Should drawings be RG, casual, or evaluated by supporting info. case by case?

Current iNat curator guidelines consider drawings in obs. to be verifiable/RG. But the following suggest there’s some complexity in or lack of consensus in if this should remain the same:

  • Although guidelines approve drawings, their first reference caveats “unless someone repeatedly posts such content,” and the second discusses what to do if there are reliability problems.
  • Most iNat areas besides curator guidelines only refer to photos (above all else) and sounds.
  • Forum topic discussions on drawings and evidence expressed some views that drawings lack the verifiability of photos, although others felt otherwise.

My view is drawings may make most sense to become categorized automatically by iNat as casual, for similar reasons non-media obs. are. Some have argued even no-media obs. can be evidence (e.g. if including written description, observer’s comments, or taking into account observer reliability). So, the issue isn’t whether drawings are evidence, it’s if they meet the standard of sufficient evidence RG should use.

RG translates to sufficient evidence to transfer to GBIF (iNat has previously clarified it’s not intended to differentiate obs. ‘value’ in any other way). There is a question of if all drawings being verifiable lowers data quality. With drawings (vs. photos) it’s less possible to know an organism was observed (like it is for non-image obs.), or that an organism wasn’t drawn in a non-misleading or limited way, even accidentally (which could in theory even apply to some scientific illustrations). Drawings aren’t comparable to blurry photos, since those can be ID’d to broad taxonomic ranks or marked no evidence of organism in DQA if needed. (edit- photos so blurry nothing is visible)

Update: my current view is the original question remains worthwhile to discuss. But since some regard some drawings to have sufficient supporting info. to ID (e.g. written description), I’d be open to the compromise that all observations (photos, sounds, drawings, and no-media) begin as verifiable by default, and then are sorted by community ID and DQA votes, which can be chosen freely by users.

3 Likes

They’re legit RG observations.

8 Likes

Observations with drawings (instead of photos) can become RG. You’re right, many people do end up downvoting in the DQA. Usually what happens is other people will upvote and leave some comments explaining that drawings are allowed.

8 Likes

Update: the following comment (and some comments above) are from another topic which was split and unlisted, which no longer perfectly relate/fit to the current topic:

@brennafarrell @marina_gorbunova Posting the drawing isn’t the subject of the current topic, and I didn’t clearly state whether drawings must be RG or casual grade. I just noticed related past topics about drawings and what is evidence, and there looks to be much lack of consensus there. For that subject I’ll add a most recent post to each of those topics, and suggest those would be a better place for if there’s more discussions on evidence.

iNat staff allows it, there shouldn’t be any consensus from users really.

2 Likes

I’ve personally seen so many instances of drawings being faked or abused as evidence, or even innocently given biased detail because of the observer’s opinion of what they thought they saw, that I struggle to trust them for research grade.

But the site rules say it’s acceptable, so that’s the unbiased answer.

6 Likes

i haven’t observed this, and also isn’t it easier to just fake an observation with a stolen photo than to draw a fake observation? That seems like a lot of unnecessary effort. Of course some drawings are not possible to identify to species, but i don’t think iNat needs to be an art critic website. :rofl: I also feel like it’s a pretty rare case for it to come up.

7 Likes

A careful drawing which shows diagnostic features - yes - much better than a blurry photo. A drawing is a greater investment of time and effort, than click click.
I haven’t ever seen a drawing on the obs I ID, only on discussions here in the forum. As Charlie says - drawings are rare.

I never listen to audio, but that doesn’t mean I don’t think they earn Research Grade.

2 Likes

I really don’t know where you find all those drawings to start with and how you find out they’re faked? There’re x10 more stolen photos obs than all the drawings uploaded.

4 Likes

I agree one should be cautious about confirming an observation with only a drawing as evidence, especially for a rare species or one seen out of expected range. But I wouldn’t go as far as to prohibit them from being research grade. Sometimes a drawing leaves no room for doubt. If an observation in Tanzania drew an elephant, there would be no room for doubt over the species observed (to give an extreme example).

If the fear is about deliberate falsification, then that can just as easily be achieved by stealing pics from the web and posting them as your own sightings, which I have no doubt happens. I’ve not seen evidence of anybody going to the effort of doing a drawing to falsify a record, although I appreciate there is some scope for false memories or wishful thinking to creep into an illustration.

3 Likes

One issue for drawings is there’s no way to know an organism was even observed. I don’t think this is comparable to photos. Curators routinely find copyrighted photos and remove them. The fact that photos could be faked doesn’t justify trusting all drawings. Drawings by their nature aren’t actually recorded using an instrument (camera, recorder, spectrogram), so at least in some ways or circumstances there’s far less initial reliability/verifiability.

