Dealing with low quality observations and inappropriate content on iNaturalist

As the City Nature Challenge 2018 is beginning to ramp up (Edit: and because iNat was recently chosen as Apple's App of the Day), you may have noticed a ton of new users on the site. You can see the increase happening at the stats page...which is great! But many new users don't know that iNaturalist is a valued resource used by land managers, researchers, organizations, etc. and treat it more like any other social media site. I'm certainly seeing a lot more flags coming in this morning than usual.

Use the Data Quality Assessment section below the observation to mark observations as captive/cultivated, not containing an organism at all, clearly incorrect location or date, etc.

Several people have helped prepare these common responses to inappropriate content & other frequent issues, which include:

  • Not an Organism/Test Observations
  • Observation of Human
  • Multiple Species in One Observation
  • Captive/Cultivated Organism (abundance of or not marked as such)
  • Copyrighted Images
  • Photo Far Away/Not Cropped/Unidentifiable
  • Missing Location
  • Imprecise Location
  • Private Location
  • Duplicate Observations
  • Bad Identifications (jokes or malicious IDing)

I'd recommend bookmarking that link! Do you have anything to add to that page? Let me know... or just add it if you're a curator.

If you see something inappropriate, you can always flag the offending content (ID, observation, comment, and/or photo). A curator or site staff can take a look and hopefully find a resolution. But if you can, try to use one of those canned type responses before flagging the content. I generally try to engage the user before sending a message to Some people do need to be suspended right away; check out the Community Guidelines.

Curators, I use these boilerplate responses to flags suuuper frequently. If anyone reading this has been on iNaturalist for a while and is interested in potentially being a curator to help resolve these types of issues, you can reach out to me or directly after reading through the Curator Guide.

When in doubt or if something is extremely inappropriate and should be deleted immediately, you can always email

Posted on Απρίλιος 23, 2018 0558 ΜΜ by bouteloua bouteloua


Yep, the low-quality observations are pouring in around the Bay Area. This is a fantastic and helpful post, thank you!

Αναρτήθηκε από naturalisttrent περίπου 6 χρόνια πριν

I have two requests for prepared responses:

-observer clicks "agree" to an ID immediately, and it's fairly obvious they cannot verify it independently (e.g. their previous ID was way off, or only at kingdom level). I think there used to be a response for this called "trigger-happy agreeing"? I see this pretty frequently, and since it only takes one person + the observer to bring an observation to Research Grade, it can be concerning.
-be skeptical of the suggested IDs from the iNat app

Thanks again for the responses page-- I use it very frequently and it's SO helpful!

Αναρτήθηκε από mcaple περίπου 6 χρόνια πριν

I used to have a response for both of those on that page but people generally do not respond well to being called out on them, so I removed them.

Trigger-Happy Agreeing: Please do not simply “Agree” with an identification that someone else has made without confirming that you understand how to identify it too. An identification confirms that you can confidently identify it yourself compared to any possible lookalikes. It's okay to leave it blank or let the community identify it for you. If you agree with the ID without actually knowing the taxon, it may reach Research Grade erroneously.

Computer Vision Fail: If you're not relatively certain of an identification, it's okay to identify it to a broad level such as "plant" or "animal." Then other iNaturalist users can help you identify it to species. Please don't select the option presented by Computer Vision without confirming that it looks similar and is likely to be found in your area.

If you can think of a good way of phrasing these that doesn't make people bristle so much I'm all ears!

It's also not super apparent when it's a Computer Vision ID (you have to view json to see). Sometimes users use CV as a sort of autocorrect/autofill--they are selecting it as the ID and do mean to agree with it rather than 100% relying on the computer ID. There was a plan to display whether or not an ID is from CV or not, but it hasn't been implemented. I'm not sure why.

Αναρτήθηκε από bouteloua περίπου 6 χρόνια πριν

Ah, yeah, that does make sense... I'm more concerned with the first than the second-- the second is easy to correct, regardless of if it's a CV fail or just a bad ID, whereas the first can of course cause problems. I can definitely see why people would respond badly to both... I unfortunately don't really have any suggestions on how to make it less harsh, though.

Αναρτήθηκε από mcaple περίπου 6 χρόνια πριν

Is there a way to filter for potential "trigger happy agreements" so that someone could check them more easily, rather than stumbling across them or going through every observation?

Αναρτήθηκε από mikeslater σχεδόν 6 χρόνια πριν

Not that I can think of, is there a filtering process you had in mind?

For things out of range you can check the "marked atlases" page (more info at the Learn more link at the top of the page)

Αναρτήθηκε από bouteloua σχεδόν 6 χρόνια πριν

All I can think of isn't a filtering process, but something that could help would be something that would get someone to rethink their quick agreement. I have written the following suggestion about a pop-up question based on the "trigger happy agreement" response you have above in this thread.

Similar to the pop-up question choice that comes up when a reviewer suggests a more general ID the observer had originally posted. A question would pop-up”
“If an observer has posted an observation with no or a general ID at first and then selected “Agree” with the first suggested ID”
Pop up the following statement and question choice:

“It may be better to wait for additional supporting identifications from other people before agreeing to this ID unless you have done additional research to know if this ID is correct.” with two choices

1) I’ll wait for further IDs
(the agreement is canceled)
2) I have studied this and I understand now why the ID is correct and I can identify this organism compared to possible lookalikes.
(The agreement is confirmed)

Αναρτήθηκε από mikeslater σχεδόν 6 χρόνια πριν

That's a good suggestion, maybe as one piece of a "new user" or "tutorial mode," after which it would turn it off once the user has been active for a certain number of days and/or uploaded a certain number of observations.

Αναρτήθηκε από bouteloua σχεδόν 6 χρόνια πριν

Προσθήκη σχόλιου

Συνδεθείτε ή Εγγραφή για να προσθέσετε σχόλια