Dealing with Account Deletion


Following up on this long thread in our Google Group.

There has been a lot of discussion over the negative consequences of account deletion, mainly:

  • When observations are deleted, that occurrence data is no longer available on iNaturalist.

  • When an observation is deleted, the IDs and comments on that observation that have been contributed by other users are also deleted, so the result of any hard work/research done by those users is no longer available.

  • When an account is deleted, any IDs made by that user are also deleted, possibly removing Research Grade status from those observations, as well as any other improving IDs.

  • When comments are deleted, any conversation involving those comments can become incoherent.

iNaturalist will always allow users to entirely delete their content from the site, both as a matter of principle and because laws like GDPR would not allows us retain data by users who no longer want us to.

However, that doesn’t mean we can’t give users the option to leave their iNat contributions on the site if they do decide to delete their accounts. I’m planning on writing a spec about this functionality for the team to review, but I’d like to get some feedback from you all about how this might work best. So, right now, I’m thinking it will involve:

  • A pop-up that shows how many IDs, observations, comments, etc will be deleted from iNaturalist if the users chooses to delete them all.

  • An extra confirmation step that will prevent accidental deletion and slow down hasty decisions.

UPDATE: This has been added on March 14, 2019 (thanks @kueda and @carrieseltzer!) Here’s an example of what users will see:

*The option to leave all of one’s data on iNaturalist, either anonymized or with their username attached to it.

  • The option to choose which type of content should be deleted or remain, such as Observations, IDs, or comments on observations (I’m not sure comments on flags or taxon changes would be possible and/or too complicated to get across), either anonymized or with their username.

How does that sound? We would then need to have some way of indicating the user is no longer on iNat, and choose how to display anonymized data, among other peripheral changes.

As to how account deletion affects GBIF, I’d rather not have that be part of the conversation here, we can start another topic about it. Hoping to get an FAQ up soon about exactly how GBIF and iNat data interact so everyone will be coming from a common understanding.



I admit i find this issue a bit frustrating in particular in regards to IDs. It seems odd to me that one should be able to revoke the act of telling someone else what an organism is. That doesn’t strike me as a privacy or data control or intellectual property issue at all. I am not understanding what principal would require doing this, though perhaps this thread isn’t where that should be discussed.

In particular, i wonder if it would be possible to retain the ‘vote’ on the observation when someone adds an ID, even if they delete their account. To me that seems reasonable, can that data law really require that people are able to subtract votes they make on sites? Research grade is re-attainable but I don’t think deleting an ID in this way should result in a loss of the species tag.

For instance, if I add an Abies, and user AcornSquisher identifies it as Abies nebrodensis, i typically would not ‘agree’ with the ID if i am not familiar with the species (maybe I was on a trip to Sicily but don’t have a clue about firs there). But now, if AcornSquisher decides to delete their account and delete all their IDs, will my observation go back to ‘Abies sp’? If so, i am inclined to agree with everything anyone ever identifies so I don’t lose the info if they quit. However, then that makes them all research grade unless I vote ‘needs further ID’ and that in and of itself has problems too.

In any event, I think what you are proposing is a good idea if it’s necessary to allow deletion of IDs, but I think you should consider retaining the ‘vote’ in some way either way if only in the community ID.



I don’t want to speak for everyone on the team, but what was written came from consultation with some of them. I think people should have the ability to be “forgotten” online, meaning they should be able to remove, at the very least, any public trace of theirs from a website, whether or not there’s a law requiring one to do that. And that would include things like IDs made on iNat. But as this isn’t really up for discussion, it’s best to move on with other suggestions/critiques.

I suppose the scope of this is also important and I don’t think content deletion is a big enough of a problem to make this a huge issue. I’ll try and crunch some numbers to see how many observations have been affected but my suspicion is that it’s relatively small. I think a better use of time and resources would be to understand why people leave and which improvements can be made to reduce such occurrences. But that is a different topic of discussion, don’t want to derail this one so soon. ;-)



I like this. I’m just wondering whether at least a link to a description of other protective measures that could be taken instead of deletion (such as blocking someone) might be helpful to someone’s decision without causing undue pressure.



