Step 1: Go to observations (not sure if it is all, but I’ve seen a LOT of way off IDs today, so it is happening I think on many).
Step 2: Suggestion is not close at all. Clicking Compare and visually similar (and clearing the incorrect ID appears to work, but suggestions is not working).
Step 3: I wouldn’t think much about it if it was one user or just a couple of obs, but I encountered way more than normal that are way off this afternoon and evening (I spend a lot of time IDing and this has not felt or seemed normal today).
In the above photos, the bindweed normally gives me the correct bindweed…especially if there is a flower (but it is not) and Cakile always shows up as Cakile (it’s not in the list at all). Again, not just these two, but many, many obs today.
For the first observation if you include suggestions not nearby then Cakile maritima it is top of the list. If you look at the GeoModel for the species …
you will find that location is not covered by a hexagon. I think the issue with these hexagons (and the rather simplistic niche modelling based only on altitude as an environmental predictor) has been discussed before.
I uploaded 20 or so observations today and the suggestions were either very wrong or not even allowing me to change them. I had to submit some as “something” and go back to edit them later. I thought maybe I completely forgot how to use the suggested species field. I did not take screenshots, though.
See the long discussion here (including whinges from me - since the 60km hexagons mean I need to ignore the Geomodel ‘anomalies’ for our Cape Peninsula Endemics). The Geomodel is one more iNat tool but we need to use it thoughtfully.
I get a few obs to clear each day with the filters set correctly. Change the location to suit your purpose. The anomaly may be invasive, or captive / cultivated, or flat out human error - you chose the wrong taxon, or an Endemic which is NOT an anomaly at all.
Thank you. This is a problem (it explains, though, so thank you). When people upload their observations and they are traveling to an area or do not know about the plants of an area they are often taking the suggestions and they are very wrong. I still think something is not working correctly. Today I just got a gorse observation that is suggesting juniper and that never happens (and this is GORSE in BROOKINGS, OREGON…an area filled with gorse).
It seems like a switch was flipped in the last day or two that has things all of the sudden not being suggested that normally would. I get what you are saying. I’m just adding that something might have been changed in the last day or two that has broken something that used to work a bit better.
Thank you. I wonder if something was run or adjusted in the last day or two that has now moved things normally in a range out of a range because the system is definitely not working as it has in the past in many of the areas where I ID. I only ID on the West Coast of Oregon (a small range)…so these are plants that normally would get correct suggestions (even if not the first suggestion at times, the plant would be there), but now it is not even in the list, so people from out of the area traveling here are accepting suggestions that are way off.
It is unfortunate. Thank you for explaining this and for sharing the link.
I get what you’re saying, but I had an image of a turtle out of water and it suggested beaver. It was very clearly a turtle. No way at all it could have been a beaver.
This (what blinque said)! I just got an email from someone who is a pretty strong iNat user who is helping me at an event tomorrow where we are teaching people to use the app (these will be very new and some first time users of iNat). She is concerned because today it returned “Trillium” as a suggestion for an ant and poison oak as a suggestion for Glehnia. We are worried about our event tomorrow and what is going to happen with the app as we are trying to usher in new users.
I have the same issue since a few days. It only concerns the app (android), at the website everthing seems to be fine.
Installing the last update didn’t change anything for me.
Before, the iNat results have suddenly been much more reliable for nearly a week or so than those from the competing apps I use normally because ot better results for plants. With beginning of the City Nature Challange, this has changed awfully. :(
is that starting from Unknown - or with insect at least ? (I can tell a Trillium from an ant, but I’m lost with the other pair) Good luck.
PS perhaps the glitchy app explains all the beaver obs in Ostrava. I found - one blurred live obs, one construction with chewed tree, and 3 dead beavers - beyond CNC. So the ID is possible - but not on any of the landscape with trees CNC obs.
That would seem to be a separate bug than what is reported here. If you can document and fill out the template, it would probably be best to submit as such.
Are the links and photos I provided in the initial post helpful or do you need additional information? I will say that at our event today, we only had one strange suggestion (poison oak suggested for Angelica Lucida…which is interesting because my friend had poison oak suggested for Glehnia yesterday). But other than that, all of the suggestions came back looking pretty good.
I did not do enough IDing tonight to see if things still seem amiss. We had three local experts in our outing today (including myself) and the other two said they had felt that things had been off for the past couple of days as well…seeing a number of uploads with IDs that were way off the mark.
If the above links and photos are not helpful enough, I can see if my friend has some I can share as well.
The issue I pointed out for the first observation also affects the second. It is not covered by a hexagon, and if you relax the ‘only nearby suggestion’ then C. soldanella is first in the list. So the two photos/links provided do not suggest there is an issue.
As cooperj said, the issues in your original post are not due to a bug, but due to a known deficienty with the geomodel (and any model will have deficiencies). I’d be interested in seeing the photo that was IDed as Trillium that was a concern.
FYI that I in my browser I just went under “Compare,” Visually Similar and cleared the Taxon just to see if Trillium would appear and it does, but farther down in the list. I attached a photo of that as well.
I just wanted to add that I’ve done a lot of IDing last night and some today and have not had the same issue where it is suggesting things that are not even close without including the correct item in the list at all. I’ve also seen less IDs from observers that are way off (not even close in looks, etc.), so maybe the “glitch,” if there was one, has worked its way out?
I can replicate the Trillium suggestion when I put that photo in the uploader and give it those coordinates. I just looked at quite a few pacific trillium photos on iNat and am not seeing anything like this ant photo. Closest would be insects on the light colored petals, eg https://www.inaturalist.org/observations/356321537
Not sure what’s going on there. When the next CV model gets released this month, I’d love to try it on this observation again.
