59

With the recent Code of Conduct changes (see the blog post, the FAQ), and even before this round of changes, there has been an increased attention on users interacting respectfully with each other. However, moderator tools have not been kept up to give the diamond moderators insights into how users have been interacting, leaving us in a tough spot. We must either become investigators, digging through activity of two or more users, or in a position to ask a particular user to dig through their own history. The latter option could be extremely difficult for the user, depending on the nature of the interactions. The former can be extremely time-consuming for the volunteer moderator.

I propose a tool that accepts the input of two users on a given site and returns information about their interactions. This would be all questions in which they interacted (within the question, one or more answers, or comments), chat transcripts for times when they were in the same chat room parented to the site in question, and a list of sites where both users have an account so questions can be directed toward the moderators of those sites. Some elements of user history - avatar changes, user name changes, and about me changes (including when it is changed by importing from another site), should also be indicated for both users, with diffs if they are available. If there are ways to expose suspicious voting patterns or other data or abstractions over data that capture interactions, those would also be useful.

In order to respect privacy of our users, the use of this tool should be logged much like how moderator access to PII is logged (and there's a message stating as such). Staff should be able to see cases where a moderator looked up the interactions between two users to see which moderator performed the function and what two users were searched.

I believe that having such tools will greatly improve a moderator's ability to address concerns where one user is made to feel uncomfortable or unwelcome by another user. We can learn more about their interactions to appropriately find and address instances of poor behavior and make better informed decisions about how to protect our communities.

Improve this question
11
  • 4
    I’ve been relying on data.stackexchange.com/stackoverflow/query/853808/have-they-met, my own version of the “have we met” SEDE query floating around, with several issues fixed and improved. It’s far from ideal; content up to a week old, no deleted content, no chat. Still, helpful for socks and voting rings too :-)
    – Martijn Pieters
    Commented Oct 11, 2019 at 2:11
  • I think it need to be over all sites the two users are on, not just a single site.
    – Ian Ringrose
    Commented Oct 11, 2019 at 10:09
  • @IanRingrose There may be privacy concerns with that, but I'm not sure if it's limited to the same type of data that is in the Data Explorer. There may also be performance reasons that would make that unachievable. However, since some form of the data exists for profile pages, displaying a list of sites where both people have an account (and perhaps date of most recent activity) can allow a moderator to reach out to the moderators on those other sites to investigate as well.
    – Thomas Owens
    Commented Oct 11, 2019 at 11:41
  • 8
    I don't want to do this. I am a moderator on an online Q&A site, not an interpersonal skills coach. If I have to step in and sanction people, then something is going badly wrong and the evidence will be obvious. I don't want to conduct genealogical research, read the entire history of users' interactions, or anything of the sort. Just as importantly, I don't want to encourage other moderators to do this, either. These sites aren't about people; they're about content. If people are getting in the way of the content, then that's when a moderator needs to step in, and only then.
    – Cody Gray - on strike
    Commented Oct 12, 2019 at 2:44
  • 5
    @CodyGray None of us want to have to deal with this, but sometimes it's necessary to resolve a site disruption fairly. Most of the stuff we face is easy to figure out. Having a tool available to help with the stuff that we rarely face doesn't mean all the sudden we're going to make the easy stuff more complicated. If it's not a nail, put the hammer away. If it is a nail, wouldn't it be nice to have a hammer instead of a rock (especially when you need to pry it out instead of hammer it in)?
    – ColleenV
    Commented Oct 12, 2019 at 13:39
  • 1
    @CodyGray It's also nothing new. Handling potential sockpuppets and other claims of harassment are hard without tools to inspect interactions. Fortunately, on my site, it's very, very rare. But moderators are, and have been for a long time, expected to serve as community builders and leaders. Perhaps it's different on Stack Overflow due to size or scope, but this is something I feel is necessary to make mods on other sites much more efficient. I don't feel comfortable needing to kick so much stuff to the CM team and wait days or weeks for a resolution when it comes to unwelcoming behavior.
    – Thomas Owens
    Commented Oct 12, 2019 at 15:00
  • I wonder, Thomas, do you think it would be reasonable to expect the person raising repeated flags to collect the evidence and share it with the site moderators privately? In the unusual but important situations where the flagger feels there is ongoing tension and bad behavior going on. I was once such a flagger, and I was careful to collect the evidence to share with the moderators, because I am aware that moderators are not God (all-knowing, all-remembering). They're only human....
    – aparente001
    Commented Oct 13, 2019 at 13:54
  • 1
    @aparente001 No. In a case where a person has been treated disrespectfully over a period of time, we should not ask that person to go through their history and dig up instances. That can be incredibly painful for people. Tools such as the one I propose will allow a user to flag a single instance of behavior and indicate that it was persistent over time and give moderators the ability to investigate and confirm without asking the flagger to dredge through months (or more) of possibly hurtful interactions.
    – Thomas Owens
    Commented Oct 13, 2019 at 13:57
  • I see. Okay, then consider this as a temporary workaround suggestion (from the user's point of view): the way I've been able to collect the evidence I needed was by looking through my flag history. (Sometimes it's tedious, true.) I have found that the key is to include the verbatim quote from the comment I'm flagging.
    – aparente001
    Commented Oct 13, 2019 at 14:02
  • @aparente001 There are workarounds, for users and mods. But they are clumsy. Mods can see the flagging history as well, but it would require that the old posts were just flagged. Perhaps the two users in question simply tried to reach agreement and moved on without flagging but the issues repeat - there may be no flag history to use.
    – Thomas Owens
    Commented Oct 13, 2019 at 14:04
  • Ah. Well, this confirms my opinion, that new and intermediate users should be guided to do a whole lot more flagging and a whole lot less confronting and arguing. // I've found it quite helpful when a moderator responds constructively to a flag I raise: (a) If they decline, they explain why; (b) If they found it helpful, they leave a public comment, a constructive comment, using the incident as a teachable moment for the user base. // I hope you will let us know what you think of Martijn's query once you've had a chance to work with it a bit.
    – aparente001
    Commented Oct 13, 2019 at 14:09

