Is a single additional ID enough to have an observation safely confirmed

When I started posting observations a few months ago, I believe 2 additional IDs from other users where required, now a single confirmation with the same species is enough.

I think this is not sufficient. Many initial IDs are probably based on the AI suggestion, and anybody can participate to the identification of any taxon, even if we are not specialists, so this is a bit light for a scientific procedure.

For example I have found “confirmed” audio observations without any trace of the species ID’d - and confirmed by another user. BTW I’m not sure if I need to “flag” these observations, or write a comment under the observation, and in this case, does the user get a notification that I wrote this message?

Just my 2 cents…

8 Likes

If you believe something is incorrectly IDed, you don’t “flag” the observation, just add a more correct ID. The observer and anyone else who has added IDs or comments will get notification of the new ID.

12 Likes

If there is no disagreement, 2 IDs are all that is needed for an observation to reach Research Grade; there has never been a need for more IDs, though agreeing IDs beyond the necessary 2 may be added. If there is a disagreement, it will take a more than 2/3rds majority for the observation to reach RG.

This is a heavily discussed topic in other threads. In short, there are many taxa and many regions of the world that have so few identifiers that observations would be unable to reach RG if observations required more than 2 agreeing identifications.

As Vireya said, simply add your own ID, preferably with an explanation of the dignostic characteristics you believe rule out Species A and/or indicate Species B. You don’t need to know the specific species in order to add a disagreeing ID. Simply ID to the lowest taxon you can be certain of. A disagreeing ID will bump the observation back to “Needs ID” (unless there are already 3 agreeing IDs for the original taxon, in which case, if you truly belief it’s inaccurate, you can tag another knowledgeable user).

This is true, but be aware that the notification system is pretty messy for high-intensity users. If you really need to get someone’s attention, it’s best to tag (@) them. But I would reserve this for situations where you really need to address someone specific (e.g., requests to remove erroneous IDs that appear to be the result of misclicks; specific requests for reasoning behind an ID). Always be polite even if in disagreement.

10 Likes

Thanks, in this particular case all I could ID was people talking on this audio observation, that’s why I was asking… but it’s been apparently removed since now (maybe the wrong file was uploaded, it’s easier to mess with audio files as the content doesn’t show as a thumbnail).

to reach RG, there must be two IDs which are specific, which is to say at species level (in most cases). If you added no ID or a broad ID such as “plants”, then two more people must add narrow IDs (eg., Red Oak) to reach RG.
If you suggested a specific ID, then only one other person needs to agree with you.

1 Plants + 2 Red Oak = RG Red Oak
nothing + 2 Red Oak = RG Red Oak
1 Birch + 2 Red Oak = disagreement, needs more IDs

if you are not sure of the species, it is better to add a more general guess. Or you can withdraw your ID after in the case of uncertainty.

are you on a phone, or computer?

4 Likes

Thanks. If you have a link for a poll about this on the forum, it would be nice. I can find threads but no poll so far.

If you really need to get someone’s attention, it’s best to tag (@) them

So I can tag users in the comments just by using @username under the observation, thanks for the tip.

about what specifically? what do you want to know?

2 Likes

I’m not following. Do you mean a poll about preferences on whether 2 agreeing IDs should be enough for an observation to reach RG? If so, I don’t think such a poll exists. Ultimately, iNat staff set the current system based off their own preferences (and the preferences of many/most established users).

Yes that is correct.

1 Like

are you on a phone, or computer?

I’m on computer only.

But I worded the title incorrectly - I only see the typo now… I hope it’s a bit clearer now that I edited it!

about what specifically? what do you want to know?

The poll would be about the number of agreements needed, I just have some doubts about 2 IDs being enough in many cases. The 1st one may often suggested by AI - most of the time correctly, but sometimes… not! And the 2nd may not be an expert.

I understand this is a very complex issue, and I also have unconfirmed observations, sometimes for months even t. I would just be more conservative and ask for more users to agree with an ID.

In the observation, in the section “Community Taxon” click on “What’s this?”

2 or more must agree and number of agreements must be “greater than” 2/3.

If for any reason you think that 2 (or more) agreeing IDs are not enough to qualify an observation for Research Grade… feel free to check the ‘Yes’ box under “Based on the evidence, can the Community Taxon still be confirmed or improved?”
It will keep it in the ‘Needs ID’ state (until you or someone else unchecks it), thus attracting more identifiers (maybe).
When checking that box for others’ obs, better explain briefly why you think it needs further scrutiny.

3 Likes

Given that this topic has come up many times and staff (and many users) have consistently supported the current system, I don’t think much would come of this.

While I agree this can be a useful course of action in certain situations, I want to stress, this should be used sparingly and only for specific circumstances. In almost all situations (especially those involving the observations of other users), adding your own ID and/or tagging other users is the better course of action.

