I noticed that the numbers on a taxon page and then the subsequent leaderboard page don’t seem to be equal and I was wondering why.
As an example, the home page has txlorax with 3,733 identifications.
but, if you go to the leaderboard, it only has 3,286.
Is it somehow based on the region I’m in that applies a filter? I’ve checked and no filters seem to be applied.
And it doesn’t happen for observers, the home page and leaderboard page have the same value.
The leader board page does have one filter selected, “verifiable” observations, so I think that excludes maybe captive/cultivated observations or ones that have ended up with data issues that will prevent “research grade” status (like missing date or wrong location).
To make things more confusing, the identification counts from the /observations api does not match the identification counts from the /identifications api
website
txlorax - 3286 count
==
v1/observations/identifiers using taxon_id=81792
txlorax - 3286 count
https://api.inaturalist.org/v1/observations/identifiers?verifiable=true&taxon_id=81792
==
v1/identifications/identifiers using taxon_id=81792
txlorax - 3743 count
https://api.inaturalist.org/v1/identifications/identifiers?taxon_id=81792
==
v1/identifications/identifiers using observation_taxon_id =81792
txlorax - 3312 count
https://api.inaturalist.org/v1/identifications/identifiers?observation_taxon_id=81792
Thanks @pisum, I read through that post and it lead me to a couple of other threads (I did search before posting but obviously not well enough).
I guess I kinda understand. In the above case, if I grasp things correctly…
txlorax IDed a plant as Sugar Hackberry 3,733 times, but only 3,286 of those IDs ended up having a community consensus ID of Sugar Hackberry; those other 447 possibly being IDed as Common Hackberry or only IDed to the Hackberry genus.
It would seem more useful for the Top Identifier to reflect the person who has had the most consensus IDs. I’m less interested in a person that IDs a species a lot of times, but possibly incorrectly (I’m not talking about txlorax, just making a general observation) than in the person who has had the highest number of community consensus IDs. The “Top Identifier” moniker leads one to believe that this is most expert person at IDing a species, but it could be just a person who IDs a species a lot but possibly incorrectly.
And I guess that illuminates an issue I hadn’t thought about; how are “leading” and “supporting” IDs handled in the calculation of those numbers. If you’re the 1st person to ID a species or the 10th person, do those count equally? I guess that would be a kinda a tough thing to contextualize, though, because it would prioritize speed over accuracy.
As usual, it’s a lot more complicated than it initially appears. I’ll consider the issue resolved with the hope that the Bug Reports and Feature Requests provide some future clarity.
Having a top identifier who doesn’t have the most consensus IDs is useful in the case of uncommon taxa that aren’t in the CV yet, I think. At least for plants, most of one of these new-for-iNat taxa tend to be misidentified, because iNat’s CV setup pushes users to choose a species for an initial ID, so people not familiar with how the system works or people not familiar with the organisms are likely to select a species for an initial ID and thus start off with the wrong thing. Many observers don’t check their notifications often (or at all) and so even once an identifier has realized that there’s an additional species in the mix and has started correcting mis-IDs, observers may not change or withdraw their initial IDs and there may not be another identifier who knows the new species well. Thus, these mis-IDs of uncommon (to iNat) species can hang in ID limbo for a long time without consensus before there are enough correctly identified examples that people sometimes choose the correct species for their initial ID. In this case, it can be useful to know who’s been identifying the new thing so that you know who to ask if you think you’ve found one.
I guess that’s true if they are identifying it correctly.
almost, but not exactly.
it’s not necessarily true that Sugar Hackberry observations that txlorax identified are a subset of observations tied to txlorax’s Sugar Hackberry’s identifications. it’s possible that txlorax identified something as something other than Sugar Hackberry (ex. a maverick or an ancestor), and the associated observation still ended up as Sugar Hackberry.
furthermore, the Explore page by default filters for verifiable=true (which is equivalent to quality_grade=research,needs_id). so you need to account for that, too, when looking at the two screenshots in your original post.
compare:
… vs:
… vs:
and compare:
… vs:
