Agree - well, the idea was initiate a discussion on this topic, which then someone could bring to feature request level.
But I was interested in a thorough exchange on view on the matter, as I have struggled a little bit myself and wanted to understand the point of view of other iNatters.
2 Likes
The big part of invertebrates canāt be ided by the photo lone, but I would like to still se their photos, it gives an initil understanding how the genus (at least) looks like when I do exploring on iNat or anywhere else. If the person wants to do things correctly they should do a research before adding any id. Gallery is not a point where you start identification though.
5 Likes
sorry I did not mean to remove the picture, but have them aggregated in a higher taxa level view. Like some king of redirection to upper taxa in case of ambiguity.
But this is as I mentioned above is not as easy as there are cases in which you run into āa or b not cā or āb or c not aā
But this could be at least done in cases like the pairs of undistinguishable or lichens or for instance fungi who are macroscopically identical but differentiated with examination of spores (such as Sarcoscypha)
The issue with this is that there are people who do dissect invertebrates and look at things under a miscroscope etc. and take photos and post them on iNat. So while the majority of observations canāt be identified to species, some can.
This was the case with Eupeodes americanus - there were hundreds of observations identified as it, but there is another species in the same genus that looks identical, and other species in other genera that look extremely similar. So E. americanus was a good guess, and probably correct for many, but there was no way to tell. Now all of those observations have been moved to subgenus level or to other genera where appropriate, but there are still 3 good observations because a couple dedicated observers were willing to put in the effort and capture one and get the necessary photos.
The best available solution is that everyone is careful and researches before identifying, and identifies to the lowest level that they are confident includes all the possible options, clarifying with comments if more detail is useful. Then nobody is confused about what is possible and thereās no inprecision. But it is true that more could be done to help people with the research and understanding what the limits are.
14 Likes
Youāre right of course. But I feel this situation comes from the tension of the two different faces of iNatā¦ someone approaches it with a scientific mindset (Iām sure itās at least the feeling of people working with Natural Sciences) other as a generic connection with nature who feel a āmost likely approachā is good enough.
I really think that iNat should work out something to conciliate these two souls, at least with some official guideline and maybe tweaking the interface to allow them to coexist without tensions
4 Likes
yep. Maybe a warning popping up in the space between the taxa and the comment box warning that microscopy is required to be able to confirm an ID. Could be a simple warning linked to a specific page for the taxa in question explaining the potential ambiguity, allowing user to comment adding references to literature etc.
I would definitely read that, as I spend a lot of time on researchgate and google scholar to find something useful - and the results of this search could be shared between iNatters!
Thanks, I am trying to understand the logic behind the Holding bin and related workflow.
Found this example which looks quite a good description (refer to the Comments section at the bottom of the page):
I wonder why this is not being used? Why for instance, once populated the Holding bin, IDs of the species binned are not prevented with a warning?
Some explanation about the logic applied to spiders here
Looks like
@bouteloua was involved - could you please tell us why the use of Holding bin did not spread?
EDIT: reading the profile of arachnojoe looks like this problem is felt quite a bit. Same feelings as @tom_carlberg 's?
Also found this Feature request https://forum.inaturalist.org/t/ability-to-id-to-a-group-of-visually-similar-species-within-a-genus/5655 for anyone interested
1 Like
Iāve faced similar situations to the one you describe. I always choose the highest-level taxon that encompasses all the possible species-level IDs, even though this is sometimes painful to my ego. For example, there are several Trillium species that look essentially identical when they are too young to flower. However, there are always at least some species in the genus that can be ruled out, even at this young stage, for example, because of differences in their leaf morphology. It pains me to choose Trillium genus as the identification in these cases because: āI know it isnāt just any Trillium - I just canāt tell, based on the blurry photo, between a small subset of species!ā But I suppress my ego and choose Trillium at the genus level, because this is correct and a species-level identification could very likely be incorrect. I usually then write, in the text field below the observation, which species within the genus I think are more likely than others.
