I was out in the field during this change and finally had time to read through this thread. A few comments, though I understand that this amounts to shouting into the wind.
I point this out time and again, but people die. Or get busy with a new job or family. Or are jerks. And then their observations stop responding, and the citizen science aspect of their contribution withers up. Unless iNat starts taking this kind of suggestion seriously:
Hear hear.
Here’s a real-world example of why I’m frustrated by this new change: I’m an entomologist wrapping up a manuscript where I was trying to highlight how iNat can be a force multiplier when experts help curate the flow (i.e., crowdsourcing data collection means the only expert in the country who might recognize a species doesn’t have to be everywhere at once to physically bag it). Luckily, I’d already transcribed the dates for the ~50 observations of the species I highlight in my paper. I imagine that a lot of them are now obscured and show up just as Month Year. Which for an insect researcher, is not very useful-- many insects only have an emergence period of a few days or weeks per year (as @nlblock mentioned), and knowing when in the month a bug came out can make all the difference in timing your survey to find it. Or think of how hard it’s going to be to find statistical support for climate change pushing emergence dates if half of them are reduced to month blocks. Or how does this affect the quality of the data exported to GBIF? I think I’ll tamp back some of my encouragement in the paper.
Anyway, I think this is a discouraging move on iNat’s part, and to ride my own recurring hobby horse, I see this change as yet another example of iNat seeing itself as a social media platform first and leaving citizen science to pick up the scraps. Museums do not strip the dates off specimen labels to protect the privacy of the people who donate data to them. iNat is not a museum, I know. It still hurts.
it doesn’t affect the quality whatsoever. All of the unobscured metadata relating to time/date get exported to GBIF. Here’s just one example to illustrate: https://www.gbif.org/occurrence/2429339266. No need to tamp down your encouragement :)
EDIT: whilst the location is indeed obscured on GBIF (that was my mistake, I looked at the wrong thing), the date/time are not obscured
that’s not what i see. the location in GBIF looks obscured to me. even if other data in GBIF are not obscured right now, i wouldn’t doubt that at some point in the future, additional changes to the geoprivacy functionality will close down on ways to get at this kind of stuff. (in other words, this set of changes probably did not affect the GBIF interface, but this was only the first set in what i’m sure will be many changes related to geoprivacy.)
I’m not going to get into whether or not many of the instances of obscuration are justified or reasonable. I suspect many aren’t. As others have said, observers may have many motivations for obscuring their observations. By the same token, experts who provide help with identifying observations also have their own motivations for donating their time. In my own case, it’s largely transactional. I feel that in return for all the work I put into ensuring observations are correctly identified, the data should be available for my own project and for others. Because we provide prompt and thorough vetting of all butterfly observations in Ontario, we get a lot of submissions - there’s a bit of a positive feedback loop. That’s all great, except that I have to spend a LOT of time checking observations, providing corrections, and explaining those corrections in hopes that folks won’t make the same mistakes in the future. When observations are obscured, this becomes more difficult. The obscuring of the location often isn’t a big deal - we can usually tell if a butterfly is in range or not even when the location is obscured, but obscuring the date complicates things. So now we have to spend a lot of time contacting individual observers to ask them if they will give us permission to see their obscured data. Even if we do that, it raises a different problem. For our project, data is not anonymized - the Atlas shows individual observations. Exact locations are not shown, but dates are. Butterfly observations are of limited use without dates. So even if we can obtain the correct location/date information, we are now confronted with the problem of how we can use that data. I don’t want blowback from an observer because we revealed information that they wanted to remain obscured. With over 30,000 observations per year coming from iNat, we can’t curate observations individually. To my mind, using data that has been obscured by the observer has simply become too “risky”, regardless of whether the observer has given us permission to see the unobscured locations/dates. If we can’t use the data, why should we be spending precious time on these observations? Up until now, the project has been inclusive, and other members of the project team still want it to be as inclusive as possible. I have recommended that our project become exclusive, and that it only include observations from observers who have been invited to join the project, and have accepted our conditions (real names provided, reasonable accuracy on locations, permission to see obscured locations, etc.). In addition to “covering our butts”, this will cut down on the number of observations we need to review, and perhaps the task will become reasonable (right now, it’s like trying to drink from a fire hose).
Hmm…now that I think about it, instead of trying to convince the other project members that we need to make our project exclusive, maybe I should just create my own project, make it by invitation only, and require anyone who joins it to agree to my conditions. Then, I only need to look at observations that will ultimately be useful to us. It would be like an identification service that folks can subscribe to, but to do so, they have to agree to a set of conditions. Problem solved!
Yeah. Obscured locations are about as good as specimen labels ever were up to 40 years ago (when we started putting lat/longs on labels), so when I see a location fudged by 20 miles that’s par for the course. But even the old labels that just said state or township did usually have the date!
