Apologies if this is posted elsewhere and I didn’t see it. I’ve seen a couple of examples of what I call overzealous identification, where it would appear a user has gone through and just agreed with the identification of every observation that isn’t research grade for certain taxa. I’ve seen it with two accounts in particular, both of young users who can’t possibly identify many of the observations they’ve identified, either because of poor quality photos or lack of identifying features. Both observers focused on birds but also “identified” other taxa. One has almost 57,000 identifications and the other has over 14,000 identifications.
They have gone through and made many unidentifiable observations research grade. As a scientist who uses iNaturalist observations in my research I find it very frustrating to encounter these observations, in particular on GBIF. I wrote to one as nicely as I possibly could but never heard back and the observer continued identifying.
Is there a way to fix this? Is there a way to remove a user’s ability to identify? Can curators blanket remove spurious identifications made by particular users?
Yea, that has been a problem for a while. Unless they are doing obviously bad IDs that are blatantly wrong or seem malicious (Ivory Billed Woodpeckers?), there isn’t much done. When you said you wrote, did you comment or send a direct message? Comments are much more likely to fall through the cracks… When I notice users making a lot of mistakes, but keep trying to do a lot of IDs, I try to send them a direct message.
When I notice users bulk identifying, I try to go through what I can and push back things I know. A user (maybe the same one you are talking about - animal##) recently was misidentifying a bunch of blister beetles by blindly agreeing with things.
It is unfortunate that these do go over to GBIF, which I can hope people take with a grain of salt and can see the data is sourced from iNaturalist and thus likely to have a higher error rate than normal museum curation.
I don’t have a fix or good answer to your question other than to say I feel you, but I don’t know of a solution that would work to most scientists who have a limited time.
There was a discussion about a very similar issue a while back on the Google group. If I recall, the consensus was to send a note to iNat staff if gentle nudging in comments or messages don’t seem to curb the behavior. In at least one case, that intervention seemed to help.
FWIW, I’ve reached out to a couple young users who fit this description and ended up having productive conversations with them that more or less fixed their “overidentification” issue. It’s definitely a delicate issue, as the last thing I want to do is discourage excited young naturalists.
To answer another of your questions, curators have no authority to edit or remove identifications for any reason. To me removing the ability of a person to do identifications goes against what the site is designed to be. If that is the case we may as well just restart the debate about “expert” accounts and giving “experts” overwhelming weight on ID’s.
Thanks for the reply, I sent a direct message and tried my best not to discourage one of the users, and I clearly didn’t, but I also didn’t get through to them.
I also tried pushing back by giving the observations a broader id, but then I just got complaints from the original users who were excited that their blurry blob or single black feather made it to research grade.
I probably don’t see as much of this as identifiers in other taxa, but I feel the pain strongly. What I think is saddest about it is that the less knowledgeable/qualified a person is about the taxa they are identifying, the less likely they are to check back on the observations they affect. Since I started identifying on iNat, I have always followed the observations I add IDs to. Yes, I get a ton of notifications, but I want to see every subsequent ID and every comment. This is how I learned so much in the beginning and how I stay connected to the community and keep on my toes. It’s so hard to watch people come blasting through with wrong IDs and never look back.
Sorry, that doesn’t add anything helpful. I think sending a DM is a good idea, since @ comments clearly aren’t getting through. I wonder if we could add something to the “Identifying Observations” or “How to Become a Better Identifier” about following observations after adding an identification. I know it’s not for everyone, but I think it’s a must for beginning identifiers.
I encourage you to keep doing this, to the extent you can keep up with it. Not specifically to push back, but just to add what you feel is the most appropriate ID for the evidence, and an informative comment explaining why. Then just ignore the complaints (unless you see an opportunity for further clarification). The more that you (and others) can lead by example like this, the best hope there is for eventually educating folks like that about expectations on iNaturalist. If you think it would help, you could also point them to resources like this or this.
If you ever feel like a user is showing a pattern of intentionally wrong, malicious, or joke IDs, though, definitely contact help@inaturalist.org so staff can look into it and see if account suspension is warranted.
In those threads there were a lot of suggestions about how to deal with it, and some debate about whether we even need to (is the issue big enough to matter?). There were problems with or arguments against just about every suggested solution proposed in those threads. I think it’s fair to say there was no consensus reached, and there probably won’t be in the near future.
There are two main problems with people playing a numbers game with IDs. First, observations can get to RG (and then into GBIF) with incorrect IDs. Second, it can waste a lot of time—it’s very quick and easy to click the agree button, but it takes ages to trawl through and correct IDs. And unfortunately there will always be a few people who ignore the gentle approach.
