70

Many users no longer feel safe participating here and don't feel comfortable with SE having data about them. While there is a way to delete your account which will result in all your posts being dissociated and your username replaced with a generic "userNNNN", that isn't enough.

Your old username will still be visible in comment mentions (@terdon) and can still be found in answers ("As terdon mentions in their answer...", linking to the now dissociated answer). This means that users, and especially those who have chosen to participate using their real name, are still not safe.

If someone has been made a target for trolls, deleting their account won't help them. Their real name will still be visible in the comments, or in other answers. They will still be targets. Given the current extremely tense situation, this is a far greater danger than it used to be. It is essential that SE give folks a way to entirely remove themselves from these sites.

I am not a lawyer, but I understand there may also be a legal obligation to do this, at least for people affected by the GDPR. Deleting my data should include deleting all mentions of my name that can link me to my activity on the network.

But whether there is a legal obligation or not, there is clearly a moral one. The current situation means that this network has put its users in danger, so it should provide a way for those users to protect themselves.

So, please provide us with a way to completely remove all signs of a user's presence on this site. I understand the technical challenges in this, and it may even need to be a manual process, but this needs to be done. Or at least done better than it is at the moment where it is trivial to find out the original username in a page where there have been multiple comments.

34
  • 9
    Whenever a comment is deleted, we always get told that they are transient and we shouldn't worry about it too much, so, in this case, surely that would be the justification for purging comments related to a person too?
    – Script47
    Oct 19, 2019 at 15:17
  • 16
    This is why you don't use your real name online.
    – user474678
    Oct 19, 2019 at 15:31
  • 18
    In anticipation of an unproductive point that may be raised, I will note that whether the users are correct to feel unsafe is completely beside the point. Users should have this option regardless of how safe they are and feel.
    – Solveit
    Oct 19, 2019 at 15:32
  • 11
    It wouldn't surprise me if the GDPR requires this anyway.
    – user102937
    Oct 19, 2019 at 15:34
  • 17
    It will always be trivial to find who posted what on the majority of pages on the site, regardless of how much we scrub off our site, because of scrapers and other archives. There's nothing we could possibly do to prevent that.
    – animuson StaffMod
    Oct 19, 2019 at 15:37
  • 12
    @Joeri Actually the GDPR law makes explicit exceptions for "would be extraordinarily difficult to complete" things.
    – animuson StaffMod
    Oct 19, 2019 at 15:38
  • 4
    I don't think anyone's asking SE to remove all traces of their behaviour from all existing records. Just the ones that belong to SE.
    – Solveit
    Oct 19, 2019 at 15:39
  • 9
    @animuson of course you cannot be held responsible for data stored elsewhere. But you can, and should, be held responsible for data stored here. No, this won't make the users completely safe again but anything that improves their safety (and more importantly their feeling of safety) on this site should be encouraged. So the fact that the data may be available elsewhere is irrelevant. After all, you could use the same argument to say there's no point in dissociating. So I really don't understand why you would even bring that up.
    – terdon
    Oct 19, 2019 at 15:40
  • 4
    As for "extraordinarily difficult to complete", starting by removing the name from comments where only one user with that name has been active is trivial. As I said, anything would be an improvement, @animuson.
    – terdon
    Oct 19, 2019 at 15:41
  • 2
    @terdon We do use that argument to say there is no point in disassociating, all the time.
    – animuson StaffMod
    Oct 19, 2019 at 15:41
  • 8
    @animuson then stop offering dissociation! At the very least, stop suggesting account deletion as a valid way of ensuring that users can remove themselves from the site. It sounds like you're telling me that you only do this to make people shut up and don't care about the legal or moral ramifications of it. I find this very hard to believe.
    – terdon
    Oct 19, 2019 at 15:43
  • 3
    @terdon Well, it's true. We do it because we're required by the license, and we explicitly warn users of all that information in the meta FAQ we link people to before disassociating things. As long as they understand that disassociating doesn't really do anything in the grand scheme, it's their decision whether they still want to. They've been told it's pointless. It sounds like you'd want us to emphasize that more somewhere.
    – animuson StaffMod
    Oct 19, 2019 at 15:46
  • 6
    @animuson well, ideally, I'd want it to become less pointless :) There is a very real issue of safety here. Users have had their real name shared with the press and this has led to real world consequences. They need to have a way of removing at least the most obvious mentions here to protect themselves. And I really do think (but IANAL, of course) that there is a clear legal requirement at least for users covered by the GDPR. But yes, if you're not willing or able to improve this, then make it very clear that there is no way for users to delete themselves from the site.
    – terdon
    Oct 19, 2019 at 15:49
  • 2
    To expand more, we want to inform users upfront that their name can still be found, because we constantly got back replies with link to other sites with "why does my name still appear here!?" Expecting us to be able to do something about it. People need to know their name is still out there and just having us remove it in one place is not really protecting them. That's important information. And as an aside, we do advise users to search their name and flag comments that still contain their display name if they so choose.
    – animuson StaffMod
    Oct 19, 2019 at 15:52
  • 12
    I think GDPR does not cover references. I don't see how GDPR has the right to change my comments. So if I do an observation, I see you, I mention you, then I don't think SE has any right to alter my comments based on the GDPR. My freedom of speech trumps your GDPR rights.
    – Joeri
    Oct 19, 2019 at 15:52

4 Answers 4

16

I believe this can be solved quite reasonably without causing issues if we're still going by the philosophy that comments are transient and their deletion should not be too worrying.

