I wasn’t expecting an immediate response
Okay, that is what I get when I click that link. When I do as I described previously (go to project, then observations, then Identify) I get only checkbox next to casual checked. Also just realized if I search for a project and click the link directly to observations (see screenshot) and then click on Identify I get only the checkbox for Needs ID checked. So weird!
Having thought about this for a while, it seems I had conflicting points in my own argumentation.
My updated view is: You can never be sure about the preferences that affect the data you source from anywhere. Even if you gather and analyze all the data yourself, there will be variation in personal preference at each point in time, of which you can’t be completely aware. I also include things like mood, hunger and tiredness under the “personal preference”-label. It is not an issue with just the laymen. It’s just something you have to be able to live with.
There was an interesting discussion a while ago about biases in what people decide make an observation of, that this reminded me of.
Now this is one of those things what I meant with the complexity of the subject of iNat. I have little doubt, that given a large enough amount of IDers, all the fields in an observation would start converging on “correct”, as you put it. Or, at least as much as can be done with the source material in the observation. But I have great doubt there will ever realistically be enough IDers.
That threshold view is interesting, though. Hadn’t considered it.
Thank you, at this point. This back and forth with you has been refreshing, thought-provoking and interesting. Not really something I often run into.
I just looked at the phylogenetic projects and added some IDs. Some of the observations I reviewed are research grade already with three or more IDs. Is there a way to remove such an observation from its phylogenetic project?
My point was only about people using these projects. A disagreement only with something I didn’t say is not a disagreement with what I said.
No, these projects do not need the coarse IDs since, as I explained in details, the candidate observations are observations without ID and observations with “coarse” IDs. The transition from “no ID” to “coarse ID” does not change the fact that an observation is, or is not, a canditate.
No only inexperienced users, but I would better let @DianaStuder elaborate, as she pays much attention to placeholders in her identification work.
Good intention. Thank you for having elaborated, as it makes the point more precise and the discussion easier.
To elaborate further, I suggest to check how many observations identified as “plant” are in one of the projects for “animal”, and how many observations identified as “animal” are in one of the projects for “plants”. I can’t do it in a short time for “plants” because there are many projects and URL filters on iNat are limited to 500 values, but I can do it for monocots (and all subprojects related to monocots). Results:
With your suggestion, we would not have a gain factor of 2:
Moreover, some of these IDs came after observations were pushed to the projects, so these low gain factors are even [a bit] overestimated.
Conclusion: useless, but worth checking.
(I did the checking. You could have checked either. Not a waste of time.)
PS: when I estimated that 95% of coarse IDs are useless for people using these projects, I didn’t know that I was so close to the actual value, at least for “animal”! But for dicots, it’s 99,75%, considering the gain factor 1.0025.
You used the link to observations generated by iNat.
The results you got reflect the “editorial choice” of iNat for you.
I can’t customize that!
I suggest you to use the links I generated in the “About” section of the projects, where I let you choose between several options, for example for the project Unknown / Liliopsida (monocots) the links proposed are:
Identify observations without id.:
‑ in project,
‑ in project and sub-projects,
‑ in project and all sub-projects recursively.
Identify observations with high rank id.:
‑ in project,
‑ in project and sub-projects,
‑ in project and all sub-projects recursively.
Parent projects: Tracheophyta, Plantae, Life.
Sub-projects: Alismatales, Arecales, Asparagales, Commelinales, Dioscoreales, Liliales, Pandanales, Poales, Zingiberales.
Yes.
See my other response in the message just above. I suggest you to use the links in the “About” section of the projects. Using these links to open the “Identify” page, RG observations are excluded. As you can see if you click on the “Filters” button:
Then, it’s up to you to further change the filters (uncheck the “Casual” if you prefer, for instance).
Ok perfect!
Then I see no drawbacks in adding coarse ID: it causes no harm to IDers that use yellow label projects, but helps IDers that don’t use them.
The only risk is deleting the placeholder, but I, too, think that iNat should fix this problem, and an informed IDer can copy the placeholder as a note under the ID.
Other options you could consider? Since you need clear info for local invasives.
USA has a Master Gardeners programme - you may be able to find some local help there.