4 Likes

The full text there is long and technical, so here’s a severely shortened version:

The community taxon (CID) represents what taxon the iNaturalist community thinks is depicted in an observation.

we try to choose a taxon that more than 2/3 of the identifiers agree with Sometimes this means choosing a broader taxon that contains a number of disagreeing taxa (e.g. you think it’s a kingsnake and I think it’s rattlesnake, so iNat chooses suborder Serpentes which contains all snakes).

A research grade observation must have two or more IDs.
All observations with at least one identification will also have an observation taxon. (categorization, essentially).

In cases of one broad and one specific ID, the categorization will be for the specific taxon. However the observation will not be shared with research partners yet.

the algorithm:

all agree = highest narrow taxon (needs ID if 2 or more at species*)

there is no over-2/3 majority at all = narrowest shared category (could be Plants, Life, sunflower family…) AND cannot be RG (I think…?)

there is an over-2/3 agreement that shares a category = the most specific shared category, AND if that is at species*, it is also RG.

*unless the last DQA question is checked “can’t be improved” but it is at Genus (or… subtribe???)

…honestly, and no offense to staff, but this is… not helpful for beginners nor for anyone who doesn’t know the finer mechanics of the site. The neat little bar chart is okay, but this may need a revisit. :P

2 Likes

almost every single identifier is a volunteer, so, yeah, it can be slow. but that’s okay! if you have enough time on your hands to learn about local organisms, you can always help our the volunteer team and speed it up for everyone :)
someone will come around eventually.

even in professional museums, there are many many items which have only speculative IDs, or IDs which haven’t been double-checked for decades. I know this because I’ve fixed mis-IDs in herbaria that were made fifty years ago and more!

for good and ill, the Internet is a powerful thing.

6 Likes

It depends on who the user is. If it’s a random user without any credibility, it could be just an “agree because I want to agree” ID. Most experts and serious identifiers are credible.

some connected threads where this has been discussed :

https://forum.inaturalist.org/t/agreeing-with-experts-and-research-grade/3718
https://forum.inaturalist.org/t/rename-research-grade-discussion-and-polls/590
https://forum.inaturalist.org/t/why-not-empower-recognised-experts/14419

5 Likes

I agree with you on the whole… I don’t think an AI autosuggest should count towards research grade. Fwiw the two counterarguments on other threads are :-

  1. you don’t know if people are actually using the autosuggest for the ID or just as an auto-complete for something they themselves know the ID of.

  2. that the problem is too taxon dependent - e.g. in birds, it makes total sense to have more IDs for research grade…but in slime molds and more complex taxa (where the datapoints are arguably even more valuable), this could mean nothing would ever make it to research grade, or that it would take even longer than it already does

I think issue 1 is something more for the observers themselves to solve. I don’t think it’s a strong argument.

But issue 2 is more problematic and I don’t see an easy solution.

3 Likes

A lot depends on the taxon and who confirmed it. Some taxa have expert identifiers that spend a lot of time placing ID’s. The CV suggestion being used doesn’t in and of itself make the ID less valid. If you upload something and accept the CV, and someone else who is an expert agrees, great, but you can check many of the people who place ID’s in their profiles.

If you find an observation that you know is incorrectly ID’d, than an ID of your own on it to send it a higher taxa. I would suggest that you keep an eye on your notifications though, it is possible that the identifier noted something you didn’t and will mention it.

I personally have no problem with the “Research Grade” name. I feel that many of the criticisms of that name wouldn’t be made if researchers using this data would spend some time familiarizing themselves with the data instead of making statements like this in a 2022 paper that Research Grade observations were: “specimens that were identified by two or more experts”. Because remember, RG is not an end all, every observation is up for review all the time.

5 Likes

On the one hand, 2 ID’s are a bare minimum and it would be nice to require more! Good in theory!

On the other hand, there aren’t nearly enough people doing identifications in some groups (e.g. plants) right now, and if 3 ID’s were required, the rate of confirmation would be even slower than it is. So I’m OK with the system as it is. It could be better, but it’s pretty good now. See results of a preliminary test of accuracy: https://www.inaturalist.org/blog/89255-we-estimate-the-accuracy-of-research-grade-observations-to-be-95-correct

10 Likes

I agree with @sedgequeen. Given the current identifier/observer ratio, if more agreements were required, I fear in many (even most) groups, iNat would very quickly grind to a messy halt to the dissatisfaction of both observers and identifiers. I’m not really sure how much of a problem “courtesy agreements” actually are, but I do feel the “agree” button is sometimes (often?) used as a “like” or “thankyou” button. Perhaps if there really was a “like” button that meant just that and no more, the problem of courtesy agreements could be limited. Just a thought.

8 Likes