I’ve had a bit more of a think about this and generally I agree that being able to add agreeing IDs to an observation without media is a bit pointless, but I don’t agree with removing the ability to add a disagreeing ID to these observations.
If something is clearly wrong regardless of whether there is any media associated with it, then we should be able to disagree, at least to start a discussion with the user who posted. I have come across several observers who simply don’t reply to comments but will respond very quickly to disagreeing IDs (I’m not sure if this is something to do with their settings?).
First and upfront: I think observations without media are perfectly fine. When looking at distribution maps, they often fill in gaps and give a more complete picture. There are many citizen science platforms that have specific approaches to validating such datasets, and some of them share observations without media with GBIF (eBird is one example, Observation.org is another). iNat has decided to set the bar so that only observations that include media can be validated and shared with GBIF, and that’s fine with me too.
Given iNat’s current validation system, I see the point that adding matching IDs is pointless, as @matthew_connors said, so I’d be in favor of removing that option only.
I think there isn’t generally much point in adding IDs to unverifiable observations, but disagree that the option should be removed. Here’s what I’m seeing after reading the discussion here.
Pros:
Prevents manipulation of leaderboards using one method.
Prevents incorrect RG status if someone agrees to an ID of a medialess observation, then media is added later.
Cons:
Prevents correction of obviously wrong IDs, such as those that make no sense for the given location.
Prevents IDs from other people who were present for the observation.
Prevents IDs made based on other information, such as a description of appearance or smell, from the observer.
Regarding leaderboards, those aren’t really meant to be relied upon for anything important, so it doesn’t matter if they aren’t perfect. A better solution for the problem referenced here would be to exclude IDs on media-less observations from leaderboards, but I’m not sure if that’s actually a useful thing to do either.
The second problem, brought up by @cthawley, is a real concern. However, I don’t think it happens often enough to warrant a protocol change. Considering the number of observations that are falsely RG due to more prevalent problems (like CV mis-IDs plus blind agreeing), the few observations affected by this issue are only a drop in the bucket.
In my opinion, the number of harmless or slightly useful IDs that would also be blocked by this proposed change outweighs the number of harmful ones. Why shouldn’t someone be allowed to add an ID to an observation that they were also present for, or correct an extremely improbable ID even on a medialess observation? Sure, there isn’t enough evidence to definitely confirm or deny anything, but that’s why these observations are Casual. They can’t become Research Grade anyway, so there’s no harm in making an ID if you feel confident in doing so.
Finally, I don’t think this is a big enough issue that the site needs to be altered to fix it. Neither the pros nor the cons are very important, as none of these situations come up often. It’s better to leave things as they are than put more unnecessary restrictions in place.
I agree with essentially everything here, but I will add that the problem brought up by @cthawley would be much better fixed if a notification was given when the media on an observation was changed, which I am pretty sure is already a feature request
I’ve read all sides on this issue thus far, and I’m not swayed from my initial opinion: iNat observations should include either visual or audio data. Other platforms can function in different ways (I also use eBird), but part of what makes iNat observations robust and reliable is that ‘real’ data is available. Even if there is a photo of a tree, with a note saying that a California bluejay flew off, but it was there before the picture was taken, I will not ID this. I go to Data Quality and I give Evidence of Organism a thumbs down and move on, add a comment saying that, unfortunately, there is no bird visible for us to ID, ‘but would you like to leave this observation up so that someone can ID the tree?’, and then move on.