Or Friends / volunteers at a local nature p/reserve
Also blogs - this one is Tennessee https://www.clayandlimestone.com/2024/07/wildflower-wednesday-path-rush.html
Unfortunately it applies across the board. Placeholder text is greyed out - iNat treats it as junk. It took me a while to notice it, look for it before adding an iD, and now use it consistently. The observer may use it as a ‘herbarium voucher’ but iNat destroys the info and prefers - It’s A Plant, duh! Placeholder text is only visible if the identifier makes the effort, takes the time, to … open … each obs.
iNat treats placeholder text like a high school spelling bee. One letter missing, doubled, or wrong - you’re fired! A careful and correct scientific name (as placeholder because, field work, poor internet, whatever) is binned as a Dicot, yay. Then there are missing sp which need to be Flagged for Curation.
tiwane said they will not change this, because iNat staff know Placeholder is temporary. It is absolute silence, not even a trace of Placeholder was here.
I don’t understand why anyone would deliberately choose to record temporary info, in the sure and certain knowledge that it will disappear without any warning. Perhaps iNat needs a private field on an obs - not my battle.
So we (i.e. @jeanphilippeb) pulled the next batch of 20 obs with placeholder text. No missing sp there since they are Southern Africa (my missing sp are Rest of Africa). One example of a correct scientific name dumped in dicots. 10 of the 20 were tiny typos. The problem is tangible. But I no longer have to stress about - wait! - let me catch the placeholder first.
We cannot retrieve placeholders where there is already an ID - I have to let that burden go. iNat has destroyed that info. Linnaeus spelt that wrong - bin the scruffy paper - keep the shiny new digital ID - to paraphrase @rogue_biologist
What is unkind to newbies is iNat’s
Shall we use that as a placeholder for you?
Either Placeholder should be treated like any other text / data an observer adds or Placeholder should be excised from iNat. Issue is now compounded by the insistent flashing red lights and sirens - you forgot to add an ID. I have seen an obs IDed as something (now be quiet and let me get on with this). People don’t realise you can ignore that red, and forge ahead regardless. I don’t see a benefit from bullying observers into adding a helpless placeholder or very broad ID.
I thank you all for opening doors and windows to your part of iNat.
Why do you do that?
That way?
Oh, oops, so I shouldn’t have …
Thatnks for the tip, bu honesly this works fine for me. I don’t have very much on my property, and this gives me a good idea of what stuff is. I was just trying to point out that if anyone can post an id, then there should be some way to tell who the people are that know what they’re doing. There’s a leaderboard, but that’s only as good as the criteria to get on it. Right now, from my understanding, it only shows who has the most id’s. I never would have thought there was a problem until I came across the original thread, and all the replies made it seem like a bigger issue than it is.
So, what is the consesus please? Is there any point in me getting into the unknowns and iding the plants to Dicot, Monocot, Gymnosperms, and sometimes to family, sounds to birds if they are hopefully a bird, insects etc.
I also do similar with Kingdon Plantae. Is this useful to add course identifications?
I have got into @jeanphilippeb’s projects and will find them useful to find groups to id.
I am learning a lot from reading forum posts.
A summary of what people said here: one person thinks coarse IDs are always a waste of time, one person thinks high plant IDs are a waste of time but other high IDs are okay, and everyone else on the thread either believes in making coarse IDs or did not express an opinion either way. Given what is written on iNat help pages, I would say staff is in favor of course IDs.
This is an interesting discussion. I mostly make coarse ID’s because I really enjoy spending time on iNaturalist but only have the knowledge to ID a handful of species to finer levels. I thought I was doing a good thing by reducing the pile of unknowns, even when changing a hundred blank observations to “plant” gets tedious. Adding coarse ID’s seems to be encouraged by iNat, including in the frequently used responses to provide to new users.
What I think would be more beneficial is encouraging users to add initial ID’s instead of allowing them to be left blank. If a person is interested enough in plants or fungi to add 100 observations of them in a day, then they probably know at least enough to add some ID’s to their own observations.
I think adding coarse ID’s is great! I for one will never find plants left in the “unknown” pile because I don’t search on that.
We can’t require observers to suggest an ID on every post because many people, especially students, have no clue.
tiwane said iNat should be fun. So find the part which you enjoy - where you feel you add value.
