Stay with us.
This is how I deal with privacy for obs in my garden.
So far none of the random dots for my pinned location have actually landed ‘here’
https://www.inaturalist.org/projects/elephant-s-eye-on-false-bay
I also choose to obscure my wild orchid obs. And unusual bulbs.
Sometimes, when I am in a field of flowers, or some subject is widely dispersed, I use a large radius showing the species is widely dispersed.
First, there is a difference between accuracy and precision.
I do not have GPS on my camera. I do carry a phone which could give me GPS, however, for the great majority of my observations, accuracy to a small radius is not required, I don’t believe. I believe what iNat is trying to capture is biodiversity, not being able to guide someone to collect specimens - this is not in the iNat guidelines. So if I see an bee on a field, what is the difference if the insect is accurately positioned because it is going to be gone in a few minutes. And sometimes I obscure my observations to protect the specimen or privacy.
i’ve done this too, but it isn’t really something that people are going to pick up on unless you mention it in the comments. It would be neat to be able to draw polygons of populations and do mapping on iNat, but that’s probably way beyond the scope of what the site ever would be.
I’d definitely encourage you to keep participating in iNat. I do think it’s a fair point to make that data is more useful to researchers/land managers/whoever with higher accuracy. I don’t think iNat pushes this on users because high accuracy isn’t central to its mission to increase engagement with nature (this happens more or less regardless of a user’s observational accuracy, I think) - the data is a happy byproduct in this sense.
That said, a big motivator for a lot of folks to contribute to iNat is that they want to generate data that are accessible and might be used to help conserve/understand/protect organisms. So I think it’s good to hear feedback from end users about what they need to use data effectively. iNat users that don’t care about whether their data is used or not can just ignore this. I certainly wouldn’t advocate for putting any accuracy requirement on observations uploads - data users can always filter, and if they’re not cleaning/checking/filtering their data, they shouldn’t be using it…
In terms of practical use, and speaking from the perspective of a user of iNat data, accuracy definitely does matter. Obscured data is used much less frequently in papers (I see it filtered out in methods frequently), and often isn’t captured by downloads/project for specific places (since it won’t register as being in a given place on iNat).
Two common accuracy cut-offs I see in papers are 1 km and 100 m (ie, excluding observations with accuracy greater than those lines). These are somewhat arbitrary, but also sometimes derived from common data resolutions in landscape variables that data users are doing an analysis with. I would guess that data with accuracies less than those cutoffs are more likely to be used by researchers (in general), but have no specific evidence of that.
As a specific example of how observers can take this into account, I “manually obscure” locations at/near my home by choosing a nearby point that is far enough away from my house for my comfort and adding a larger accuracy circle to it that encompasses the true location. I intentionally chose a location where I could set the accuracy under 1 km since I thought this might make the data more usable.
yeah, i do want to reiterate that just because data isn’t useful for fine scale ecological inventory doesn’t mean you shouldn’t add things to inaturalist. One of the main benefits of iNat (and cited as the main purpose) is just to help connect people with nature and help them understand and value nature. Even though I am already an ecologist iNat has been very helpful just in me learning more species, both in my area of focus (plants in my region) and other areas (insects for instance - very important, but i don’t know them very well). If you add something with low precision or even no location at all, it may still help you learn more about the taxa and thus be worth it. The main thing is just that you should make sure to use that location uncertainty correctly so projects that can’t use less precise data can filter it out at that time. I do wish the less precise observations displayed different on the maps. I love perusing iNat maps and it bugs me when an observation pin shows up somewhere way off - like a pin for a species in the centroid of Yosemite Natural Park which really only occurs at the far western end of the park. Most likely the observation was mapped correctly with the correct location certainty mapped, but looking at the range map that doesn’t display so you can’t tell. I think ideal would be to have the less precise observations vanish from the map when you zoom in a lot and only show up further out. But that’s probably too complex to be reasonable.
Anyway, sorry if i sounded discouraging to anyone. Feel free ot keep adding your location-uncertain data, just use those uncertainty buffers :)
Thanks for sharing your garden, so different from mine, half way around the world!
One thing I will do now is review how I have set my obscure location, to see if I can tighten up the size of the circle.
I’ve been searching for orchids- mainly in Portugal - and always obscure those locations. It has gotten almost too disheartening to keep on with it. In 2020 there were large and many holes dug at numerous locations.
Are there species you think should be automatically obscured in Portugal? If so you can flag the species on iNaturalist.
