this is what i do. i just manually select Houston, TX, which points to a spot in the middle of downtown, and bump up the range of uncertainty. i know where i live. so if i ever want to go back and change any of those, i can go back to observations at that point and find them relatively easily.
2 Likes
i think you might want to test this out on some real-world cases. i know GPS technology is pretty good nowadays, but sometimes i think even the best non-professional devices might be off just a bit, and it would be unfortunate if that led to a the creation of sort of a donut ring of observations around a an otherwise obscured geofence.
2 Likes
i used to do this for stuff at my house, and it resulted in a smaller uncertainty area (i don’t care if people know I live in the Montpelier area, and everyone who cares knows that anyway, i just didn’t want my address on here and easy to find).I was doing that because the points used to obscure in a circle around a central point and with lots of observations it was easy to see where the center was. I stopped when the obscuring method got updated because i didn’t like creating ‘imprecise’ data and liked to know where on our land i saw things, plus i wanted to be able to share that data with a conservation organization. But… I think the ‘big circle’ approach is mildly frowned upon but not wrong. The problem is it may be harder to filter out when you filter out obscured things, but there are some filters for low precision observations now too.
Personally, I don’t understand the need to restrict the area to “about a football field”, which is apparently about 1.4 acres. All it is doing is auto obscuring observations within a certain range of home, and if we can do that manually out as far as, say, 100 football fields, then doing it in an auto-form to, say, 3 football fields should be fine? Maybe obscuration for anything within 100m of home (effectively 6 football fields) would cover 99.9% of observers… I noticed a while back that here in New Zealand (roughly 38 deg S) the 5th decimal place is roughly equal to a metre, so rather than geofencing with a boundary, just applying the obscuration if the lat/lon matches for up to 3 decimal places! Perhaps have a user setting for “small section obscuration” being the 3dp, or “large block obscuration” at 2dp matching.
I think this is a great start, tiwane. We can see if it is sufficient and look at refinements later if the users want more. Thanks!
3 Likes
This topic is specifically about creating a personal “geofence” and not about the broader topic of obscuration on iNaturalist in general and it seems like your comment here addresses the latter. Please keep comments on topic.
The football field was sort of an example. A little bigger might be fine, but it should still be small. Calculating which observations belong in which place is one of the jobs our system does.
Also, I assume this is implied but the geofenced area should not be a public place, correct?
4 Likes
i agree, not a public place. maybe if you want to get fancy make it shareable with only some people (town conservation commission or whatever) but that probably isn’t needed because you can give them trust to your observations anyway.
3 Likes
I agree… making it public kind of goes against the grain of what it’s trying to achieve
2 Likes
definitely needs to be big enough to avoid the “donut ring” effect. I also would like for more than 1 permissible geofence. My reason for that is again related to the donut ring effect. A very regular, circular ring is STILL going to make for a very obvious “I LIVE HERE” indicator. A complex shape is more likely to avoid the ring altogether, but it’ll also be more difficult to figure out the location being obscured. When I’ve used geofences on other sites, I commonly use a radius of no less than 1/4mi. And oftentimes larger than 1/2mi. I would like to obscure my whole neighborhood, at minimum, since I frequently make observations while I’m out walking the dog. My neighborhood is easily more than a mile across at its widest point.
And FWIW, I currently obscure all observations in my neighborhood manually. It can be annoying at times because I have to remember to change the setting. And sometimes I forget it and have to change it later. Would prefer for this to be done more programmatically. Even better if I could define the boundary with a .kml file instead of setting a radius around a point.
3 Likes
if you’re conscious of the donut hole effect you can counteract it, just obscure things nearby too, you can also cruise the road by your house and add a few things from your yard and your neighbors but not obscure those. or whatever. Once you’re aware of it you can subvert it.
if I have to manually address it, anyway, then the user-defined geofence is pointless.
3 Likes
i would say that it’s almost worse to have a geofence that people might realize later is ineffective. not everyone is going to be as technically savvy or as vigilant about their privacy until it’s too late.
