Searching for observations with more than 3 identification agreements

Is there any way to search for observations that have a certain number of agreeing identifications. basically research grade plus more.

I know there has been discussions about increasing the bar to become research grade, but I guess that has it’s own issues.

If there was a way we could search for observations that had more than a given number of agreements we could select the grade depending on need.

Is there anything in URL searching perhaps?

1 Like

Hmm… interesting idea. I would imagine someone out there could tweak that info out of the API.

Bearing in mind that not necessarily all of the additional agreeing IDs represent greater confidence or knowledge. There have been a few incidents of users who seemed to be numbers chasers; who agree to thousands upon thousand of records for the fun in seeing their position on the leaderboards.


Well, you can find these in most of birds, many mammals, very few in fungi or insects (except for several charismatic and recognizable species).

Not supported by any user facing tools, nor is it in the API.


Which goal do you have from searching those - checking given id?

Yes. at least trying to refine certainty of ID. I’ve also just found a work around. When I download the data there is a column for num_identification_agreements. which pretty much allows me to do what What I was trying. albeit after download.

I’ve also wanted this feature, because I’ve found that there are many misidentified observations with 2 identification agreements.

However, I think on a deeper level, what I actually want is a bit more subtle or complex than just searching for ones with 3 agreements, because I have occasionally seen misidentified records with 3 agreements.

What I would really love is if there were a way to filter out casual users. In an overwhelming majority of the cases where I see a misidentified record with 2 agreements, the users have made few ID’s, and in many cases they aren’t even active on the site any more, which is a bit frustrating because they don’t respond to discussion of ID’s.

To me, these records seem obviously different from one where there are two ID’s both by experienced users who have ID’ed a dozen or more species of the relevant taxon. Yes, there still could be an error, but the chance of one is much smaller in this case.

I would love for the site to have some sort of probationary or training period that people go through.

iNaturalist already auto-calculates things very accurately, like listing taxa that are frequently misidentified. It seems like it wouldn’t be that hard to assess things like how often a user’s initial ID’s are deemed correct long-term, and also perhaps to calculate whether or not a user is responsive to discussion or disagreement.

I think there’s a huge difference between “drop-and-run” ID’s where the user is no longer active, and ones where the user is actively examining it, researching, and engaged in discussion. I would ideally want these to be treated very differently.

It’s also frustrating when I encounter a record with 2 or 3 mis-ID’s and I post on it and then none of the users replies for weeks or months. In that case, my only recourse to fix the record is to @ - message multiple users, which kind of wastes their time, as I try to be sparing about bringing in other users.

I don’t know. Sorry that this is really rambly. I just have thought about this a lot, because I’ve had the same desire to search for records with 3 or more ID’s, but then I started thinking about what I really wanted, and it was actually just that I wanted records I could trust more, and to that end I think there are probably more sophisticated ways to filter them that I would strongly prefer. I.e. I would prefer if there were slightly different, perhaps more stringent standards for making something research grade. And then this extra search feature wouldn’t be necessary, because the things would naturally have been filtered out.

i think this generally will be okay for research grade observations, but if you’re dealing with some atypical RG observations or with observations not yet at research grade, you’ll have to keep in mind that num_identification_agreements works using the community taxon as the basis for agreement (see

if you use the API, you can get a little more granular in determining what kinds of agreement you have (ex., but that determination also can be done only after download.