Ok. Fair enough. I don’t really understand but it doesn’t sound like something I have any control over. I’d just ask you to really think about whether the “vote” - just the number in the community ID algorithm - without any name or anything - could be considered a “trace”. To me it’s hard to imagine how it could be. Comments and of course observations are another thing. And you are right. It probably is a minor number of observations and IDs.

I also think it’s inportant that people realize that you can’t really rely on anything you put on the internet ever going away. There are archives, screen shots, etc. I know you are aware of that but I just think any inat messaging promising that isn’t telling a full story. Once you post something, it’s in some ways totally irrevocable. Which is scary.

1 Like


Just a thought about factoring in the scope of the issue in terms of how many obs are affected…
since we can’t predict the future impact of inat nor the future behavior of users of the site (or really the internet generally) it seems that this concern should be weighed less or at least considered as variable. Who’s to say that ten years from now those of us (like myself) who have left sites like FB or its predecessors might not be ultimately fed up with the loss of privacy or some other concern and leave this site in either impulsiveness or planned withdrawal? Potentially, this could leave holes or other messiness in the data that could be avoided and could be significantly greater in scope than present day. In my non-tech non-legal opinion it would seem that finding a solution to make anonymous the valuable information from deleted users would be an important consideration. ebird has anonymous accounts, can this be looked to as an additional offering to folks considering deletion?

As for the other suggestions, especially from paloma about offering other non-deletion solutions, I agree. I would also repeat what was said in the google group (although I can’t remember who said it) suggesting a cool-down time or a warning that account deletion will go into effect after a period of time during which the choice to delete can be reversed. (as a sometimes impulsive type myself this would be a welcome option)

As with all of my well-intentioned comments I’m just trying to be helpful but if I don’t display a proper grasp of the subject at hand please either kindly correct me or just ignore my attempts.



Given that users can delete all their information if they want to, and that there is some concern about losing IDs made by such users, I’m curious whether it is okay in certain observations to either restate the ID (without the user’s name) in either the description section or in a comment, as a way to preserve the possible ID in the event of deletion of the actual ID. Or is there something unethical or otherwise problematic in doing that?

1 Like


I think a ghosted version of the ID is a good idea, although it could be problematic if the reason for the delete is because a relatively large number of IDs have been made in error. I’ll cite the case of my Hackfalls account as an example. I opened the account, bulk uploaded about 3000 records with IDs, and then encountered considerable taxonomic issues with those observations. I was going to have to alter many of them, and it compromised what it was that I was trying to achieve, so I opted to delete the account as a way of removing them all, so that I might start again with a better data source. Granted this is IDs on observations that are going with the account, so there would be no ghosting of them, but perhaps similar needs of deleting might occur. I could imagine a situation where a rogue student went crazy with ridiculous IDs, and deleting such an account and wiping as much trace of it would be hugely beneficial to the community as a whole! Of course, this last one would assume the account was somewhat under the control of an administrator or teacher…

I am very much FOR retaining the right to delete, but I am even MORE FOR reducing the impact of it. If it meant the only way to prevent the confusing dialogues, for example, I would gladly surrender the right to delete. But I would like to think a reasonable effort to retain that right was made. I volunteer my time and energy to iNat, very much BECAUSE it doesn’t come with a contractual requirement behind it.

I really like the idea of having the option, even if it was a default choice as well, of having the data retained and made annonymous. In situations where one wanted to scrub the data completely, as in the examples I cite above, you could change the delete settings to fully remove it. It would also be good to retain the identity of the account as an option, so that in the case of deceased persons, the account could be made to reflect that the person will not be changing their ID’s anytime soon.

Perhaps in the case of a deleted account that has persistant IDs, or even an abandoned account or one from a member that has passed away, other members could mark an ID as being problematic because of it’s persistance. Maybe 10 votes to remove a problematic ID could have it removed or at least withdrawn. Just throwing some ideas on the table, might trigger other useful ideas :slight_smile:


Deceased observers - and a deceased symbol on their icon

Here’s a concrete example of what i am concerned about

i don’t know anything about that organism, but am I forced to ‘agree’ with the ID or else risk losing track of what it was identified as if roman_romanov decides to delete their account? I don’t want it to be research grade based on my ID but if i mark it as needing further ID, it means i have to do that every time this happens, which is common, and it doesn’t turn off automatically after I get an ID so they get stuck ‘on’.

Not sure i am explaining it right?

