Obscuring observations now obscures the date of comments and IDs

[zdanko]

I guess some silent update must have recently been implemented.

Comments and identifications of other users (on observations of taxa where the location has been obscured) no longer show the timestamp (i.e. 1d ago, 1min ago, etc). Instead, they show the month that the comment was written in, with a little eye symbol, that, when hovered over, tells you that the “date [was] obscured to protect personal or organism privacy”:

I don’t understand this. The dates of these comments and identifications have nothing to do with the actual record, so why are they ‘obscured’? I think this is unintentional.

What seems intentional is that the date at which the record was observed and submitted is also obscured:

However intentional this may have been, I am still confused. The dates now only show the month and the year. Why? I don’t understand why it is beneficial to remove mention of the day of the month that the record was observed.

Obscuring (including taxon obscuring) is usually used to stop random people on the internet from coming to a private location or going somewhere to collect an endangered organism. It’s also used if you just don’t want people to know exactly where the organism was found. I can’t think of a similar reason why one would want to obscure the date. Especially the date submitted: why does that even matter? Perhaps obscuring that is also unintentional?

So why was this added? Are there people who like this feature? If so, can we make them a separate option to obscure only the date?

Please fix the comments (and date submitted?) issue, as I believe it is unintentional. If not, I will have many more questions.

Edit: This is intentional and there are many good reasons to implement this. I am no longer of the above opinion.

[thebeachcomber]

@zdanko

I don’t understand why it is beneficial to remove mention of the day of the month that the record was observed.

@zdanko

Obscuring (including taxon obscuring) is usually used to stop random people on the internet from coming to a private location or going somewhere to collect an endangered organism. It’s also used if you just don’t want people to know exactly where the organism was found. I can’t think of a similar reason why one would want to obscure the date.

To answer this query specifically: say I go on a walk, and observe 10 things. One of them is a rare species, and gets obscured, the others are common things. To figure out the location of the rare thing, someone can go to that user’s observations, match up the date/time from the obscured observation with the non-obscured ones, and deduce the location.

Of course this system is not perfect, and there are still a number of ways to guess obscured locations, but this change is just one small incremental improvement, and iNat is continuing to work on ways of improving the system.

See the updated explanations at https://www.inaturalist.org/pages/help#geoprivacy

[thebeachcomber]

this is not a bug, so I’ve moved the topic to general

for future comments (by anyone), please keep them related to the mechanics of obscuration, and don’t conflate this with which taxa get obscured and why (which has been covered on many other threads)

[zdanko]

@thebeachcomber

To figure out the location of the rare thing, someone can go to that user’s observations, match up the date/time from the obscured observation with the non-obscured ones, and deduce the location.

Ah! Perfect explanation! I now understand (at least that part).

I believe this is the most important excerpt from the updated help page (reference for others):

• Interpolating from nearby observations: If an obscured observation was made at a similar time to an open observation made by the same user, the true location of the obscured observation can be guessed. iNaturalist avoids displaying the exact date of obscured observations publicly but this information is still available in the API responses and nearby observations can be inferred by searches, the observation_id, or even just making guesses about a user’s behavior including the places they frequent. We are working on structural changes to make interpolating locations more difficult, but in the meantime we recommend not posting any open observations nearby in space and time to where your obscured observation was posted.

[zdanko]

@thebeachcomber

this is not a bug, so I’ve moved the topic to general

The bug report isn’t for the date of the observation being obscured, it specifically addresses the dates of comments and IDs being obscured as well. I don’t think that that being the case is explained by the help page. As that is the title and main topic here, I think this should be in Bug Reports.

[thebeachcomber]

It’s intended for those to be obscured as well, as they can also provide info on the obscured location.

For example, go to https://www.inaturalist.org/observations/87821078. The date/time of the first ID (mine) matches the submission date/time since it is my observation, and I made the ID in the upload screen. Someone with malicious intent can apply the same process as above, and use my other observations with the same date/time for the first ID (assuming I uploaded them in a batch) to guess my obscured location.

[zdanko]

Hmm, okay, this fits #general then.

What if staff make it so that the first ID is obscured but others aren’t?
Edit: Or maybe make the first IDs all made within, say, a day of uploading obscured? Or would that defeat the purpose again?

The reason I ask this is that it is excessively difficult to read the comments of an observation when they all just say Jul '21 and I have no idea if it was yesterday or two weeks ago.

[thebeachcomber]

@zdanko

What if staff make it so that the first ID is obscured but others aren’t?

