Description of problem
I’ve noticed a handful of recent observations being uploaded with the observation date Dec 31, 1969 7:00 PM EST (or whatever the timezone offset). I suspect it is due to particular Samsung models or app version.
This is generally a sign of a null or zero value in a date that counts from the Unix Epoch time. I started notifying people of their inaccurate date, and often pointed them to the photo data to get the correct date/time. While doing that I realized every single on of these I’ve looked at have this in common: They are all taken on a Samsung phone. Looks like mostly Samsung Galaxy S10e.
So I started looking at the model and software version, here are some:
This is beyond me at this point, but I think it is likely a bug for some model or software version of iNat on Samsung.
Steps to recreate
I have not personally had this issue, but some people seem to have reoccurring issues. @andrew54 and I were messaging and they indicated it happens inconsistently.
I had begun to go through these observations, add a comment about the wrong observation date/time, requesting it be corrected and voting “No” on the DQA “Date is accurate?”
But is there some way on iNat’s end to correct such observations en masse by reading the actual creation date rather than that erroneous GPS date?
This data set with erroneous observation dates is screwing up some historical overviews of observations I’m trying to compile.
The number of erroneously dated observations is substantial and probably increasing rapidly: https://www.inaturalist.org/observations?on=1969-12-31&order=asc&place_id=any&subview=table&taxon_id=48460&verifiable=any
The above search (as of 1/3/21 1:30 PM CST) finds nearly 3,500 observations with bad dates. Probably only the first half dozen of that set (earliest ones by upload date) are correctly dated Dec. 31, 1969. All the remainder appear to be fallacious, even though metadata on many of the 2019 uploads do not include the source equipment or model.
That’s what I’ve been doing for a small subset of observations of interest to me. I may also add a polite comment under the observation. A large proportion that will remain uncorrected are those by casual iNat users (students or CNC only) who use Samsung products but haven’t returned to iNat.
I think I also noticed a spike of unintended dates for “January 1, 1970”. I assume staff has the tools to at least recognize these likely bogus observation dates? I would be interested in seeing a small list of the most frequently seen erroneous dates.
Is there any opportunity to semi-automate this process (there are still thousands of mislabeled observations, many of which are at GBIF) or assist users in any way?
In particular it can be a frustrating issue for users to fix the date since it’s not possible to type in a date on Android.
Since I can’t determine which dates were caused by the bug and which were intentional, I can’t automate the fix for everyone. I can fix observations for individual users if they request it.
Would it be possible to at least automate posting a comment requesting review, but only on posts with the Unix Epoch as observed date AND where the photo’s metadata datetime is not the same?
It doesn’t seem like there’s a ton of these anymore, but there’s still some that haven’t been corrected, and many haven’t had votes on Date Accuracy or comments to review the observe date.
A quick “Explore” of observations on “December 31, 1969”–one of the known erroneous Unix dates–yields just under 300 observations and I only see one or two that might be legitimate 54-yr-old observations. https://www.inaturalist.org/observations?on=1969-12-31&place_id=any