Platform(s): website for sure, I’m not sure how the mobile app works…
Description of need:
Currently, new users sometimes use the CV before adding the location. This leads to a whole lot of misidentifications. For example an Indian Cormorant in Mumbai was identified as “Neotropical Cormorant” by the observer using the CV. This observation now needs three more identifiers to correct it. It gets even worse when another new user goes ahead and agrees to it just because it looks similar, that would then need 6 more identifiers, (unless the identifier corrected themselves, which most don’t). Don’t we have a shortage of identifiers already?
Feature request details:
While uploading, if the user clicks on the show suggestions species name input field and the location isn’t provided, then it should say something like, “Please provide location so we can come up with accurate suggestions”. Once it is provided, then the CV should work as normal.
small progress. We have a Disagreements filter. Easier to find, and fix, the problems. After the fact.
Perhaps the bigger issue is that many iNat users neither know nor care about - this Pacific snail won’t be found on South Africa’s Atlantic coast, nor our Indian Ocean coast.
Better onboarding for newbies would make those who care - aware that location matters, and gives a better CV suggestion.
This is why the CV distinguishes between visually similar and expected nearby. Sometimes it is helpful to see suggestions that are not from your location, particularly in underrepresented regions.
In theory I would agree with your request. In fact, the CV suggests things from different continents for me all the time, so I don’t see the point. Maybe some day.
I don’t think location necessarily needs to be required for CV to work, though I don’t really see any reason why it shouldn’t be as all iNat observations should have a location. That is the most logical/useful solution.
If that can’t be implemented, I do agree that a prompt when trying to add an ID when there is no location entered yet would be useful. Something like “ID suggestions may improve if you add a location first”. Of course, there are currently a lot of problems with the “Expected nearby” suggestions, but hopefully the geomodel issues that have caused that will be fixed soon.
Would this cause the CV to not work if location couldn’t be loaded? In bad service Inat tends to slow down and only gives CV without location knowledge(even though I know the correct ID without the AI). The AI for me is a shortcut, so I don’t really like this idea.
That wouldn’t help in the situation described in the OP. The Indian Cormorant is in Genus Phalacrocorax, whereas the Neotropical Cormorant is in Genus Nannopterum.
I absolutely get that there are a ton of different possible ways to try and encourage users to enter location before checking CV suggestions, and they come with tradeoffs worth careful consideration and deliberation. That being said, I really wish we had some sort of bandage solution in the interim.
It’s been two and a half years since that reply was made. I absolutely get these things take time. iNaturalist is a very small team relative to the size of the platform they handle, I am not at all condemning them for not coming up with the perfect solution in that timeframe. However I do really wish they just did something to make this happen less, no matter how inelegant or flawed.
Anecdotally: there’s a genus of grass skippers called Lon. A few species look extremely similar, e.g. Lon melane, Lon poa, Lon vitellina, Lon chiapas. Fortunately only one of these occurs in California, Lon melane. However every now and then someone will report e.g. Lon vitellina here, always with the CV badge. Seeing as Lon vitellina shouldn’t be in the geomodel here, I am all but certain this happens because someone checks the CV suggestions before entering a location. On its own this wouldn’t be too much of an issue, but the problem is that when this happens the geomodel starts to expect this species nearby, and suggests it to others. Eventually this happens enough that even slightly more responsible observers will start to think their skipper might be L. vitellina, because there are so many pins nearby.
Even this wouldn’t be an issue if it was just happening with California grass skippers, but I’m pretty sure this is happening with thousands of taxa in hundreds of places. I can’t help but think this has created a fairly substantial amount of unnecessary work for identifiers, which I worry about when I see increasing numbers of posts about identifier burnout here. Every time a fix to this has been suggested, people rightfully point to drawbacks and why the suggested solution is flawed. I tend to agree. While I voted for this feature request I don’t think it’s a good solution in the long run. But I really don’t see how even implementing the hackiest slapdash fix for this would be worse than maintaining the status quo for years.
I wholeheartedly agree with this suggestion and think it would be excellent, but I also see why it may not be an ideal solution and why staff are unlikely to implement it. By removing CV unless a location is given, then you are essentially restricting access to the CV and that’s not generally what iNat is about. I can quite easily imagine there would be situations where you would want to use the CV without inputting a location.
So, I have an alternative suggestion: if the user doesn’t input a location, have the CV still show its best guesses, but disable the ability to choose any of them. That way a user still gets to see what the CV says, and can still put that in by typing it out if they want to, but it’s more difficult to do so, and more importantly it’s more difficult to do that than to just put in a location. This encourages people to add a location without removing any functionality, and I think it is a pretty good solution for both viewpoints. I’m definitely open to suggestions though.
Yes, this is my current approach. It’s a little discouraging that I keep seeing it happen again and again with the same taxon in the same place though…
And the ones I know about I can handle, but it makes me wonder how many more obscure/less curated taxa have this happen completely unnoticed at a much greater scale! I think there are many more people IDing/curating skippers than many other invert families, for instance
Another solution is to have the CV only suggest up to a certain Taxon Rank if the location is not provided; i.e. Kingdom, Phylum, Class, Order, etc. There could then be a pop-up saying ‘please provide location for a more accurate suggestion’.
Well it works vice versa. When I am on a distribution map to see ifThis species is supposed to be found here … if there are not too many - I go and whine - try again - not a South African Endemic in …
When a distribution map looks off, bash it into shape again.