For some reason as soon as that “identified=false” enters the URL I can’t find anything before 2020. I know there used to be totally unidentified observations that were a decade or more old because I’ve ID’d sponges that way before (if I remember correctly). If an observation has any kind of ID, I can easily find them going back over a decade. What’s going on here?
I get about 300: https://www.inaturalist.org/observations/identify?per_page=60&order_by=observed_on&order=asc&identified=false&d2=2020-01-01
They might mean “date added” before 2020, in which case I don’t get any results added before Oct 5, 2020 (by sorting “date added > ascending”).
Since literally any ID takes an observation out of “Unknown” / “identified=false”, I’m thinking that maybe anything older than that may have gotten cleaned up by someone? I don’t know if there used to be so many that that’s impossible or something.
1 Like
Well that’s what it looks like, and if so that is the absolute worst possible solution to a long-standing problem on the site, which has had many good suggested solutions.
I’m a bit confused by what you mean here; what ‘solution’ are you referring to?
1 Like
If that’s what indeed happened to cause all unidentified observations before 2020 to become unaccessible. Still trying to figure out what the cause is. But if the admins just chopped off everything before 2020 that hadn’t received an ID that would be awful.
that is not what @cigazze is saying, and is not what happened. Their hypothesis was that an iNat user/users recently went through and added IDs to all the old unknowns to help get them IDed, and thus why you can no longer see the old unidentified stuff (because now it is ‘identified’, even if that is only eg a phylum ID)
8 Likes
maybe staff implemented that use of AI on unknowns over a certain age feature request?
nope, IDs were added by iNatters
7 Likes
I misunderstood.
That’s…incredible. There must’ve been hundreds of thousands of observations that got cleaned up in this way. And it must’ve happened within the past several months. Was there some sort of concerted effort?
1 Like
@lotteryd ? With help I assume.
1 Like
when was the last time you checked? i assume you’re looking only for verifiable observations because there are a ton of unverifiable observations still.
…
over the years, i recall there being some concerted efforts to clear Unknowns (by adding high-level identifications). so i’m sure that over time, the oldest identifiable observations would get worked out of the system through these efforts.
there was a new Data Quality Assessment item added earlier this year (i think) that allows folks to identify observations which have different organisms captured across different photos and make them casual. there seems to have been a small effort after the addition of this DQA item to move some of the lingering Unknown observations out of the verifiable set.
it looks like a bunch of the unverifiable observations were updated in early 2023, and that seems to coincide with the creation of a “No evidence for organism” project. i’m not sure what the lore behind that project is (or why that would be necessary), but the project owner (@jeanphilippeb) might know why this was created. maybe it helped to streamline efforts to move things out of the verifiable set somehow?
3 Likes
The project does not add an ID or remove obs from the Needs ID pool.
A human iNatter must do that - but the project is a way to find problem obs.
Random example I checked - date and location are wrong, but the iNatter chose to tick ‘no evidence of organism’ - which does not apply.
Most of the correct obs are variations on photos of photos, or signage.
But for Porifera there is literally one obs.
Yep, before CNC24, my 2024 project has been to clear the pre-2024 Unknowns by month. Right now I’m in Aprils and I’ll be there for a bit yet because of old CNCs.
One minor issue I’ve run into is an objection during CNC24 from a “sea things” person, that my slapping Life id onto sea-things Unknowns is “useless”. Not in my workflow. I encourage people who know beaches and oceans to go through some of those localities, to advance those coarse ids farther if they can. Thanks for help there!
3 Likes
Yeah I don’t see any harm in doing that. All it means is that you have to search for un-ID’d observations by “unknown category”. (That also includes viruses and bacteria, but I’m sure there’s some way to limit it to just Life + Unknown that I don’t know about)
The url in my prior post already excludes the viruses etc. Here is a just-Life version without Unknowns (that is still pre-2024, but not limited to the Jan-April window I’m working on).
Categorizing things as “Life” is not helpful. It actually makes things even harder to find than if they had no ID at all. I was wondering if that was what had happened to a lot of these observations. Better to call everything a coral or sponge. Life is so high level it won’t be normally picked up by anyone.
It seems to me that if IDers of various taxa (sea things, plants, etc.) keep asking people not to add broad IDs to unknowns, at some point we end up with a problem – nobody identifies unknowns because they don’t want to get yelled at and then the pile gets so big that it is no longer practical for the specialists who wanted things left at unknown in the first place to pick through the pile and find the observations they are interested in.
Would it be helpful if there were a “sealife” project that identifiers of unknowns could add observations to? i.e., if people put them in an appropriate project, it would pre-sort the sealife observations for those identifiers who can tell their sponges from their corals from their anemones.
This would seem to me to be more productive than arguing about whether the observations should be left as unknown or ID’d as “life”.
6 Likes
I don’t think anyone is actually doing that. I think that’s the perception of identifiers that want to leave everything at unknown. I am a specialist of marine invertebrates and I’m telling you guys openly and directly, yes ID things at higher level taxa, because otherwise they may never get identified. But don’t put them at “life”, because not only is that useless, but it’s worse than having them labeled “unknown”. I can search unknowns. I can’t specifically search “life” outside of the very poor identify tool.
And the solution to this issue has been known for years, the administration just hasn’t implemented a solution: either have the computer vision auto-ID observations with no ID or simply disallow users from uploading observations without IDs. I’ve never uploaded observations, so I don’t know how unintuitive the observation upload process is, but it’s clear enough people are putting their ID’s in notes, rather than any identify field so clearly it needs to be streamlined. Every time the issue comes up though, a bunch of observers complain they can’t upload 100 observations with zero ID and the issue gets swept back under the rug for another year.
Hello! This project, and a few other “blue” projects (see this umbrella project), is needed in order to be able to filter out the observations that are not candidate for the “yellow” projects (details here). Without the “blue” projects, downloading and checking observations for new candidates for the “yellow” projects would result in downloading again and again the same observations that are not candidate (making the whole process fail). iNat does not provide any API filter to rule out these observations, but a project containing these observations can be used to filter them out.
4 Likes