Subsequent IDs by other users could also provide clues. If an obscured observation is something like a bird, it is highly likely it will get ID’ed within a few minutes/hours/within 24 hours. This then gives an indication of roughly when it was uploaded (of course assuming you’re not uploading an old photo from months/years ago), which then facilitates interpolation. That is my take on that specific aspect, but you would have to ask an iNat staff for confirmation/clarification.

@zdanko

excessively difficult to read the comments of an observation when they all just say Jul '21 and I have no idea if it was yesterday or two weeks ago

As an aside, and not saying this isn’t a valid concern/question, but why is it necessary to know this information?

[zdanko]

@thebeachcomber

If an obscured observation is something like a bird, it is highly likely it will get ID’ed within a few minutes/hours/within 24 hours.

That’s a good point.

@thebeachcomber

As an aside, and not saying this isn’t a valid concern/question, but why is it necessary to know this information?

I’m not sure if others have this problem, but when I’ve read comments on obscured obs recently, I just get kind of confused because there isn’t much of a timeline to follow. I like to know how old comments are on observations when I respond to them. Preference, I suppose. Also, since my own comments aren’t obscured, the time looks kind of weird, but that’s also just a preference issue.

[onyxrat]

If a user takes three observations in this order: open, obscured, open, and then another user looks at the first user’s list of observations, will the observations still be listed in the order they were really taken? If so, someone could guess the location because it is bracketed by the two open observations.

Update: I just tried it. Yes, you can still guess the location of an obscured observation if it is bracketed by open observations. Maybe iNaturalist needs to pretend this observation is taken at 12:00 am on the first of the month when sorting the list in order to better obscure the location.

[thebeachcomber]

there are multiple other ways of guessing as well, but the obscured dates/time change is designed to add an extra layer of difficulty for the interpolators, not necessarily be an absolute solution

[mws]

If all the observations made within the same obscurity square and within 12 hours in either direction of the obscured observation were obscured, would it matter what the times are? The times are only useful for matching up to the locations, so if there’s no locations to match up, the times can be as open as possible

[thebeachcomber]

@mws

If all the observations made within the same obscurity square and within 12 hours in either direction of the obscured observation were obscured,

the issue is that this will create a huge number of needlessly obscured observations of common things. Many users make several hundred observations in a day, so I don’t think obscuring all of those in addition to the 1 or 2 rarities in the batch would be the best solution

[mws]

But the date obscurity will make it impossible to count any obscured observations in bioblitz projects, won’t it? And it’s also still pretty possible to match a date-obscured observation to a date by other means. If a rare dragonfly is obscured and the observation photo shows the observer holding the dragonfly, you can guess that the observation was made on the same day that the observer caught a bunch of other dragonflies. A rare alpine plant would presumably be from the same day as common alpine plants. The only way to be certain that there’s no way to match up an obscured observation to open observations is to make those open observations obscured

[thebeachcomber]

@mws

But the date obscurity will make it impossible to count any obscured observations in bioblitz projects, won’t it?

This is a good question to which I have no idea what the answer is, will have to wait for staff to answer.

@mws

And it’s also still pretty possible to match a date-obscured observation to a date by other means.

Yes for sure, but as I mentioned a few times above, this new feature is not designed to be a blanket solution, but to provide an extra hurdle to get over

[mws]

It feels like another hurdle isn’t that useful if we already know that the most reliable solution would have to rely on obscuring the locations of the observations near in time. At least, I can’t think of anything with fewer loopholes, but I could’ve missed something big

[cthawley]

I get the rationale for this change (although it’s not that recent I don’t think, has been out there over a month I think). However, there are still numerous ways around it. If someone is already on iNat and has the expertise and desire to use the breadcrumbs approach to find locations, I think that they’ll still be able to track down those locations fairly easily in most cases.

My own guess is that the change won’t do much to prevent determined people finding locations and isn’t worth the decreased site usability/data usefulness.

[trscavo]

@thebeachcomber

for future comments (by anyone), please keep them related to the mechanics of obscuration

What is supposed to happen if the location is missing? (It seems a missing location has the same effect as an obscured location.)

[trscavo]

@cthawley

I get the rationale for this change (although it’s not that recent I don’t think, has been out there over a month I think).

This new feature was rolled out yesterday: https://github.com/inaturalist/inaturalist/pull/3142

[thebeachcomber]

@trscavo

What is supposed to happen if the location is missing? (It seems a missing location has the same effect as an obscured location.)

Interesting, do you have an example of this? I’m looking at an observation atm with no location, and the date/time haven’t been obscured