When people receive replies (@Script47), you get a notification which if clicked takes you to the exact comment. This tells us that the comment ID is being stored somewhere with the details of who was notified.

So, essentially, we could just delete all comments associated to a user (using the notifications link) or replace the usernames within those comments.

Though this won't catch all instances, it would catch a major instance that you mentioned.

5
  • 1
    Agreed. This would at least be a good start.
    – terdon
    Oct 19, 2019 at 17:58
  • 5
    Every comment has a "reply to" thing internally where it marks who it's a response to. It's visible to moderators. So yes, that is very much stored.
    – Mithical
    Oct 19, 2019 at 18:03
  • 1
    @user58 yep - I was just explaining for the non technical users who visit meta too.
    – Script47
    Oct 19, 2019 at 18:07
  • 1
    But obviously the comments pinging a user can't be deleted by that user.
    – DK Bose
    Oct 20, 2019 at 1:06
  • 1
    @DKBose of course, this is intended for the system to accomplish.
    – Script47
    Oct 20, 2019 at 12:32
11

One thing that shouldn't be too hard to implement and would go a long way towards addressing this issue, is to focus on pages the deleted user has been active on. So, if user Jon has posted an answer, and Jon is now user123 because they've deleted their account, then all mentions of @jon in the comments should be changed to @user123.

Yes, that might mean some other Jons are affected and their name erroneously changed, but that is the lesser of two evils and a small price to pay to make the real Jon safe again. The good news is that even in cases of exceedingly common usernames like Jon, chances are that most mentions will be of the relevant one since it is less likely (although far from impossible) that two completely separate Jons have interacted with the same post.

Alternatively, remove all Jons where there has clearly been only one user with that name, and flag the harder cases for manual review. Then a human can read through them and figure out which to delete.

Finally, even if this isn't perfect and still lets some mentions fall through the cracks, that would still be a significant improvement over what we currently have. Dissociating a post from its author but leaving dozens of comments that refer to the author with their original name is a band aid. Any improvement would be a step in the right direction.

10
  • This won't be enough. For example, say Jon Snow deletes their account, but in some comments people have referred to him as Lord Snow (not pinging them, but referring to them nonetheless). A partial (even fuzzy) match of the username would have to be purged.
    – muru
    Oct 19, 2019 at 15:28
  • 6
    @muru yes, of course. Partial matches should ideally also be caught. But this is just a proposal for a small step in the right direction. Any improvement would be welcome and I would rather see a small improvement than no improvement.
    – terdon
    Oct 19, 2019 at 15:29
  • There are thousands on Jons that could be on the site, some even in the same comment threads. It would be unrealistic to expect this to be possible.
    – user474678
    Oct 19, 2019 at 15:32
  • 4
    @JL2210 yes, that's why I used such a common name as an example. And that's also why I specified that they should first focus on the threads where only one Jon has been active and flag the rest for manual review. In any case, even only getting unique names such as "Goldifrapps Fliperton the third" would be an improvement. Anything would be an improvement.
    – terdon
    Oct 19, 2019 at 15:34
  • Yep, that's what happens then I don't read.
    – user474678
    Oct 19, 2019 at 15:36
  • 2
    [...] then all mentions of @jon in the comments should be changed to @user123. ... except that causes a massive problem for common usernames. Take "john" for an instance - there's literally over 1900 pages of users matching that search on SO.What if there's two in one discussion that both are either pingable by @John or have the exact same username? There's no way to sensibly disambiguate it. Or what about people like me, who have changed their username multiple times? (I've, in the past, changed here on meta just for internal fun in a chat room on SE) Oct 19, 2019 at 16:01
  • 1
    Catching all the cases and figuring out what pings reference what users is a pain in the ass with SE's current mention model. If they built off something like Discord (<@veryLongNumericUserIDHere>), that would be possible, because each ping refers to exactly one user, not 1900 pages of users. Not to forget the caching element - it would require to edit each comment, which causes a mess. Not to forget about non-pings and convenient overlaps between examples with random names and usernames Oct 19, 2019 at 16:02
  • 3
    @Zoethetransgirl I'm not sure what point you're making that hasn't already been raised. As I said just above, I know, and that's why I chose "Jon" and why I suggested that a good start would be focusing on pages where only one Jon has been active and flagging the rest for manual review. As for the SE model, then maybe they should change the model. But, as I said multiple times, even a small step such as dealing with the unique usernames would be a good start.
    – terdon
    Oct 19, 2019 at 16:04
  • 1
    If this is done, I swear, someone is going to mess so much stuff up. It’s one thing when @jon could point to one of several people, it’s another when @section could point to a person or code.
    – Laurel
    Oct 19, 2019 at 16:36
  • 1
    @Laurel not necessarily, please see my answer.
    – Script47
    Oct 19, 2019 at 17:52
1

GDPR does not apply to Stack Exchange as it's not an European entity, but if it wishes to do business in the EU, then it will have to comply.

GDPR does have the right to erasure enshrined in it, so I can ask for my data to be removed, and it would have to be done.

I think the two points that are relevant to Stack Exchange are:

  • you are relying on legitimate interests as your basis for processing, the individual objects to the processing of their data, and there is no overriding legitimate interest to continue this processing;

  • you are processing the personal data for direct marketing purposes and the individual objects to that processing;

They may be able to keep my questions and answers, but my username is used by me elsewhere and thus can easily be described as personal data (i.e., something that could identify me as an individual). For example, IP addresses also fall into this category.

4
  • 3
    If your username is mentioned in a comment or question/answer I don't believe it'd be covered by GDPR anyway. GDPR only cares about data used for processing, and SE isn't crawling questions or answers for peoples personal infomation. Also arguing it needs removing because someone else put it there fails under There are some exemptions where the company or organisation can refuse your request. These include: the right of freedom of expression and information Oct 19, 2019 at 17:25
  • 10
    It does apply to entities outside the EU. Oct 19, 2019 at 18:15
  • @djsmiley2k-CoW not so, this bit of the GDPR applies to personal data that can ifentify someone. I put a link in for you that has a section describing what "personal data" is.
    – gbjbaanb
    Oct 19, 2019 at 19:29
  • 4
    Yes, but it's not personal data that SE is 'handling' as per GDPR. - Obviously the true test is in a court of law, as the wording is very unclear. Is facebook responsible for every time someone is mentioned on someone elses feed, even if it's not directly linking to them? If I say 'Bill Gates' or 'Linus Torvalds' here, and they issue a GDPR take down(for what - SE have never 'collected' his information) does SE have to comply? Oct 19, 2019 at 21:04
-4

SE will never delete anyone's data - they'll just switch ownership to the community wiki.

This is bad because even if internally within SE it might be hard to connect post no. XYZ to a non-existing user, there may still be links to that post in other sites with anchor texts like "according to Username's answer on thid SE site...". I believe this is not GDPR compliant, but I am not a lawyer.

I.M.O. things like GDPR and other privacy-upholding codices are great, but people shouldn't depend solely on those things to protect themselves - most specially here, since SE is fighting against privacy by keeping your content.

If you wish to leave and want to anonymize yourself, you can at least delete your comments (those get deleted for good) and posts (they'll just be harder to find and read, requiring 10k+ rep to view). This may be hard for people who are not power users, or who have a lot of content though. There is also a limit on how much of the content you authored you may delete, so for people with a lot of content (which might be most of us), unfortunately it won't be possible to delete all.

5
  • 10
    No, you cannot delete your posts, nor should you. When you post here, you license your content to SE. This has always been the case and has never been hidden. You do not get to delete the content you have posted here. This is a good thing. Anyone not OK with that should not participate. This has nothing to do with any of the recent issues, it is as old as SE.
    – terdon
    Oct 19, 2019 at 18:54
  • @terdon I think you meant to day only that we shouldn't. I can see a delete link below my own answers here and mostly elsewhere, and in two sites where I have 20k+ rep I can even vote to delete other people's posts (under certain conditions), so I totally can. Oct 19, 2019 at 18:57
  • 4
    No, I mean you literally can't. More specifically, when a user deletes many of their posts, an automatic mod flag is raised by the system and mods (and high rep users) can always go and undelete it. I have done so many times myself. So yes, you can briefly delete all of your posts, but they will almost certainly be undeleted.
    – terdon
    Oct 19, 2019 at 18:58
  • 4
    Not to mention, you are hard limited to how many of your own posts over a certain age you can delete in a 24 hour period (I believe it's 10). Hitting that limit will block you fully from further self-deletions for the rest of the day.
    – animuson StaffMod
    Oct 19, 2019 at 19:36
  • 1
    And what do you mean by “your comments … get deleted for good”?  Comments, like posts, only get soft-deleted.  I don’t know whether it’s straightforward to undelete them, but they are stored in the database forever, and ♦moderators (and appropriately privileged staff) can see them. Oct 20, 2019 at 1:41

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .