology of the species? Asking the observer is an appropriate response. Marking the obs as inaccurate seems arrogant.

1 Like

The issue is that the location accuracy of the observation (I don’t really care about copyright, although I think the iNat admins might). If the location is not accurate, then the scientific integrity of the observation (and by association, iNat) is in doubt. I’ve been marking these as “location is not accurate” to give them an instant death so that they will never be used in any scientific database ever, but the bottom line is that these “observations” undermine the scientific integrity of the process.

With all due respect, how is it up to you to determine if the location is accurate, because the picture is a photo of a photo? That hardly means the person doesn’t know where the original picture was taken.

I don’t know why someone would use a photo of a photo, maybe they aren’t very tech savvy but, that doesn’t necessarily make them liars.

3 Likes

Yes, my question exactly. I wonder if I can revive these observations where I don’t see evidence of inaccuracy. I think I’ll comment in the observation about the possible but unlikely to be malicious issue.

A comment to the submitter, asking for clarification, is better than marking a record as flawed in some way unless there is very clear evidence in the record that the location is wrong, the organism is not wild, etc.

1 Like

I feel like the whole point of iNat is that someone makes an observation of a living thing in the wild, not an observation of a photo they took. It is possible that they take a photo of a photo and then adjust the exact spot they took the original photo in the new photo, but if they are that savvy, why can’t they just use the original photo in the first place? Other questions include, is it their photo? It seems like there’s more assurance if they upload the original.

@karen33317 to answer your question more specifically, iNat has a box at the bottom of every observation that asks anyone if they think the location is accurate, so I guess that’s how I think it’s up to me to determine this. I also think by submitting an observation to iNat, the user is implicitly allowing all kinds of things up to me, such as what the family/genus/species/ is, is it wild, etc. I don’t necessarily think people are lying, but if they are unsure how to use the technology and a record is deemed “research grade” it is eventually uploaded to GBIF and other bioinformatics sites which scientists use in peer-reviewed publications. The utmost accuracy should be used.

One of the core iNaturalist community guidelines is that we always start out by assuming good intentions. I hear your good intentions of wanting the data on iNaturalist to be as accurate as possible. That said, though, good intentions are a two-way street. If you are downvoting Data Quality Assessments based on assumptions of what a photo-of-a-photo means, instead of on clear evidence, then it would be better to follow @jnstuart’s suggestion first, and contact the observer with some polite questions about their photo, location, etc. You might discover that they are a new user still experimenting with how the site works, and would respond to some gentle guidance. Or you might discover that they have a very well thought out work flow, and are in fact providing accurate data. It would be important to know these things before making Data Quality Assessments. The iNat community is hugely diverse and international in scope, and approaches to using the site are just as diverse.

As a scientist and data user myself, I am definitely with you on wanting information on iNat to be as accurate as possible. As the site admins have pointed out before, though, the primary purpose of iNat is not to be a data repository, it is (paraphrasing) to be a social network connecting people with nature. By fulfilling that mission, the site does generate a huge amount of very useful data, and it also provides a lot of tools to maintain and refine its accuracy. By its very nature, however, the data will always be noisy – though I would argue no noisier than the data in your average museum or herbarium database. We would not expect scientists to use GBIF data from those sources uncritically, and the expectation is the same for data sourced from iNaturalist. We are not expected to do that whole job for them.

Hope that helps clarify where some of the other responses to your post are coming from. It’s definitely a worthwhile topic to have brought up, and a good discussion to have, so thank you for posting it!

3 Likes

One possibility is that a submitter might only use the phone app for uploading their pics, in which case they get a photo on their digital SLR camera then take a phone pic of the image on their camera screen so as to upload it to iNat with the coordinates at the location. The camera image gets downloaded on a computer later but isn’t submitted to iNat. We can’t say how every submitter might go about getting their records posted.

I thought the replies on the original observation were good - questioning the observer for more information on the picture. Not threatening, but asking. It appears like the data may not have been downloaded by a serious user. I know I have said this before, but if a person is using iNat data for research, then it is beholden on them to ensure the data are correct. I do my best to ensure the ID of anything I confirm or post is correct but I will, and do, make mistakes. People with less experience with a group will make more mistakes. Any ID’s I make for reptiles or amphibians should be immediately suspect! As for the technical stuff, I have no opinion - I use a camera, edit the images on a PC, then upload them. I don’t know what’s involved with the phone app.

1 Like

I’ve seen several times where the first image is a screenshot of the observer’s gallery app on their phone, with the relevent photo opened. Sometimes the actual photo is also uploaded though. I don’t understand why someone would upload both… I comment on these asking the observer to just have the original photo but I don’t recall receiving any replies.