Observations without media should not allow IDs by other users

Platform(s), such as mobile, website, API, other: All

URLs (aka web addresses) of any pages, if relevant:

Description of need:
There have been a few flags recently about users who add large numbers of agreeing identifications to observations that have no media. Although it does not have an effect on the observation itself, which stays casual, it seems to be a poor-faith attempt to gamify the leaderboards and increase ID numbers in a way that few people will notice it happening.

Feature request details:
Similar to this suggestion: https://forum.inaturalist.org/t/prevent-dqa-voting-when-pre-requisite-conditions-arent-met/49859

In order to add an ID to someone else’s observation, there should be a prerequisite of some evidence to base it on. If an observation has no media, only the uploader should be able to create an ID, since there is no way for others to know if they did or did not see what they claim.

By no media, you mean no photo, no sound, no drawing?

Without linking, what might an Observation contain, then?

I have corrected the IDs of casual observations a couple times when I know that their ID is geographically impossible and often misidentified from a much more likely species.

11 Likes

I’ll link to an observation of my own I just created. Last night as I was getting ready for bed, I heard a great horned owl hooting in the tree outside, as it usually does. I didn’t bother getting a recording or trying to find it to take a pic - but for my own record-keeping, perhaps in the future I would want to remember where and when I heard it. So I can make an observation like this: https://www.inaturalist.org/observations/254245054

It has no pictures or sound, it is just my own ID, and the time and location. It’ll stay casual forever unless I add some media to it.

7 Likes

I admit I’ve done that a few times myself - but I still think it’d generally better if others can’t add IDs. Since medialess observations don’t impact anything like range maps, there’s no harm in them existing even if incorrect - and it’s often just as likely that the observer’s ID was correct but the location they input was wrong. If you want to let them know, it should of course remain possible to leave comments.

3 Likes

Maybe you could add a comment such as:

1 Like

I certainly see the logic in this feature request. Thinking about the scenarios where I have come across observations without media, they are:

  • Real errors (the majority), most of which seem to happen when the user’s mobile app doesn’t finish syncing content to the iNat servers. I try to leave a comment for the user so they’ll realize there’s an issue to fix. In quite a few cases, the CV icon shows next to the ID, indicating that the user did have a photo, but it didn’t make it into the iNat record. On occasion, I’ll be confident enough that the ID is incorrect that I’ll add my own ID as a correction. For example, the observer may have chosen an ID that is not plausible in this location, but has a common name very similar to one well known in that location. Or the observer may have chosen a CV ID that is commonly confused with a plant known from that location.

  • Intentional medialess observations, like the example @graysquirrel gave. If this seems to be the case, I don’t bother to interact with the observation. When I add observations like this I try to make sure I add a note making clear the lack of media is intentional.

  • Very old observations from when iNat used to link to Facebook photos that may have received multiple IDs before the media elements were later deleted

Looking through medialess observations that I have reviewed, I did find at least one that appears to be intentional with a confirming ID from an iNat curator who may also have been present at the time, so that might be an edge case to briefly consider.

Trivia: The oldest observation on iNat (by Ken-ichi) is of a cultivated organism with no photo or sound, and eight confirming IDs.

5 Likes

So we have a couple scenarios I can see of varying degrees of plausibility:

1.) ID which is clearly impossible. In an extreme case, say I upload a medialess wild passenger pigeon observation, annotated ‘alive’, dated December 2024, geotagged to Boston Common. Can we really not vote that that ID is inaccurate? Most cases won’t be that extreme of course, but sometimes it just feels like a rule of never voting that medialess observations are inaccurately ID’d can be a little overly cautious.
2.) Observation that IDer was also personally present for. Maybe an unfair advantage for that particular IDer, but it doesn’t take it RG or anything
3.) Observation with text description so detailed that it is possible to add an ID from the caption alone, such as
a.) a report of taste or smell (e.g. ‘I crushed and tasted the leaves and they had a distinct garlicky smell and taste’ could be enough to ID to at least genus allium, maybe even a specific species given range and phenology information).
b.) a description of visual features so vivid it amounts to a sketch. Or there could be a
c.) a DNA barcode is pasted into an observation field (leaving aside the debate about how DNA barcodes should be used on inat in general)
4.) Observation that originally had a picture, that has since been deleted by the observer; what happens on a technical level in that case? Does it remove the IDs when that happens? Or are they just frozen forever as they were when it was deleted?

In any event, I think it does little harm to allow IDs on medialess observation, and it is perhaps not a good use of time/developer effort to worry about preventing ‘gaming the leaderboard’ by adding such IDs, because the observations won’t be RG. I think some of the statistics don’t include IDs for casual observations by default anyway, though I’m not entirely sure which ones.