1 Like
by the way, here’s an example of why i think it’s important to look at some real world cases before you go too deep into development. here (https://www.inaturalist.org/observations?place_id=83006&taxon_id=542500) we have 4 observations of what i’m 99% sure is the same cluster of plants in the same 1 sq m spot, but the locations of observations are scattered all over the place. even if you factor in their (in)accuracy ranges, only 2 of these observations even overlap each other. if you look to the east, there happens to be a football (soccer) field. compare that field to the locations of the observations (which, again, are probably all in the same 1 sq meter spot).
1 Like
Having the geofence means you only have to check occassionaly to make sure it’s doing what it should, as opposed to having to manually obscure every observation that is too near to home… so it is hardly pointless!
That is a botanical gardens… I know when I go around a place like that, I take dozens of photos and if my camera doesn’t pick up GPS (perhaps I took the photo before it could lock a location) then it will be up to a hundred metres away from the last successful location lock. Also a possibility, how many of those 4 observations were manual pin placements “from memory” of where the observer thought they saw the plants? Perhaps they wanted the name of the gardens in the location field, and after they typed it, they realised it moved the pin, so they moved it back (not quite to the correct place)… etc
There are known issues with GPS accuracy, +/-20m (although it is often as good as +/-2m) as well as the distance from camera to subject (which for trees/birds can be an additional 100m!). But when you are photographing at home, you tend to be right on top of the subject.
[edit: arboreta, not botanical gardens! Same issues though]
1 Like
a good number of photos that i take from home are taken when i have just returned from being somewhere else. i bet this is true for many people’s observations at home.
look at the timestamps of when the observations were made vs uploaded, look at the photo metadata, or look at what those observers have done in the past, and you can decide for yourself whether those are likely to be manual and/or from memory.
I check the locations when I’m at home all the time, anyway. I have a decent mix of stuff taken with my big camera of birds, from when I’m inside the house where I have to manually mark the location, as well as stuff taken with my phone, either from inside the house or out, where the location is off enough to put it outside the proposed small geofence.
No, I’d be checking the obscuration for EVERY observation even still. Maybe the small area would catch most of them, but I’d still have to check all of them. My property is an odd shape, too, so unless irregular shapes could be drawn/uploaded manually, chances are accurate locations within my yard would still fall outside the proposed small area geofence.
And then of course still every observation when I’m out on my neighborhood walks would have to be manually obscured. And if I allowed myself to get lazy and NOT check the observations at my house, then I’d get into a bad habit and wind up with a lot of observations that SHOULD be obscured but aren’t. And it would happen, no doubt. I know how my mind works. I still say if I can’t make the geofence the size and dimensions I want, then it’s a “feature” that’s not worthwhile for me to even use.
I’m just going to chime in and say that it seems like most of the nay-sayers are people who would not use the feature as described. Development is incremental. Let’s get a basic feature out there and if people have more suggestions for improvement (give me an option to make the fence bigger, let it automatically change geoprivacy settings when you move the fence, change what happens when posts are obscured, etc. etc.) then we can address them later.
The kernel of this idea is a basic feature of many other apps that use geo data. You can geofence on Strava. It’s not perfect, because if people were really stalk-y, they could still find my neighborhood and the ‘donut-hole’ around my house, but if I were that worried about stalkers I wouldn’t use the internet. Can we agree to bump tiwane’s proposal to development and revisit this later?
4 Likes
yeah this stuff is just to make sure it isn’t easy. I have friends who dealt with stalkers and unfortunately with the real bad ones, nothing will be enough. In most states it isn’t hard to get an address for any landowner if you are persistent enough, for instance. All social media with photos might have something someone might recognize, etc etc. And then there are the hackers. So if you have a real serious problem you probably shouldn’t add anything to any internet or social media site from your home. Which is a bummer.
1 Like
Yes. The geofence is just to add a layer of difficulty, so that anyone who looks at my posts can’t just say, “Oh, look, his address is…”
1 Like
the way i look at it, if you’re going to advertise something as a security/privacy feature, it should actually provide security/privacy in practice. speed of development does not take precedence.
1 Like