Yep. And iNat wonders why they don’t get more expert buy-in to help them fix the tricky groups…
Not sure if this has been suggested, but it seems like having an option to obscure the username who posted an observation would cover a lot of the concerns about stalking/harassment and interpolation for poaching. You could still have the initial ID be identified to their username if they want. Obviously that makes some things like life lists not work, and nothing short of not posting an observation will ever guarantee that it cannot be interpolated somehow, but I’m not sure there is any other possible obscuration step that would be as high impact at making it harder.
That’s a reach to say ider problem has something to do with how platform decides to obscure data, if date is all that you base your id on, maybe it doesn’t have enough base to be added at all? I see end of July shown for August observations, so it’s probably a couple of weeks from original date as max.
Iders shouldn’t expect something more than community taxon and appreciation from what they add, it’s a platform for all and not for science per se, it shouldn’t be thought about simply as a database for your next paper.
Making this change optional would solve most of problems with it.
There are at least 3 additional relatively easy things, possibly more (which for obvious reasons I wont list here) that could be implemented to further secure the backdoors/loopholes which render the current methodology flawed and easy to defeat.
Certainly not the only one, just the highest impact I can think of. I can think of other things which mostly involve applying some kind of distortion to the photo, although that is sort of getting into fairly niche highly dedicated interpolator scenarios and some of them would meaningfully impact the usefulness of the observation. Also my area is pretty mountainous and in a lot of cases the bounding box is just not big enough to really obscure what trail the observer was probably on, especially trails running along canyon bottoms.
the biggest issue with anonymization, i think, has to do with the way attribution / licensing works in the platform. since you can’t properly credit an anonymous observation, i can’t think of a reasonable way to anonymize observations in iNaturalist unless there are are very fundamental changes to the way observations are recorded and licensed.
so while i like your thought process, i’d be surprised if anonymization was part of future changes related to geoprivacy.
The existence of these back doors and loopholes makes me think that the iNaturalist help page on obscuring observations is highly misleading. https://www.inaturalist.org/pages/help#geoprivacy Who has access to these back doors and loopholes? Staff? All curators? Additional people/organizations?
For some species, date can be one important piece of information that is used to distinguish two (or more) very similar species. Location/habitat can also be very important. Considering how crummy a lot of the photos are, we need all the help we can get. If we have location and date, we can usually pull the rabbit out of the hat. Hide them, and we’re hamstrung.
Note that wanting to use iNat data for science isn’t necessarily for personal benefit (and one could argue that using data for a published paper benefits everyone, not just the author of the paper). The project I work on is a publicly accessible Atlas. We combine the iNaturalist data with data from other sources, including old observation data going back over 100 years. It’s available for everyone, and it’s partly because of it that we are able to do a lot of our ID work on iNat so well. We already have a great deal of data about ranges and flight seasons, and we use that data to assess new observations on iNat. New observations help to fill in the gaps in our Atlas data. So it’s very much a two way street - our Atlas helps improve iNat, and iNat helps the Atlas. Both benefit from the free flow of information. Obscuration is an impediment to that flow. Working around it has been a challenge, and each new “enhancement” makes it more onerous. As I said earlier, I think the solution (for me) is to strike a bargain with observers. They provide data for the benefit of the community at large, and I help them with IDs. It seems perfectly reasonable to me, considering I get no direct benefit out of it at all.
I like your idea and also the whole thing of making observers knowledgeble how their actions affect what they upload, there’s always will be people turning off commuity taxon without reason as well as those who obscure locations or make them private, because that’s what uploader asks them - if location is private. But again with making all of this optional you’ll also have all those who wants to participate but not disclose precise location.
I haven’t seen too many instances of folks turning off community ID. As I understand it, this will result in the observation becoming “casual”, and we already exclude all the casual observations from the data we get from iNat. I guess it could mean I’ve wasted time trying to ID an observation that will be subsequently excluded, but as I said, it rarely seems to happen with the observations I work on. That’s my main concern really - wasting time on observations that nobody can use. There are only so many hours in the day that I can devote to this, and I need to pick my battles. I’d like to be able to put my time towards the observations that, at the end of the day, are going to be of most use to everyone. Currently, I don’t know if there’s an easy way for me to filter for only observations where geoprivacy = open.
No, it won’t become casual, it just means you need tons of people to id it to change the id it has if is wrong, often it has the right id, but if not, it’s just really painful to move it out of the map as you can’t just vote in DQA to make it casual unless it has enough ids to mark it as good as it can be.
Oh, maybe it just doesn’t go Research Grade if you opt out of Community ID? That has the same result from our point of view. We only use Research Grade observations in our Atlas.