For example - I can pick crabs out of unknown - no idea how to take the ID further. But, I can pick out our African FRESHwater crabs and savel will filter them out. All the other crabs … that is loarie’s corner.
Retrieving birds - yes, done while you wait!
And insects too.
Can we find you a better slice to enjoy? Have you tried annotating for phenology graphs, for a taxon which does make your heart sing?
Or for the handful you know - try going up a taxon level? Or (what I loath) a taxon sweep thru the RG obs to pick out - wait, that’s not …
That would indeed solve the problem for the vast majority of organisms, but I doubt it will get implemented to account for edge cases such as “dead matter” being mistaken for an organism that wouldn’t even fit in “life”.
You are. I think if there is one thing that this thread shows, then it is that different identifiers have different workflows. So while you may take some observations away from one identifier that way, you will also make them be found by another. In any case, with the amount of unidentified observations, I’m confident that there’s still enough in both unknowns and in Plantae for everyone to be busy. Haha
If it gets tedious, why not switch it up and learn to identify something new?
I sometimes need a break from my usual IDing habits too, so it’s nice to switch it up. (For this, I can only recommend finding a small taxon you like (usually genus/tribe) with only a few species in it that are identifiable from photos, but have no real active identifier on iNat.)
Many different ways and ideas on the tandem thread
https://forum.inaturalist.org/t/prolific-identifiers-how-do-you-filter-for-observations-to-id/54481
Despite my expression is clear-cut, I don’t want to make a big issue with the discussion about coarse IDs.
Sorry for this long response. Drafting a solution is long.
What is the big issue?
We need observations to be identified (the iNat added value) but there are too few identifications (all requiring some expertise, academic or empirical, and time) and there are more and more incoming observations. Yet, CV is available to help identifiers and has considerably improved for years.
The big issue is the difficulty for identifiers to find the observations they can ID efficiently. The big issue is the lack of support for CV-based filtering in the iNat web pages that filter observations (“Explore” and “Identify” pages). Imagine a checkbox to enable CV-based filtering in your search queries, so that you can find all observations of birds, with or without IDs. CV-based filtering by iNat (the true solution) would be better than phylogenetic projects (the palliative). It would require storing precomputed CV suggestions in the database, to make them readily usable for filtering.
Phylogenetic-projects-based CV-based filtering is not well enough integrated, as someone said, but it is usable right now. BTW, would you say that the unidentified observations are well integrated in iNat (not mentioning the placeholder)?
What’s my point with the discussion about the coarse IDs?
My point is to suggest you to use the palliative solution, until a better solution comes from iNat (if it comes), because our resources are limited and because we do need a more efficient identification work. For the same reason, my point is also to prevent wasting time.
I never said that coarse IDs are “bad” besides the placeholder issue. Moreover, as @DianaStuder mentioned, we are working on a solution to the placeholder issue, presently operating at small scale, and very promising. Will you like it? Made possible thanks to our expertise, will and efforts : I precise that not for self-promotion but to make you aware that you should not wait too long for a solution from iNat to some issues.
Lacking CV-based filtering, we are discussing the benefits of adding coarse identifications, concluding that you need these coarse IDs. Be aware that you express this need after considering the solution you got, and after being teached that some key aspects of the solution won’t change. This is a considerable bias in the vision.
With CV-based filtering, most coarse IDs become useless. (Just as computers, scanners and printers made carbon paper unnecessary in most circumstances). Of course, if a CV-based search for animals yields 5% of plants, we would ASAP identify these 5% as plants, to let the ID-based filtering have precedence over the CV-base filtering. When I say that 95% are useless, I mean that 5% are necessary. CV-based filtering can spare the tedious 95%.
what is too few? i don’t see any evidence of there being too few identifications. whenever i’ve taken a snapshot that shows research grade vs verifiable, that ratio seems to be holding steady over time.
this seems to lead to such a pessimistic view of the value of human experience and knowledge. there’s value in going through a bunch of things you’re not familiar with and trying to figure it out without a computer telling you in advance what it thinks it is. sure, a database can replace a file cabinet as a means of recording information, but do you really want a future state where people only know what something is because an algorithm told them that’s what it is?
this also seems to value only the taxon specialist workflow, while ignoring the value of all the other identifiers out there.