The second issue is many problems associated with trying to ID based on drawings. Specifically, we’re discussing whether drawings should be RG/verifiable, which mostly refers to species IDs (and some genus etc. IDs). If a community ID is only as fine as family, setting DQA to “community ID is the best it can be” will change the obs. to casual (and so no longer what we’re discussing). At the genus and species level specifically, there are numerous ID issues with drawings (described in the body text).

Lastly, I made the comparison to how non-image/no-sound obs. are currently considered non-verifiable. Even those could be argued to have some reliability to them, but not enough. Likewise drawings have some, but not enough. RG only means they transfer to GBIF. I doubt even GBIF would want drawings. Has anyone asked them?

4 Likes

The forum is a place for curators, mods, staff, and users with none of those affiliations to discuss all parts of iNat, including proposing changing or re-evaluating existing policy/guidelines. By “consensus” I mean the existing topics on this subject reached no consensus (including by some users close to the team or admin decisions of iNat). By reaching consensus I mean all of the above kinds of users (not only users who are non-staff, non-mod, etc). That said, staff continue to make final decisions. So, no harm in discussing.

3 Likes

I agree that audio can receive RG, because there’s more verifiable evidence. As I explained in the body text and some comments, there are basic verifiability/reliability limitations for drawings somewhat similar to the ones obs. lacking any image or sound have. Secondly, there are many problems with determining if and how drawings can be identified, including for some technical illustrations even. I don’t think a comparison exists between drawings and blurry photos. Blurry photos either can get some taxon ID (even a vague one), or if entirely blurred will receive DQA: no evidence of organism.

1 Like

Not easily, which is part of my wariness around them. In many cases the observer has admitted to it, eventually. Sometimes it’s easy to tell because people copy a photo from the top of search results on google images, in a very specific pose and no other sketches or field notes. Other times people say they weren’t sure but they “thought” it had these features. In minimal cases photos or other follow-ups showed something that had features unlike the drawing, which confirmed the observer guessed or again was biased and assumed it was X species and had X features rather than what they could actually see. And the number of times people on outings with me have been absolutely sure they saw one colour, or one feature that the organism didn’t have is more than I’d like to ignore. There’s enough motivation for people to “cheat” just to have their sighting verified and justified. Photos can be faked, but drawings are much more easy to fake and even harder to disprove.

Field sketches are very, very difficult. It’s a level of detailed observation that’s really unlike anything else. And so I only trust a very small number of people with them.

12 Likes

Considering a lot of databases including GBIF seem to accept records with no mention of verification, it seems this “proof” verification iNaturalist deals with is a bit unique. Some other sites like ALA take observations on the faith that people are issuing correct IDs, though a large portion of their dataset is sourced from museum collections and actual determined specimens, there’s another portion that comes from random observers. A fraction of those are backed by photos, but not all of them.

The risk of incorrect and false data has been a problem for decades, possibly more so in recent years because of the accessibility of citizen science. It’s meant that there’s a larger number of enthusiastic people contributing, but not necessarily making the best choices and decisions with their data. Note I’m not talking about how professional people are, just the abundance of observers – even experts are lax with their data sometimes.

6 Likes

I agree that GBIF is imperfect. I think it makes sense to set a good standard of accuracy for anything that is in iNat’s control which is being sent to GBIF. And possibly GBIF itself should use higher standards. A difference between various non-iNat unverified records being sent to GBIF and iNat drawings is, a verifiable drawing obs. can change into an ID which the observer didn’t even specify or may even disagree with.

When I was young, I enjoyed drawing or painting pictures from photos - photos perhaps taken by me or my family; and often ones found in the encyclopedia or newspaper or magazine. In fact, my parents did that, too. (We did not know about copyrights back then, but such was just done for our own entertainment anyway).

I think a photo provides a much better verification of observation than a drawing, particularly when submitted with metadata.

While I personally think drawings are iffy evidence of an observation; scientist and naturalists have provided drawings as scientific evidence for hundreds of years, if not millennia. So, I guess the practice is “grandfathered”.

3 Likes

I really have no opinion on this, except to say that when I first saw this bird ( Purple Finch (Haemorhous purpureus) from Riverview, Winnipeg, MB, Canada on September 14, 2021 at 10:07 AM by Ian Toal · iNaturalist.ca) it looked orange, and the shadow (best seen in photo 2) looked like a black mark around the neck. If I had been skilled enough to draw a field picture, it would have looked different than the actual photo.

2 Likes

Sure, you just said it as if it wasn’t true and was still under discussion while it’s up there working.

There’s a whole website working on CV only and nothing it suggests is checked before getting to GBIF, so drawings are the least thing to care about, if they’re working as photos, then they can be forged, it’s easy and we just need to be careful, with any evidence we see, it’s really weird people are doing that though.