Perhaps for those of us who aren’t working or studying in fields where accurate data is so useful, it is more difficult to keep those needs in mind. It’s not that I don’t care, but apparently I had not learned, or forgot, that I was making my data useless for research by obscuring the location.
Going forward I will use your tip to ‘manually obscure’ my location. Is there an easy way to change the accuracy of observations already uploaded?
As for the topic: how to encourage users… it seems to me that more straight-forward explanation and instruction would go a long way. Clearly what is on the Help page isn’t enough. A link on the Home page would help.
Thanks @tiwane, yes there are, although I have nothing but my own limited experience to justify my opinion. Part of the issue, and not a small part, is that areas are being built up in the name of “progress”.
I wouldn’t say obscured data is “useless”, just less useful/used less.
There is a “batch edit” functionality which I think can change obscuration, location, and accuracy. I’ve never used it so I can’t provide detailed help, but you could try searching for threads about batch edit and seeing anyone has used it for similar before. Here’s one to start with:
https://forum.inaturalist.org/t/is-there-a-way-to-batch-edit-my-old-observation-locations/28996
You could try changing the “true” location and accuracy first and then, once you’re satisfied, taking off the obscuration.
Plus, given how concerned people are with privacy, they may be discouraged from participating simply by being asked to provide a precise location.
Thanks for your help Chris. Spent some time on the batch edit and can see that it’s do-able. Plan to make some of my past observations less useless/more useful.
Wasn’t that (entering fake values for coords) regarded as “wrong behaviour” (for lack of a better term) by some users in another forum thread recently? Even if a poorman’s fix for the broken (i.e. overly broad) obscuration system here.
IMHO encouraging users to go from “accurate coordinates + high precision + obscuration” to “inaccurate coordinates + low precision + no obscuration” has the potential to make things less usable/useful.
Instead of that rare (hence the obscuration) parasitic plant growing next to its host tree, you end up with that plant pinpointed in the middle of a pond +/- some overly large uncertainty. That’s pristine information deliberately removed, now unavailable to e.g. researchers (who could have instead benefited from very precise data by kindly asking for unobscured access).
Some users might not like it, but it isn’t forbidden by iNat’s guidelines. As long as the true location lies in the accuracy circle, the observation is meeting the guidelines as far as I understand it. The way the data works, you’re not really saying that the organism is exactly at the point in the center of the accuracy circle, just within it.
Additionally, I would note that it doesn’t mean that the information is “lost” - just not on iNat. If I ever move, for instance, I’ll probably go back and edit the locations of previous observations at/near my house to the correct one with finer accuracy and then they’ll be available.
A user can also always share true coordinates with scientists or others if needed via other methods - if the observer is being messaged to ask for access then they can reply/share it via DM, email, etc.
I’m not talking about guidelines here, just about data quality.
Well… the way a GPS/Camera works, it IS really saying that the observation is taken exactly at the point (+/- some Dilution of Precision).
Then upon uploading to iNat, unless some tech-savvy user chooses to voluntarily alter the EXIF data (by adding some random value or a systematic bias to geolocation tags), the pinpoint WILL be at the exact EXIF location (the ‘accuracy’ circle/value shown on iNat is irrelevant to the debate, as long as the pinpoint in its center reflects the EXIF location).
And later, there is no easy automated way to alter (by adding random noise, or systematic bias to) the EXIF location in iNat’s database. To replace the 25km-obscuration with a 100m-pseudoobscuration, one has to manually edit each one of thousands of obs + invent some wrongly large “accuracy” to it.
“Lost to iNat”, if you will. Thus giving users the task to have to store somewhere else the precise coordinates, be it at home (oops, hard disk toasted, true coordinates lost forever!!!) or on some other online database (boo, risk of duplicates!!! boo, time lost entering the very same data across several sites!!!)
I know these solutions exist, thanks. That’s overkill and a daunting task to have to maintain the very same data - data already present on iNat in its full glory, with pics and coords - across different DBs, e-mail exchanges, rehosting the exact same photos on FTP etc… just because iNat staff does not see (or understand?) how things could be even better and easier for everybody.
iNat is a pretty tool, but believe it or not, like other tools it has its shortcomings the staff ‘wontfix’… even though not all users’ suggestions or expectations are unreasonable or uneducated ;)
edit: tl;dr: imho, tampering/degrading observation metadata to compensate for the lack of a proper obscuration system on iNat… is not something to be taken lightly.
is an appropriate way to describe this practice. For one, it’s the user’s own data, and additionally, they aren’t misrepresenting it. They may be choosing to show it with less accuracy/precision than other user’s would prefer, but that’s their choice, and they likely have reasons for it. Other users are free to post true locations or use the obscuration feature as they will.
