I am not sure where it is best to post this topic:
I noticed that a lot of users initially post an observation without identifying it in the first place (which is fine), but as soon as an other user suggest an identification they (dumbly?) agrees to it.
So we can end up with research grade observations with very few validation and a wrong ID.
I do not know how it is best to deal with that behaviour; perhaps iNat should raise a confirmation message when the same user propose (or agrees to) an ID that is different of what he proposed earlier. Also I would recommend that at least 3 different users suggest the same species ID before it go to research grade status.
I see this often, what I do is advise the person nicely that it is not always wise to immediately agree with an ID, I have had great success doing that. Or you can always just tag someone who will know for a confirmation. But yes, itās probably a good idea if iNat can possibly set up a system that helps decrease this, I canāt think of a way that wonāt annoy proper identifiers though.
This is quite the issue. I do a lot of Salticidae IDs, which require counting hairs, looking at eye ratios, or any other very subtle sort of distinguishing. Iāll bring the ID from Family to Species and immediately have the user agree with my ID.
Maybe if we as identifiers had some disclaimer to copy/paste such as the words written in the āhelpā section under identifications on this site: āPlease do not simply āAgreeā with an ID that someone else has made without confirming that you understand how to identify that taxon. An identification confirms that you can confidently identify it yourself compared to any possible lookalikes.ā
Having a copy/paste at the ready would be very useful, especially if you ID like I doā¦ dozens or hundreds of IDs in a day.
I wonder if this is any different than an observer blindly picking the first species that is automatically suggested and guessing ārightā, or an identifier blindly agreeing to an observerās id. Iāve noticed the latter happening as well and that results in even more inaccurate ids than the situation described above. If a changed id by an observer is discounted, then perhaps observers shouldnāt be counted among the two ids. Or rather, there should be at least 3 ids. That would cover all bases. An immediate change of id may simply be because the observer was debating between two possibilities and views an identification by a more knowledgeable person as confirmation of their alternative possibility. It is no different than if they had proposed that alternative to begin with.
It would be a problem if ResearchGrade was irreversible, but misidentified RG observations can be corrected, so I donāt really think it is too much of a problem in reality. RG obs arenāt harvested immediately anyway by data partners (possibly in part for this reason). The onus is on identifiers to be vigilant for misidentified RG obs.
I donāt see this as a problem. RG observations are meant to be used by researchers. When going over these we can expect them to check them individually to make sure they are correct and to discard the iffy ones. These users may also feel comfortable trusting the IDs because the identifier has a lot of identifications or the site calls them a top identifier.
I do not think that you can expect researchers to go over each one individually to checkā¦ but you might be able to expect them to add some uncertainty into their calculations to account for misidentifications (I do at least).
Youāre probably right. I had to go over observations for a paper but there were less than a 100. It would probably take too long for someone working on a different taxon
This is an issue, along with blindly accepting the first suggested ID, where I am. If itās in an area where there are a lot of regular users the errors get weeded out, but it youāre in an area where the users are transient and iNat doesnāt have enough observations over-all to provide good recommended IDs (and also where the IDs are complicated and people make incorrect IDs often) itās a real challenge.
Iāve introduced my staff to iNat and have had to spend a of time explaining this issue to them as iNat often suggests IDs for species from a completely different part of the word and users will do the same when making identifications. Iāve told my staff to double check the suggested species range and to cross-check with the species lists we have for the area.
That helps, but itās not a solution that works in the majority of situations.
For the greater number of users Iām no sure what a good solution is, but I do think itās an issue that needs more attention.
Personally, I think the āagreeā button should be removed. That would cut down on this sort of behavior a lot. From a psychological perspective itās very easy to off-hand hit agree, but less people will go to the trouble of putting in the species name in to their own suggestion unless they are sure.
The agree-button as it is seems like itās asking a question, and psychologically people are inclined to answer.
It has been pointed out how āAgreeā button IDs in the thumbnails of the Identify page donāt show descriptions or dialog, so often arenāt seeing crucial info relevant to an ID.