3 Answers 3

Reset to default
11

Another situation has come up on ELL that might have been prevented if it had been easier for the mod team to understand how accounts were interacting with each other's posts. As I mentioned in a comment, this isn't a tool that would get used very often on most sites, but when it's needed, there's no reasonable substitute.

Our team has a need for this tool currently. We have a user that is abusing the kindness the community shows new users by making multiple accounts to ask proofreading questions. Normally, once the community saw the pattern, the questions would be downvoted and closed, but because these all look to them like different users asking single questions that might be salvageable, people let it slide.

We’ve been monitoring these 10 or so accounts for a while to try to determine if there is any voting fraud going on, and just stumbled across an example today. If we had access to a tool to identify interactions among accounts, we could have fixed it before it got to be this much of a mess. It also turns out this user has all those accounts on two other sites and is probably sock puppeting there as well.

(And before you ask, no, these aren’t roommates or coworkers who are innocently participating and don’t realize the problem.)

Improve this answer
2
  • Just curious. Do the sock puppets get identified by matching up IP addresses? I wonder if there could be a query for that. Or is there some other way of finding sock puppets?
    – aparente001
    Commented Oct 15, 2019 at 2:22
  • @aparente001 Regular users don't have access to the information needed to find sock puppets.
    – ColleenV
    Commented Oct 15, 2019 at 10:23
8

I don't think this is useful. In general, if I am going to take moderator action, I want the justification to be obvious. I am not going to do that much digging to find evidence of wrong doing. I suggest if you are concerned that you hand it off to a CM and let them handle it.

Improve this answer
3
  • 16
    I agree that I'm not going to do much digging. However, I'd rather have the tools to support cases where a user is indicating that they feel unwelcome. Consistent with the new CoC updates, for example, let's say that they claim they are being misgendered by a user. I'd rather have the tools to investigate this easily rather than just handing it off to a CM. There are less than a dozen CMs for 170+ sites. I'd rather be able to take action quickly than hand off to a CM.
    – Thomas Owens
    Commented Oct 10, 2019 at 17:01
  • 4
    I agree: The CMs thought this change up. Let them deal with the difficulties of it.
    – Cerbrus
    Commented Oct 12, 2019 at 10:04
  • 3
    Also, this tool is useful for figuring out if accounts are interacting when we've identified someone with multiple accounts. Some folks try to breathe life into their sock puppets by answering their own questions and leaving comments.
    – ColleenV
    Commented Oct 12, 2019 at 10:21
2

It might be possible to do this using https://data.stackexchange.com/stackoverflow/queries?order_by=favorite and look for questions where both users left comments, posted answers, made edits, etc. I wouldn't be surprised if there's a canned query for this.

This one might work: https://data.stackexchange.com/stackoverflow/query/50166/where-have-we-met-before

Improve this answer
3
  • 2
    Maybe? Building this into moderator tools would be useful, especially if it also looks over deleted posts and chat. I know that chat is definitely not part of the data explorer. However, this may provide some of the inspiration to build a moderator tool.
    – Thomas Owens
    Commented Oct 10, 2019 at 23:49
  • 1
    I long since fixed several of the issues with SEDE query you link to here: data.stackexchange.com/stackoverflow/query/853808/have-they-met
    – Martijn Pieters
    Commented Oct 11, 2019 at 2:09
  • 1
    I think it would need to somehow denote if/when the user changed their username and/or avatar.
    – StrongBad
    Commented Oct 11, 2019 at 18:39

You must log in to answer this question.

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