Also, treating a GPS/camera sourced point as a true location is a fraught assumption, and I would not recommend it. There are many articles/links about how GPS/cameras/phones (though mostly cameras and phones) overestimate precision (some on the forum), and the actual location is not included in the precision radius around the recorded coordinates a not-insignificant amount of the time.
This
is not correct, as a batch edit option is available (see posts above in the thread) which many users make use of.
In regards to data usage, in my experience many (most?) end users of iNat data are going to be accessing it via GBIF without interacting with users on iNat at all (or attempting to get “true” locations via projects or messaging). For those users, “pseudoobscuration”, if done with a reasonably small radius (ie <1km), almost certainly makes the data more usable than obscured locations.
So yes, there are always going to be trade-offs with any method for obscuring the data. It’s definitely more work to sharing true coordinates with “pseudoobscuration” than the built in obscuration feature - but I think this approach also increases the usability of the data for many applications and may be preferable for some observers.
I agree with a lot of what @cthawley said!
I am one of those Users “tampering” with each location of my 1000s of observations I uploaded. It’s just a very normal part of my uploading process… it’s not such an unlikely or immense effort as you make it sound actually.
Why do I do this? Well, because my usual photo equipment does not have GPS included (antique, I know). So I do it all by hand.
I recently started using my smartphone to take some pictures as well and started to use the GPS function there… I quickly learned that it might be of use in many cases but is not as reliable as I thought it would be. Suddently I had observations from the other side of a lake 15km away where I never had been… Or I uploaded observations near a waterfall and later by accident discovered that were uploaded to iNat far (kilometers!) away from the river… sometimes I have datapoints somewhere in the woods when I never left the road… I now do use this phone GPS data as a starting point, but I am far from trusting it and will always set the location and accuracy manually to the best of my knowledge (or willingness to share). Not everyone uses high end GPS equipment, and I would not blindly trust the built in alternatives in phones or cams - especially when leaving urban territory
Sorry if ‘tampering’ is not the right word (my English is definitely rusty). Altering? Degrading? Misrepresenting the physical location of the organism in the real world?
As for GPS accuracy, precision, and other geodetic subtleties (I actually posted a link about that above!): I wholeheartedly second that - few people have the tools and time for sub-meter GPS positioning, what I dare call a “true” location for practical purposes: accurate, precise, consistent. I reckon that posting “false” (altered/tampered/edited/whatever, as opposed to “true”) coordinates on iNat is tolerated as part of the “rights of the user over their own data”. Which somehow answers this thread.
As for adding a systematic/random geographic bias to the location of existing observations (e.g. rounding their values to nearest 0.05° automatically; or moving each obs by some random value e.g. [+/-] 25 to 50m on both lat and lon, then adding a 50m radius so as to encompass the initial true location): I haven’t found how to do it on the website. Would someone care to elaborate? Or is it feasible only through the API? This would make converting my thousands of “accurate+obscured+barely usable” obs to “inaccurate+unobscured+very usable” very easy. (This would make disappear the need for a finer obscuration grid, and/or the need for user-level editing of GPS EXIF metadata prior to uploading to iNat).
edit: If the function is already there to batch-round/batch-shift the coords, maybe implementing it on the ‘Add Observations’ page could be convenient? Thus avoiding the extra steps of ‘upload accurate+obscured, render inaccurate by rounding/shifting, unobscure’. Suggestion: a 4th option “Location is to be rounded”, next to “Location is public” “Location is obscured” “Location is private”?
I think it is important to remember the economics of all this.
People posting observations do it for free. Don’t tell them (us!) what is “acceptable”. Some of us don’t know where we are. No map-sense, drifting with the wind for a week, too close to military targets (Putin’s house) where GPS is obfuscated, etc. Hassle us and we’ll stop posting any observations. OTOH some of us don’t want to say where we were. Was I on my neighbor’s land? A top-secret research facility?
OTOH, data-users are getting data for free. And may be doing good things with it! But they can accept that this is not finely curated data but a bunch of casual free labor. Crowd-sourcing! And no control on who contributes. Only a gross cross-check in that if I see a horse and post a cow, someone may disagree and suggest “horse”, or suggest that cows are unlikely in the Mariana Trench. There will ALWAYS be outliers that a good protocol will discount. Cheaper data IMPLIES more filtering and cross-checking.