1 Like


I think a way of being able to see the ID history of an observation, or at least the history of what the Community ID was (not linked to any usernames) would solve this issue. Alongside being notified if the Community ID changed because of an account deletion.

If there is a deleted user or deleted ID, an unobtrusive small grey bar could appear in its place in the “Activity” section saying: “A user has deleted their account/ID. Click here to view the full Community ID history for this observation.”

Because the Community ID is not added by any individual, but is a sort of consensus, I don’t think preserving it would raise any problems of data sovereignty.



Your linked observation is the type of observation I was asking about, charlie. If you are concerned it might be deleted in the future, can you just state the suggested taxon in the Description or in a comment of your own, instead of agreeing with it? I just wasn’t sure that would be considered okay, which is why I was asking.

1 Like


i could. In reality, with so many tens of thousands of observations, I’m probably not going to do that for all of them. And even if I did, i’d then have to somehow know the ID got deleted and then go back and manually change them all. Ugh. If i’m going to bother with that I might as well just agree with the ID. I don’t like to give them research grade, but I am hoping there’s another solution so i don’t have to figure that out.



But does EVERY observation have to count? If the ID is lost, does it make the observation useless? It will appear back in ID reqd, so will hopefully attract specialist attention from someone else. If they are no longer present to defend or explain their ID, can you even trust that it is right? Can you trust that they will CONTINUE to be right. iNat is growing, and for every specialist lost from deleting, we gain (how many?)… heaps more. Sure they might not be specialists in the same field or taxon, but it is just a matter of time before that gap in expertise is filled again.

If that identifier had NEVER put that ID, would that negatively impact what you are doing? If so, then what other IDs are NOT being applied because some other person NEVER joined iNat and applied their ID? How will you even know? And does NOT HAVING THOSE IDs have the same critical impact on what you are doing?

This issue over the loss of an ID is the “loss aversion” (bad terminology) that has been referred to previously. Think of it as running up a slippery slope, where for every 10 steps you take forward, you slide back 1 step. It is the NETT progress made that is useful. You can be upset about that loss of 1 step distance, and ideally if you can find a way to go 10 steps with no slide backwards, then terrific…

A child at the beach is likely to be happy, and if he is given an ice cream he will be happier. If he drops that ice cream in the sand, he will be unhappy! That arises out of being focused solely on the loss. If you can help the child to see that they have had a wonderful day at the beach, played with their friends, saw some awesome fish caught by the fisherman on the rocks, and that it was unfortunate that they dropped the icecream, but they can have some cake when they get home, then they see the nett position and will be back to being happy again. If we change the experience of the child to being followed around with a big tray under his icecream trying to catch it when it falls, then we are kind of destroying the whole beach going experience for him! If we can look at the bigger picture, and accept SOME SMALL LOSS while making GREAT GAINS, then we have a good outcome.

I think we are better off focusing on the nett gain that iNat as a whole represents, rather than the (very few) IDs lost in the sand when someone deletes. If we can come up with a way to minimise the impact, then that is good. A single data point being reduced in quality (due to a species level Id becoming a genus level ID, for instance) is far less impactful than a conversation that might be happening on that observation about the ID, which can potentially contribute to the education and up-skilling of countless persons who might then go on to be able to make those IDs elsewhere. The whole give a man a fish (ID) and he’ll eat for a day, but give him a rod and teach how to fish (link to literature and explain differences) and he’ll eat for a lifetime… This is where I think iNat rocks… not only have I had a large number of observations ID’d by experts, but they have been sharing with me their knowledge and giving tips on how to differentiate the whole time. I look back at what I knew when I first joined, and what I can ID now, and the difference is astounding. And I am passing on that knowledge to others, and the discussions around the taxa are going viral. One person saw sea swalllows up in Northland, and people all over the world get to see it. Many of us that have seen that observation can now identify sea swallows, should we ever encounter one in real life. That still stands even if the original identifier deletes their ID! That conversation is a growing thing, and we now know there are two very similar looking sea swallows, and how to tell the difference.

