In that case, that would only leave subspecies, which isn’t required by iNat. With the Computer Vision being so accurate, I kind of makes anything but a Supporting ID difficult.
No, sorry, I changed my thought mid-sentence and wrote it incorrectly, you need to id higher groups to have less supporting ids, I agree sometimes, so I have 24% of supporting ids, but most ids I do are the first of that taxon for observations.
2 Likes
Interesting stats, thanks for posting the link! Turns out I have about 27% supporting IDs and the rest pretty evenly splits up into leading and improving.
I’m not sure though if I understand these categories correctly. I know supporting is e.g. when you help with getting something to research grade by agreeing with an ID or add agreeing IDs to an observation that is already RG. But what exactly is the difference between leading and improving? If I put an ID on a previous unknown, it says leading. I see that those turn into improving if someone else agrees with them. So it seems those are dynamic and depend on further community input. Is that how it works? Initial IDs are “leading” but get converted to “improving” once someone agrees with you?
It’s explained on the help page (@bobmcd first linked it) https://www.inaturalist.org/pages/help#identification
3 Likes
Go to Unknowns and pick out the bugs - first ID will always be leading.
My rough sorting often goes on to Marina.
iNat counts quantity not quality, so my Insecta or Pterygota counts as one, ‘equal with’ an entomologist’s can’t take that past Genus.
3 Likes
Yes, it would be nice to see out how many of my Improving IDs were at species level vs. higher levels, e.g. kingdom. Is there a way to further split up the stats to get an idea? I’m guessing a lot of my species level IDs are probably sitting in Leading though, waiting on someone to confirm.
Unfortunately, he can’t.
I think a way does exist but I don’t know it. I feel like I read somebody talking about it at one point on here though.
2 Likes
I believe the API would do it. Perhaps I will attempt to make a link.
3 Likes
https://jumear.github.io/stirfry/iNatAPIv1_identifications_identifiers.html?own_observation=false&is_change=false¤t=true&per_page=90&order=desc&order_by=created_at&user_id=arboretum_amy
shows all my IDs for others that are current (not withdrawn) and not the result of taxon swap: 85,921
Stick in your own user name instead of mine, and then add the following as you see fit:
&category=improving (or you can say leading or supporting)
&rank=kingdom (for IDs exactly at kingdom)
&hrank=species (returns IDs species level or lower)
&lrank=genus (returns IDs genus level or higher)
or any other parameters the API will accept, see list here
So I have 17,651 improving IDs of which 12,974 are species or subspecies. Or in general I have 63,995 IDs at species or subspecies, regardless of whether they are categorized as improving.
Credit to @pisum for this API tool thingy
9 Likes
Thanks, awesome! I’m already getting side-tracked putting all my data into a spreadsheet. To answer my own question, about 78% of my improving IDs for others are at species level, about 22% at other levels. For leading IDs, the numbers are about 55% at species level, and 45% others. The overall numbers of leading vs. improving are almost identical. I guess it is fair to say based on these numbers that species level IDs, especially on “unknown” observations, are more likely to get confirmed than IDs at other levels are getting refined/confirmed. Now I wonder how much of that is observers simply agreeing with the first ID that someone puts on their unknowns. I’m guessing though that trying to figure out how many of the confirming IDs are from observers vs. non-observers is not that easy.
1 Like
I think there’s another factor to keep in mind: observations that get species level ids fast likely to be easier to get that species id, if id is higher it means either observation has unsufficient evidence or it needs to be seen by more people with deeper knowledge, so it means more time is needed for an id to come. Also, observers are more likely to agree with any id they get than other iders, so there’s less following confirming ids if other people can’t id the narrower group.
3 Likes
I guess you have to be smarter than the av-e-rage bear.
Another thing I’m noticing that is really frustrating is identifiers who “do their best proofreading right after clicking send,” if you get my drift. Meaning that they don’t proofread. A couple days ago I came across a very clear, unambiguous observation of a lacewing. The reason it was at State of Matter Life is because the third identifier chose Neuropteris, a genus of ferns, instead of Neuroptera.
I get that a lot of you guys want to do a lot of identifications quickly; but what good is speed at the expense of accuracy? Rushing through in an effort to get more done is harmful in that you leave stupid-typo misidentifications in your wake. Optimizing productivity isn’t just about speed; it is about finding the right balance between speed and accuracy. A few moments to make sure you selected what you thought you did may slow you down a little bit, but it will be more than compensated by preventing harmful misidentifications.
7 Likes
From an earlier post (Homonyms happens) in another topic, I talked about these kind of entries into Life. Since then I have learned so much more about homonym and near homonym errors. Much of these errors are what I have called mis-clicks where the identifier is knowledgeable to the point that they can type in the name but the drop down menu trips them up with with the little thumbnails that can be hard to discern or the wrong taxon shows up first or the mouse hits the wrong one - sometimes the more prolific the identifier the more errors - but when the identifications are in the mid range of six digits and most of those are leading, what are the odds?
Even more interesting is when a common name was typed in and a homonym is picked forcing the ID to State of matter Life - this is a rabbit hole I go down once in a while when I can’t believe the identifier selected something so off base so I go to the taxonomies and try to figure out where there is overlap in common name because I believe they knew what was correct but things got messed up - more of a challenge when there is another language involved.
8 Likes
Homonyms haunt me and I check Kingdom Disagreement (for Africa) every day. Mostly easy to help it back in the right direction.
4 Likes
I have an identify filter question that maybe some of you who know more about URL hacks can help me with. Sorry if this has been asked before but I searched and haven’t found an answer yet.
I know I can target multiple places at once by adding them all to the identify URL (e.g. https://www.inaturalist.org/observations/identify?place_id=95113,72645,9012,95209 to cover all the National Parks along the Appalachians). But is there also a way to exclude a particular place, or combine existing overlapping places to narrow it down to “Place A but only the parts overlapping with B, not C”?
The particular use case I have in mind is making an identify link for the Blue Ridge Parkway (id=95113), but only the section that runs through NC (id=30), not VA (id=7), or vice versa. Trying something like https://www.inaturalist.org/observations/identify?place_id=95113&without_place_id=7 doesn’t seem to work.
Once I was identifying a hoverfly. It was of the genus Eupeodes, though I couldn’t go further. I typed ‘Eupeodes’ and clicked ‘Eupeodes’… (or so I thought…), then after a brief interlude of the spinning circle, my ID popped up: ‘Piciformes: Woodpeckers and Allies’.
I have no idea.
Weird stuff happens. I did correct it though :)
5 Likes
Sadly the opposite of place_id= is not without_place_id= but instead ¬_in_place=. When staff wrote these terms they must not have considered consistency. I was just complaining about this last week on the second search URLs tutorial thread.
2 Likes
A like by Diana reminded me that I had not updated the State of matter Life clean up wiki in quite a while (fortunately, life happens 
) From the table at the end of the first post (the wiki) March 10, 2021 total was 22,265 while April 10, 2022 is 37,186. One community behaviour change (possibly just because I did not notice before) is that users are actually choosing to identify things as Life for various reasons (which is kind of counter to my State of matter Life limbo post where I felt things fall off the radar for IDing for most users and takes some effort to resuscitate).
1 Like
It amazes me how often bits of plant debris get mistaken for insects. The strangest thing I saw today in State of Matter Life was a bit of woody debris floating in duckweed – the observer thought it was a newt. Admittedly, one dead duckweed leaf was stuck to it in such a way that it could look like a closed eye, but still…
3 Likes