7 Likes

I think they’re frozen. Looking through no media obs based on another thread, I’ve come across IDs I made on observations that now have no media. On some of these my ID was the first suggestion made. Since I don’t ever suggest IDs on observations without media, this seems to indicate there was originally a picture that I based my ID on but it has since been deleted. Apparently nothing happens with the IDs in those cases, and I found at least one observation were this was the case and it was still RG.

3 Likes

I see the issue, but I’m here to say that I have both given and received legit IDs on observations with no media. For example, we were both there for the observation, but neither of us was able to collect any confirmatory media. One of us posted it and the other provided a confirming ID on iNat. Please don’t block this.

5 Likes

In this case, an observation without media shared by several users should allow all users sharing the observation to identify it:

Share observations between users

2 Likes

I’ve seen a couple of observations without media but with extensive comments on how they IDed the organism, and I still think it would be very helpful to be able to add IDs to these sightings. If their comments are good enough to support the ID (or good enough to suggest a better alternative) then I would like to be able to add a confirming ID, especially if I consider the observer to be someone trustworthy. It doesn’t matter much to others because the observation will remain casual, but as someone who checks through casual observations for my own research it’s helpful to have them sorted still.

If the issue is that people are using these to gamify the leaderboards, perhaps a better option would be to not count these observations on the leaderboard?

4 Likes

why is that a problem? If you are an active identifier, you quickly realise who racks up ‘empty’ numbers (while I am not sure they have the knowledge to support their numbers)

I have one case, where I got a PM. Please don’t encourage … their IDs are enthusiastic but uniformed and ‘we’ work hard to sort out their problems. But those obs have photos with adequate / good info.

A better solution for medialess obs would be to force the observer to add notes. Hello? No picture or sound - can’t upload without a note - what did you see / hear? In one of those flashing red boxes like no date / location / ID / media

Or is that already so? I have never had reason to try. But I do hear invisible owls in our carob tree …

2 Likes

I support this feature request. One case in which I think allowing blind confirming IDs can be an issue is when:

  • observations are uploaded without media (which happens somewhat frequently due to connection issues with the app)
  • identifier blindly agrees without evidence
  • observer then goes back and fixes observation or adds photos
  • The observation is now RG without any identifier seeing the relevant evidence, and any mistake is less likely to be corrected due to identifiers looking at RG observations much less.

On a side note, iNat intentionally doesn’t allow observations without media to reach RG, so I personally don’t think it’s justified to add IDs based solely on text descriptions by observers. Anyone can write anything that they want - there’s no guarantee that it is correct. A comment suggesting an ID would be more appropriate in these situations in my opinion. Of course, an observer may be correct with their initial ID and text description, especially if they are an expert, but as identifiers we are not supposed to rely on our perception of another user’s expertise when making our IDs, only our own evaluation of the evidence in the observation itself.

5 Likes

For what it’s worth, eBird observations have roughly the same standard of evidence as iNat casual observations (except with some oversight for anything that gets flagged as unusual), and they’re treated by GBIF as equivalent to iNat’s RG observations.

3 Likes

I am aware of this. However, I don’t think it’s a good reason to ID in a way that goes against iNat’s guidelines. Data users expect eBird data to be generated according to eBird’s guidelines and iNat data to be generated by iNat’s guidelines. They can then make informed decisions about whether and how to use that data.

3 Likes

Also, I see plenty of observations without media that are straight-up imports of eBird lists to iNat. If iNat changed the way these would be shared with GBIF they would all be copies since they’re already there from eBird.

2 Likes

Duplicate entries from different database sources are common in GBIF.

As to the feature request, since there isn’t a way to vote against them, I’ll say that I’m in favor of the status quo for the multiple reasons listed above.

6 Likes

Older observations with now-missing photos may result from the Facebook-decoupling I mentioned above.

If you’re seeing no-photo observations still listed at RG, I’d suggest adding and then removing a DQA vote in order to force RG to be recalculated. I think that should be enough to cause the observation to become casual.

1 Like

Not saying I disagree, but we have traditionally accepted field sketches, which have the same issue, as verifiable evidence.

@tiwane:

Field sketches are accepted. It should be one you made while observing the individual that the observation documents. Please don’t add drawings from memory or drawings of photographs of the type of organism you think you saw.

Another relevant thread by @david99:

When identifying organisms (mostly birds, for myself) I sometimes come across observations where the photo itself is unidentifiable (due to distance or camera limitations), but the notes associated with the observation add some insight into the probably ID of the organism. Sometimes the photo is just a landscape photo, with the notes that the organism in question just flew off, and a brief description of the organism.

Should identifiers take these notes into account when identifying organisms?

1 Like