As a scientist, I definitely share @jmaley’s frustration with this, and would like to see the data gathered in iNat as accurate and useful as possible. However, it’s clear that the site’s primary mission is to engage as many people as possible with the natural world. Certainly we can and should continue to try and engage and inform users so as to improve data quality—no question about that. We’ll never reach perfection, however. So I guess the bottom line is that if anyone wants to use the data, the onus is on them to check/filter/prune etc until they’re happy with the quality of the sub-set they’re left with.
I too continue to follow everything I participated in. As a result of some folks who were kind enough to point out the periodic error of my ways, I too have learned a huge amount - e.g., species that cannot be IDed to lower than genus with only pix (e.g., green tree frog and Copeland’s frog, both Hyla). I rely on these, as otherwise I get paralyzed by not knowing what I don’t know and being afraid to try anything outside of a set I am absolutely positive about…
I also make a point of trying to ID for many different people to prevent setting up an echo chamber!
So after that long unhelpful ramble, I’d just say Polite Persistence - post a more appropriate ID with the brief explanation of why it’s better. Hopefully in time, as RGs turn back to NIs, they will pick up on the hints.
In my day life I’m a data geek so I too feel your pain!
Variations: I see some people overidentifying geographically: they want to see all observations as RG from their countries/region. Some are following well-known scientists and confirm everything they ID. Some confirm everything inside one particular field (like birds…). Some want to confirm the rarities…
I fear that such behaviour may demotivate the more prudent identifiers …
What about adding a “Disagree” button that provides you with a short list of broader (more inclusive) taxa on which you just have to click on? This would make correcting things much faster, nearly as fast as agreeing.
It would be best if scientists would re-evaluate each observation by themselves before using them, but very large analyses involving many observations (e.g., multiple taxa and/or global) might not be possible if this is required?
Question: Is it possible for scientists to get user behaviour data (to let algorithms filter out problematic identifications based on known behaviour patterns)? Or is this not possible due to data privacy concerns?
Adding a dissenting ID is the disagree process. Trying to link it to autogenerating a list to choose from just reopens the same computer vision issues constantly brought up on the site. I don’t think there should just be a ‘you are wrong’ process, you should have to express what you think it is if you disagree.
Simply disagreeing with no alternative proposed breaks the algorithm to generate community ID unless you simply say once disagreed that vote is disregarded. If that is the case I also fear it being a tool for vandalism.
The principle on the site is each person gets one vote towards the identification of a record, not each person gets one vote and the ability to delete or invalidate other votes they don’t like or disagree with.
This wouldn’t solve the problem entirely but I think it would help a lot. Some have expressed concern that this would create a huge bolus of suddenly NI obs, but I think it could be applied as of a given date. In particular it would prevent an observation that receives one ID that is confirmed by the observer from becoming RG.
I know of a user who only confirms research grade observations and doesn’t id non research grade observations. I noted the person has confirmed a few mis-id research grade observations and he is in the top 3 identifier position :-(
Recently I’ve had a user acting like an agree-bot running through hundreds of my observations with a single species ID in the needsID category, and not once have they corrected a bad ID or improved an ID. When I correct a bad ID (the first one, mine), there is no evidence that they return. I now have hundreds of observations that need a confirming ID but got an agree-bot ID instead, and so hundreds of observations have a low-quality RG status. I would like a better chance for my IDs to be looked at and find misentries and misidentifications, but that chance was reduced by the user, since the Identify tool search filters excludes RG observations by default.
Is there a way to search for my observations that have been identified by a particular user and automatedly check the “can the community ID be improved = yes” option?
I would say keep messaging this person/ these people. Stay polite, but get a little firmer and more explicit each time. It’s important for them to know that iNat should not be mis-used.
I think a lot of people don’t understand that the data on iNat is often used by scientists, and therefore needs to be as accurate as possible. Some people think that iNat is just a game.
I think the leaderboards are the issue. Maybe a pipe-dream, but it would be good to have a way to identify the experts for a taxon without encouraging this type of behaviour. In fact, I’ve found that many of the most knowledgeable people only have a few IDs, because they are usually only called in as pinch-hitters to ID the really tough or interesting ones, while some of the top identifiers have only ever agreed with others.
If this continues to be an issue with a single, unresponsive user, you might consider the “block” feature. I’m not sure this is one of the intended use cases for it, but I see that behavior as essentially spamming your personal records with bad IDs and justifiable.