my idea for the different organism thing is to let people ‘tag’ different organisms on the photo, just like you can tag peoples faces on facebook. See a white pine behind a bird? you can tag it and a tag gets attached to an observation linking to that organism. Can also do if you see a white pine in the same observation as a deer photo with no visible pine. It could send a notification to the observer that if they want they could clone the observation for white pine automatically. if they click yes it makes a new observation. If they ignore it or click no it just stays as a tag. (you’d have to track somehow so the same pine isnt tagged an cloned multiple times). The cloning process could also include automatic cropping to the area clicked on or even inform the algorithm where to look.
Maybe too hard to develop, but would be really neat. Kind of like the ‘darwin for a day’ thing that let you tag google street view organisms.
Yes, of course there are multiple differences between iNat and a museum. I meant that I don’t see the difference between them in that they are both subject to errors. It would take considerable research to determine the frequency of errors in each and I imagine the errors of museums would vary from museum to museum.
I’d be so nervous about letting people split observations - there are limited instances, IMHO, where adding a seemingly separate species may be useful for IDing the primary species. For example, sometimes I’ll have fungi where I’ll take a quick pic of associated trees to remind myself what they were around - if I know the species, I’ll usually just note it in my notes, but I’m unsure of the tree species I may include a shot at the end of the observation. Mind, I do make note of why its there, but I could see someone who isn’t paying a ton of attention (or a curator that speaks a different language) splitting an observation like that without realizing its relevant. (Regardless, I do usually try to just make a separate observation for the tree, but you there’s no guarantee another IDer would see that as they’re IDing)
For what its worth, too, I think there’s a distinct difference between a single photo that has multiple species and an observation that has multiple photos, all with distinct species. The former is a natural result of animal behavior on occasion (see this instance of a bird flock I ran into recently - there’s at least four species hanging out together here), and can easily be solved by posters/IDers just noting which animal out of the picture is being IDed, whereas the latter clearly requires either ID of Life or lowest common taxa
It certainly does. That often depends on when/if an expert in the group you’re interested in last looked at and annotated the specimen.
It is usual for an expert going through a taxon in a herbarium to annotate the correctly identified plants as well as correct those that are wrong. Not consistent, though, because that takes time. Most likely to happen where the herbarium can offer the services of a student worker to glue the appropriate labels.
I don’t want to see any more data quality issues (such as multiple species in one observation) added to ‘Casual’. I don’t always ID Casual, but when I do, I don’t want it to be difficult. It’s already bad enough that I have to specifically exclude other problems such as No Photo.
What I would suggest is a new category be created other than “Casual” which is something like “Needs attention” which puts the observation back to the observer for some action, whether it’s separating out the photos into different observations, deleting duplicates, or adding missing information and data, fixing anything that needs fixing. Those should not show up in “Needs ID” and should not show up in the Casual queue either, which should only be for high-quality captive/cultivated observations (yes it should also be renamed to avoid the stigma of being “casual” - just use “captive/cultivated” since that’s already in use).
I have seen such observations with a single photo, multiple species, marked “life” in the US and the obs told it couldn’t be IDd. Another variation on a theme.
That can be an appropriate response, but in cases like this the DQA votes are more clear cut, most likely “no evidence.” In the case of multiple pictures with different organisms there is no simple DQA solution, however, so people try different approaches.
If you just care about plants, and not whether they are cultivated or wild, you want “needs id” OR “captive” AND “has photos”. If you only choose “captive”, you can’t choose “Needs ID” and you also have to choose “Has Photos” if you don’t want to be bothered by blank observations.
I guess it could be technically possible, but even if you are lucky enough to stumble on this combination of settings, it will never be “sticky” and you have to reset this complicated set of options every time, OR go through the extraordinary trouble of creating a separate bookmark or cached link for your custom query.
I tried to construct a combination of settings that would give both “needs ID” and “captive” (and of course, must have photos), using the iNaturalist.org site. If you choose “Captive”, Needs ID is unselected.
I would have to study the UI to try to ascertain what checkboxes mean an OR clause is added versus what are effectively AND clauses. It seems that some can be multi-selected to gather up each category, and some operate differently. In other words, some of the checkboxes function as additional constraints (AND operands), and some function as additional options to include (OR operands).
What do you mean by “needs ID”? Everything in casual “needs ID” because it will never be at RG. That’s why the button is unchecked. What are you trying to add to your search or eliminate from your search with that button? Maybe there’s another solution.
We might think we want IDs for everything whether or not it is casual, but iNat as designed doesn’t think casual/captive/cultivated need ID. The designation “needs ID” is only ever applied to non-casual, non-captive, non-cultivated wild observations. That is what I have observed, but if that for some reason is not the case, I’m ready to hear it.
The goal is to return all “good” observations that are not RG.
By “good” I mean it has a photo at least, probably a Location. For birds, perhaps an audio file for a recording of the bird’s sound. I think this might be considered “verifiable” in iNat’s terminology.
I think a close approximate answer is to check both Casual and Needs ID, AND Has Photos. You suggested I also choose “Captive” but I leave that blank, because it means it must be captive/cultivated, whereas I just wanted to allow Captive/Cultivated, not restrict to only that.
So the bottom line, it takes a lot more clicks than it should be to construct a query that simultaneously gets Captive/Cultivated AND (Wild but Needs ID) AND verifiable (Has Photos, and Location).
I think I understand it all now except for the details of verifiable since I haven’t yet located the other filter options that might be required to recreate “verifiable” for casual observations. So my question now comes down to - when you include Casual in the filtering, what else do you allow in besides “no photo” - I suspect you allow in other data quality problems, and if that’s the case, how do you filter them out? Is there a URL parameter like “verifiable=true” perhaps?
When I look at casual observations to ID, I will also sometimes filter specifically for those NOT marked captive/cultivated (done with a URL change to captive=false) to look for otherwise good observations with identifiable evidence but issues such as a missing date or location. I’ve had some success with making comments pointing out the issue and observers coming back to correct them and thus bringing the observation out of casual into Needs ID/RG. Especially on endangered species, this may happen because people don’t feel comfortable sharing locations and don’t know about obscuring, for example.
Other things I look for is commonly confused species I’m familiar with at species level & casual with a community ID to correct wrong IDs if needed. As I understand it, if left uncorrected, these may feed into the CV training set. This is tedious because you can’t filter casual observations for just those with a community ID. If anyone has a trick for doing that, I’d like to hear about it. I just go through page by page looking for obs with multiple IDers.
I was looking at one species of butterfly last night, and I found several observations, by different observers but in the same country, that were marked “can still be improved” – even though some were already at subspecies. Those, as I counteracted with my “as good as it can be” vote, I also left a comment explaining that there is no further to go than subspecies and that, by definition, subspecies is as good as it can be.
I rarely see that problem, and was surprised to come across several in one night. I wonder if that is a cultural habit associated with the specific country.
Could be that the observation was marked “can still be improved” initially. I’ve seen many newer users tick that box before the first ID is even given, probably misunderstanding how the DQA is used. Then a number of identifiers look at the observations, ID, and forget (or don’t know how) to check the DQA!