Identifiers as Volunteer Curators, and Obs with Probs

I would find it helpful to have separate icons.


Thank you for bringing this up again. Unfortunately requests to change this dynamic have been shot down quite some time ago and there doesn’t seem to be any appetite to move:


I think another (two-part) solution to the multi-species obs would be introducing a new flag that IDers can set, which might even automatically insert a comment with instructions. The second part of the solution would then be authorizing curators, admins or whomever to split the observation into reasonable subsets e.g. 3 months after flagging and no action by the observer, keeping metadata and observer intact and add a very visible notice that the observation was split administratively and by whom. Since this would be a rather invasive approach, it could only be done after the rules of the site are changed, to not infringe the rights of observers, i.e. only for future observations.

1 Like

That’d be ripe for abuse and conflict, and I’d strongly oppose any move in that direction. Let the community (curators, whoever) give their input, up to making an observation casual, but please leave the observation as such intact. Even if it doesn’t meet any particular quality standard.


Is it really that bad? I also walk thru unknowns, trying to give coarse IDs that aren’t wrong (as in 0 legs=snakes or snails, 1 leg=fungi, 2 legs=birds, 3 legs=martians,… 8 legs=mostly spiders, 14 legs=baby butterflies; weedy stuff: ignore unless I can give the family or better). My main approach is to strive for speed so the real IDers get as much as I can produce. When in doubt, I opt for the coarser choice because I fear I might miss the notification about a disagreement and then the poor being is stuck in limbo by my fault.

I see the occasional Agobacterium or supposedly virus-infected thing, and rare streaks of 30 micrographs, but not so much that it disturbs me.

1 Like

I would also oppose this option.

For one, I agree it’s ripe for abuse. The observation is fundamentally the property of the observer. I think having power to hide things is where the line for curator power should be drawn - editing users’ observations is too far for me.

And two, as a curator, there’s already too much to do and not enough time. Being asked to fix multispecies observations would be a huge time suck. In my opinion, not a good use of time.


Indeed I myself am not in favor of changing the “can be improved?” DQA line. I think the “can be improved?” question was created to be used for situations in which an observation of a single organism also happens to be a taxa which cannot be identified to species from the photos. I think the question works just fine for that case. I find it justified that two identifiers need to be present to use it for that case.

I think we additionally use the “can be improved?” question for “photo dump” situations (the many wildly unrelated taxa in many photos situations) simply because there is no other tool to use. For that use I think there is no reason to ask for two identifiers present, and requiring two is setting us up for conflict and irritation. Although I did not explicitly say it in my previous post, I am more in favor of creating a new line, “do all photos contain the same subject?”


That just makes your situation all the more unfortunate.

1 Like

Thank you.

I think there may be an error in your html or something, because the link at the top of the page is not clickable, although the text is there if one scrolls down. And there’s a bit of extra verbiage at the bottom that is repeated from the previous FAQ entry.


If photos were viewed only in connection with the observation, this would not be a big deal, because one can see an explanation about which photos are what. But this is not necessarily the case – photos may be viewed or distributed in other contexts (taxon pages, downloaded/shared with other sources) with the ID that is assumed to be correct. If it is not obvious (like a giraffe being labelled a rose), then you end up with mislabelled photos that are not recognized as such.

Thanks. I hadn’t considered those cases.


Ok. I’ll modify my behaviour. But, experience tells me that we’ll lose some (not much) valuable distribution data for central African butterflies.


If we could, sort of keep up with Unknowns, while the observers are still active, we could hope to get the observer to respond to requests to split / combine / delete.

But if I check for recent activity most of them have gone, dormant. Others, despite being active, never respond to notifications. They may simply be overwhelmed, since we lack management tools.

If we had new labels - instead of Casual (an unkind word choice disparaging a ‘good’ obs)
Duplicate to delete
Multiple to Split please
Multiple to Combine please
then both identifiers and observers would know where they stand, without reinventing the wheel for each subsequent - why is this STILL in Needs ID - identifier.

1 Like

“Ripe for abuse” has been brought up a couple of times by different people. There are many other things which could be abused here: add a wrong id (wrong selections from the list do occur, so it there is no proof of intent), disagree with an item in DQA (date, place, wild, evidence), mark “as good as possible” while at family rank or worse, and likely even more.
It seems to me that such abuse hardly ever happens.
Hence a “split it up” request button (or similar) available to all identifiers, with curators or the observer being able to perform the split on click of a button (that could be a valuable menu entry for the observer anyway), would likely not cause a lot of abuse, too.


Aren’t those just different aspects of the same process? To answer your question, it is common in UK for specialists to work through museum collections to check the identifications and extract the distribution data. Anyone producing a distribution atlas for, say, a family of insects Britain and Ireland is likely to have visited the main natural history collections or got a helper to do so. I’ve done a little of it myself, and yes, you do find errors. The museum curators will have expertise in some groups of wildlife but not even the London Natural History Museum has a specialist in everything, so on the whole the museums welcome this type of outside help. And if a family has not been reviewed in this way for a few decades, there will have been taxonomic changes that need bringing in, which are updates rather than errors.


Abuse of all features on iNat is fairly rare but it does occur, but takes a decent amount of curator time to deal with.

Any feature can be misused unintentionally or intentionally. The difference to me is that the examples of functions you provide range from 100% essential (adding IDs) to important (DQA). I do not think the ability of curators to edit observations would be an important/essential feature, so the tradeoff of the downsides is not worth it to me

That said, I think it’s probably best to focus the discussion here more on the existing system and

You could make a feature request for what you propose, or I am also happy to split out the posts about adding editing functionality for curators into its own thread if folks want to continue that conversation.


I would not be willing to split someone else’s obs.

Only the observer knows

  • oops that one shouldn’t have been on iNat at all
  • that one belongs with my Tanzania holiday
  • this one is from the zoo so captive exotic
    and and and. Also the location could vary widely across habitat, or even geography. And the date and time.

I think I’ve suggested this on a previous thread somewhere, but there might be a less heavy-handed approach to flagging multiple-species observations (although I don’t know if this would be possible to program in the iNat backend).

Create a new DQA checkbox, something like “Conflicting evidence of organisms?” (i.e., the evidence provided with the observation suggests more than one taxon). Checking this off would turn the observation Casual, etc. However (this is the architecture-dependent part), if the media attached to the observation is changed, that would trigger the automatic unchecking of that property in the DQA.

My rationale here is that most of us are very conservative about using the “cannot be improved” checkbox, because that observation will probably never be reviewed again. Of course, if it’s never fixed, we want that to happen and not waste the time of a succession of identifiers; but if the observer does come back and quietly fix it without adjusting the DQA, we’d like it to resurface.

Most observers with multiple conflicting photos will probably not fix them (so their observations will stay “Casual” once marked); the few that do respond and change their photos will automatically come back into the identification pool again; and the very few of those that have changed media but have not resolved the conflict can pretty quickly be checked off again to return the observation to “Casual”.


Sorry, I meant that to be a rhetorical question. It is obvious that museums don’t have the staff to go through their entire collections to check for accuracy. So, I don’t see the difference between data on iNat and data in a museum. There will always be mistakes.

1 Like

In general, specimens have usually gone through at least one set of IDing when they were created by the collector (and often at least a quality check when accessioned into a collection for the first time, though sometimes this would be by the collector themselves).

While I agree that both sources can have mistakes (and there’s research showing that for some groups, iNat observations are IDed just as accurately as specimens in NHCs) - I think that there are a whole host of differences between physical specimens and photo IDs - they are complementary but very different.


1 Like