New users typically see āAgreeā as a form of Like or a thankyou to an identifier, so much more on-boarding about that needs to be done. We try and educate new users when we encounter it, butā¦
I will often confirm an ID, not fully knowing for sureā¦ but only if I have come to trust an Identifierās accuracy and knowledge. I consider it a form of adding weight to an expert, but I donāt give such weighting ājust because they are an expertā, I have to see and experience and develop that trust. I wonāt add the weight where there is any sort of dissention however, and if dissention develops, I will often pull or withdraw my supporting ID.
IDs are dynamic, as Stephen pointed out, so CID will eventually settle in the right place. There are many of us that go through and review ALL obs (wellā¦ to a filter for place and/or taxa) including the RG and casual ones, and we pick up many of the mis-IDs. If you are a specialist or expert, or have resources that could make you one, then doing likewise for an area of observations can add value in a very real way.
And it is Community ID, so you donāt need to worry too much about them being āwrong IDsā. Make your ID as you see it, start dialog if you think it will help others learn or teach, and then let CID fall where it may. You can tag in others that you think might contribute their views, but avoid the temptation to āfixā wrong IDs by gaming the system. Weāre not voting in a new Government hereā¦
Iām certain that iNat staff and developers are aware of these issues, and will do what is best for the future direction of iNat.
Iām not sure what type of research you are referring to but I donāt think most field biology research gets into the ābig numberā category. And much of it deals with ranges, so misidentifications that erroneously represent where it occurs are problematic.
And, I agree, that automatically agreeing is a problem. Even if an expert IDs it, before agreeing with that ID, it would be best to investigate to see if that seems right. That way you learn a little more about the organism and experts arenāt infallible.
For sure, I understand that this makes the data less scientifically valid, but I try to follow some kind of process for Agreeing other usersā identifications of my observations.
Firstly, what do I know of the identifierās expertise? Iām not going to doubt a beetle identification by borisb, for example.
Then, what was my original impression? I also realise this is not always watertight, but I try to do some research into species I am unfamiliar with to get some idea, but then leave my own identification for the observation at āClassā, āOrderā or āFamilyā level if I have no real knowledge to back this up!
Next, I (almost) always take a look on-line to see how the new identification compares with information available from web resources. Often, I have no idea how close one species may be to another nor how to differentiate them, but Iāll Agree anyway if it looks right!
I think for new users itās very difficult to know if other usersā identifications are accurate, and the desire to learn new species and find out what one has seen can override scientific integrity, but this also leads to new learning, and corrections are always there to be made.
On the other hand, I also agree that any scientific use of the data should require extra validation by the scientist, depending on the context. Observations of Eurasian Magpie, for example, can surely be taken at face value without too much scrutiny, whereas observations of Grey Dagger moths should be taken more cautiously.
Thereās another embedded issue here. As it stands now people can delete their account and remove all their IDs. If they do this and you havenāt agreed to their ID, you lose it completely. So unless they somehow make the community ID retain the deleted ID you have to do this if you want to ensure the ID remains. So Iām tornā¦ but I do this sometimes when I otherwise wouldnāt because of that factor.
Me too. Also, I think new users especially sometimes think that getting to Research Grade is so important that they want to reach that goal as soon as possible, or help their friends reach it. Maybe de-emphasizing Research Grade would help.
I came here with a similar reply in mind. It can be hard in some cases to tell the difference between thoughtless reactive āagreeā button hitting, vs. a quick review of an unfamiliar ID provided by another user and subsequent willingness to confirm.
This morning I added an observation of a water bug- I am no Hemiptera expert so I left it at the infraorder for āWater Bugs.ā Another user quickly added an ID to genus, so I clicked on that genus, browsed through the info and species, checked a couple related genuses, and feel comfortable confirming that the observation belongs in the suggested genus and is probably one of just a couple species in that genus). So I hit the agree button.
All of this really just took ten minutes or so, so if I hadnāt left additional commentary on candidate species along with my āagree,ā it would be easy to think that I had just automatically hit the āagreeā button. So while I agree that there is a widespread problematic behavior here, thereās probably also a whole spectrum of increasingly diligent behavior thatās not necessarily possible to tell apart from the problem behavior at a glance.
What is the recommended way to disagree with an ID suggested by an observer? I ran into this quite a bit recently.
Is there any way to encourage comments to show why one agrees with an ID.
I had the window come up a few times that questioned if I was sure about the ID or not. It made me pause to go over why I either Agreed or Disagreed.