It is very frustrating to me when people offer identifications, however well-intentioned, that are more precise than is possible given the evidence. This results in wacky data in the long run.
For example, Solidago canadensis and Solidago altissima are two similar-looking plants in my region. They can be told apart easily, but it takes special care to photograph the relevant details. From iNaturalist, you would think that S. canadensis is far more common than S. altissima in the northern USA. There are several thousands more observations of it, many of them research-grade. But very few of these observations have enough detail to be reliably distinguished from S. altissima. In reality, S. altissima is well known in the literature to be far, far, far more prevalent than S. canadensis. Iād wager that a huge majority of the S. canadensis observations are actually of S. altissima. But itās such a mess due to overzealous species-level identification, that now weāre left with poor quality data. (In this instance, most of these ambiguous observations could be usefully IDād to the taxonomic level of subsection: Solidago subsection Triplinerviae).
19 Likes
reading through this topic, iām thinking of @upupa-epops 's calligrapher fly observation field https://www.inaturalist.org/observation_fields/8909 , which looks like the bird holding bin that was just brought up. i find that holding bin very helpful when my flies get a value added (the link code broke?)
these are definitely a solution to just picking one of the possible species and just going with it, which is in my opinion an awful way to go about things. thatās just misinformation. like was already said:
and actually your reply to this ^^ worries me
hereās the thing. that animal is literally, factually, one or the other. and to an observer that is so casual that they donāt care whether their IDs are factually accurate or not, why would it even matter to give them a specific ID in the first place? for someone who thinks āmost likelyā is āgood enoughā then just say it in the comments, donāt throw an ID out there and bungle data for scientists or casuals who would like to learn.
only ID as far as you actually can. the rest can be commentary or observation fields or something. that is the peaceful middle ground.
9 Likes
I understand.
The problem I cannot make my mind over is how to guarantee consistency. For instance given the mess I reviewed nearly all the Araniella, took a couple of days. If you consider the possible ambiguities, out of the 4-500 obs only maybe 20ish had picture which allowed some hypotesis based on epygine. And then you discover even epigyne is not that clear.
So the problem for me became: should for consistency flatten out all Araniella IDs to genus level?
If you do the same everywhere I am not sure how many id we should sanitize
I opted to apply the most likely criteria because it looked more reasonable.
But I understand this is considered not acceptable.
But I feel that before using the flattening out option there should be some official guideline otherwise people might think you are just an overzealous pain in the neck.
About this, just to understand, if I had to use the data for research purposes Iād review them all anyway since knowing the difficulty of a proper id I could not use the community consensus in any case, as there is no guarantee the IDier is a pro (maybe save the ones confirmed by known pros). Why do you feel this is not satisfactory?
5 Likes
I hear you. That sounds like a frustrating quandary. There are a huge number of observations in the Solidago example I gave, (and it sounds like this is true in Arianella as well), where it would be useful if iNaturalist had a button to register something like:
āI seriously doubt that this was identified to this very fine level of precision, but I donāt want to explicitly rule out the precise designation youāve given it.ā
For me, the best case really would be the āflattenedā scenario you give. If the data is mostly only of genus-level quality, then iNat should reflect that, in my opinion.
7 Likes
Look at the evidence provided. Make the ID you are confident that you can based on that evidence. When you do that at a coarser level than others have already, then it will ask you if you explicitly disagree with the CID (and the wording in that question is whack, but essentially it is asking you if you can see reason the CID is wrong). If you can see evidence to say it is not that species, choose the bottom yellow option. The green top option is if you just canāt tell for sure.
Itās that simple. Just worry about your own ID. Donāt worry about whether the CID is wrong or differs from your own. YOU are not responsible to āfixā the IDs in iNaturalist! YOU donāt need to guarantee consistency in anything other than the IDs that YOU put!
If you think you have expertise or resources that can help refine IDs, then by all means review observations. But just make your identifications as you see fitā¦ you DONāT need to create a debate about changing how iNat handle IDs, especially after only 3 months of experience doing so.