There are always going to be knowledgeable people that make IDs, but are just too busy in real life to follow them and change them if new information comes to hand. That is where we, as active participants in iNaturalist, can “add weight” to the IDs of knowledgeable people. If someone I have great confidence in the IDs of, makes an ID, I can agree with them to give their ID more weight. I can “drill down” from time to time to validate my confidence in them. Others can do so with my IDs as well. I used to get annoyed by observers agreeing to an ID as soon as it was made (I mean, surely if you thought it was that you would have put it in the first place!), but now I tend to look at that as just something that is going to happen. Unless you explicitly asked new users NOT to do that, it is going to happen. The concept behind community ID kind of takes that anomaly into consideration. So I suggest to you, if someone IDs your obs, and you have confidence in their IDs (to the point you would consider the loss of the ID as being an impactful loss) then add your agreement. It has a twofold benefit… it adds weight (in a system which doesn’t differentiate quality of identifiers) and protects an ID from the effects of deletion.



Well, I described a specific scenario and it’s one that has imho an easy and fair solution and in that scenario, if the person leaves, I lose the ability to see what some things I found were identified as. I guess another option would be notifications and options to add the ID myself if that person left. Remember, I’m not talking about research grade even here. Just IDs that are totally lost in the scenario I describe above.



I like the idea of being able to see the history and getting a notification, but to help with an observation like charlie’s example it would have to be the ID history rather than the community ID history, because, according to charlie’s observation, it takes two IDs for there to be a community ID.



to my earlier point: we can’t know the scope of an issue in the future. the loss of a few community IDs may be one thing but the loss of observations (which we can’t predict the nature of i.e.: possibly rare or otherwise uniquely valuable to the data on the site) or useful interactions (like the swallows example) could be an issue in the future after a decade of further use. Generally, while I appreciate the concept and point about loss aversion, I disagree that that is the prominent concern just simply put. I think it’s a little more nuanced based on my reading of other comments and to me just seems like there’s an opportunity to deal with not just a few losses now but a potentially larger and more disruptive situation going forward. Also, I don’t see why seizing the opportunity to proactively keep data, community trust, site integrity etc. in mind with an eye to deletion, regardless of the current issue’s scope, wouldn’t be a positive activity. Sorry if I’ve misunderstood the comments.



charlie, I am coming around to thinking you are right about just agreeing to identifications you would not want to lose. There are many, many observers on iNaturalist who didn’t know what the ID was when they posted the observation but then immediately agree to IDs made by others. Their observations won’t lose possible IDs in the event an account is deleted–so why should yours? And in observations such as your example, they won’t become Research Grade at a higher level anyway at this point. I usually don’t agree to these higher-level IDs I know nothing about either, but since the observation ID and Needs ID will stay the same if you agree, why not? I think from now on I will agree to IDs on my observations that are higher than species just to preserve a record of what the organism might be in case the identifier’s account is subsequently deleted.



I like all of the bulleted suggestions in the top post and would be perfectly satisfied with them.

Besides “right to be forgotten”/privacy laws, account deletion could be important for someone’s physical safety, like not letting their newly-acquired stalker see years of their geotagged whereabouts (!). While it’s often impossible to fully remove all trace of yourself from online, it’s certainly possible to obscure the traces and make them much harder to find.



just to clarify, at this point there is no temporary hold that can be put on accounts, right? I’ve just now found myself wanting to leave inat for the first time after having my first ever unpleasant exchange with someone in almost 2 years on the site. I am trying to “know better” and get over my emotional hurt and tears as I’m aware of how sensitive I am but I do find myself wishing I could go incognito or put a hold on my account. Because the nature of the comment probably wouldn’t rise to the level of reporting the user or blocking them (anyone else would probably roll their eyes and write off the user as nasty while I feel beside myself and no longer want to be vulnerable [aside from communicating the situation here] ) I am wondering what advice there is in this situation; what action can I take short of deletion? I know I’m not supposed to use humor (which is possibly what upset this person although I have no clue) but up until now my instincts have been pretty good for when that’s appropriate and I’m also not interested in being a robot version of myself. I’m sensitive, somewhat humorous and verbose and I am really trying to offer as much of myself to the community as possible, and clearly failing in some respects. I put this comment here because of the relation to the issues of deletion but I will delete or move it if that’s recommended.



Well, i for one have enjoyed having you on the site and hope you stick around. You can send me a message and I can try to help. If nothing else I can give you context if the person has caused other problems, you can always email admin too.

There is no rule against using humor, unless it’s fake IDs.

(to clarify I am not an admin just another user/curator)