2 Likes
Hi - it looks like youāve edited your post since I responded to it (or else it was truncated by my browser somehow?) Please donāt do this; it makes it difficult to follow the conversation.
To your new point, beginning with: āif I had to use the data for research purposesā¦ā
Iād say that the fact that downstream users of iNaturalist data should double-check the collected data is a little beside the point, here. We should always strive for accuracy. This would, by the way, make our data more attractive to downstream researchers in the first place. For example, when I put myself in the shoes of a researcher new to iNaturalist, Iām troubled by the flagrant inaccuracy that I see in some genera. I think many of them would simply walk away - much like the lichenologist who posted above.
4 Likes
I find this approach to debating the problems quite annoying and not proper for a community.
I see you are a curator, I would expect a bit more of restraint, probably your thousands of id makes you feel in the position to tell people to shut up which is an impression I got from the tone of replies from seniors in other topic to other inatters.
Maybe a reason there are not so many people active on the forum?
Just my observation, which is off topic but maybe you should consider that the fact a person has not thousands of IDs does not as well mean has no experiences in other sectors like quality control that might be applicable.
Once again, you are a curator, if you think I am wasting your time just do not reply. I think there is some option to unsubscribe threads, ignore users or the like. As a curator, you should know.
Sorry @ddennism, I edited in the same for the suggestion of the system of not posting multiple replies but quoting in the same reply.
I see your point and I agree.
I think iNat should prioritize some mechanism or instruction to handle this, so that the cases with genus level id should not be categorised as āneed idā.
Not sure if in these cases IDing to genus level and flagging the obs as āid cannot be improvedā would be an acceptable option.
Maybe using Holding bin+comment on genus id (the nasty one)+not improvable flag would be an acceptable option?
Opinions?
3 Likes
Why not? Thereās absolutely nothing wrong with having the goal of improving the accuracy of the iNat data by using your expertise - itās one of the primary reasons I use the site. I can think of a number of users who have nearly single-handedly cleaned up entire groups that were a mess and rife with inaccurate research grade IDs.
8 Likes
Yeah, maybe! (You might want to float those proposals out in a separate post at this point.) Cheers.
I think the problem with having this type of flag is that the set of observations that require it are not discrete. In the case of birds (my specialty), itās far more often the photo quality that leads to the inability to ID to species, rather than a species pair not being identifiable regardless of photo quality (many of those require a vocalization).
Because @edolis is suggesting that he should use his expertise to make IDs at species level even though there is not evidence of, in order to rule out the possibility of other species. That is the OPPOSITE of improving the dataā¦
https://forum.inaturalist.org/t/etiquette-for-id-of-species-with-no-visual-differences/9513/21
No the second is crap. One can not proof anything right. All one can do is test and proof wrong. So for every scientific hypotheses (including an iNat species ID), the first is the right approach. https://en.wikipedia.org/wiki/Critical_rationalism
An ID that can not be tested (e.g. as there are no characters - morphology or location - that allow to differentiate between several taxa) is not a scientific hypothesis and can only be subject to believe, not to knowledge.
This means on one hand: that finer IDs not supported by the evidence provided should be avoided.
But it also means, that for example the identification of all royal ferns in North America to the North American subspecies, is a valid scientific hypotheses (and therefore a valid ID), as this can be tested and falsified (at least with additional field-samples).
3 Likes
Oh sure I left that out of the box intentionally. If the pic is blurred or missing view of essential detail, I just go to genus and flag as not improvable. Letās say āeasyā and easy to explain as well. Ther real issue I was talking about is when there is no way to tell based on the pictures only. And people reply it is a perfect match of other obs they have seen in the taxa gallery. Thatās why I usually selfenroll in some review species-genus specific. Like the Empoasca vitis/decipiens mentioned above-until I run in this issue with E. pteridis.
And since you are taking quite some time to review a genus, Iād like to understand how to make it properly the first shot.
1 Like