This is a request for comments! Would this help you improve your moderator workflow? Anything I am missing or getting wrong?
Currently, if a new user signs up and is flagged by Akismet as a spammer, the only options available to the moderator handling the flag are No and Delete User.
It’s not uncommon for us to hear from customers that they are uncomfortable with immediately deleting the user, or that they can’t do it for legal reasons. They’d rather silence the user to give themselves some time to investigate and also to give the user a chance to make a case for themselves that they are not spammers but deserving of membership in the community. Doing this is awkward, because it means answering No and then going to the admin user list to find and silence the user.
So the suggestion is to add a Silence user option here. There is already robust functionaltiy available around silencing users and making sure they are well informed about why they are silenced and for how long. The user silencing fields would be prefilled. Akismet would not yet be trained that this user is spam.
Later, moderators can unsilence users or delete the users. If they delete them, then they should have the option to train akismet at this point that they are spam. I am not sure if this is currently the case, even if the Delete and block this email and IP address is chosen. It seems to make sense that it would be.
I think the third option shouldn’t say “spammer” — after all, if they’re known to be a spammer, delete is right there. Something like “place on hold” instead?
Can silenced users message moderators? I’m thinking I’d like something like this;
Possible spammer detected
Case is unclear, so moderator says “place on hold”
User is silienced, but…
User gets DM saying that an automated system has placed the account on hold for verification, and that they can respond with more information if they like
If user does not respond after some time, auto-delete them.
If they do respond, moderators get “reply and keep”, “reply and delete”, “just delete and move on with your life” buttons.
I’m actually not totally sure what the impact of being silenced is. I do believe silenced users can PM moderators, and they can reply to the email they get when they are silenced if one is sent.
I also checked and see that silenced users still have user profiles. This feels wrong to me. If the user is deemed problematic for any reason, why should their user profile still be visible to the world?
The other question I’m interested in knowing the answer to is what happens when a user is deleted later by a moderator via their user admin. Is Akismet also informed if the Delete and block this email and IP address option is selected?
I think they get silenced and the reason is shown on their profile? (If so, that too is potentially problematic.)
But: silencing is often the first step in a code of conduct response policy with escalating consequences. The goal is to ask the site member to cool off a bit, not to vanish them.
It seems like maybe there needs to be a different level — hidden? (This would also hide all of that users’ posts, I think?)
We could add a “User Profile” reason for silencing, and in that case hide the profile until the user is unsilenced?
I am interested in more input from more sites on how they deal with user spam flagged by akismet, or how they deal with users who post inappropriate content on their user profiles.
Also, is it possible to flag a user using the flagging system, when viewing a user profile? I don’t think so. That might be helpful to let community members let moderators know when they come across a user profile page that they think is inappropriate or spammy.
Silence reasons are not public but suspension reasons are.
As for why they are public I dont think it was ever said here but when suspetions were first added to Stack Exchange when Jeff Atwood still worked there (he made the blog post iirc) saying its a durtrent for other users since its something they can see moderators take actions.
As for the site I mod most reports we get are not spam. If we arent sure we check their profile mainsite (its a forum for a compeny so we have SSO) and see if we have anything there. If we are still unsure we allow and action if needed later.
Wow! That’s very interesting. So silenced users are still very much part of the community… they are just forced to lurk. I’m surprised they can still like.
I do think it makes sense to be able to change their “about me” user profile, esp if the reason they were silenced is because their user profile looks spammy or has some other problem and a moderator has asked them to change it before they get unsilenced.
One thing I am recalling now is that we have the add rel nofollow to user content which I believe also covers user profiles. So you can put whatever spam likes you want in there but they will never be picked up by search engines.
I agree with this, especially now we have Reactions to take into consideration.
I also think replying to non-staff PMs should also be removed.
This would make Silence more effective as a halfway house where community interactions were not permitted, but unread lists were still tracked, bookmarks could still function, conversations with mods in PMs would still be possible, etc.
This is a great proposal. Our community is one where we silence spammers (even if 99.9% sure they are spammers) rather than delete.
I think it might be better to divorce the Akismet feedback from the action taken. On the Akismet notification, the options could be “Is this her profile spam: Yes/No/Not Sure”.
If the answer is No, then same as now: Akismet is trained that it got it wrong and no action is taken on the profile.
If the answer is Yes, the Akismet is trained that it got it right and the Mod is given options to Delete or Silence user.
If the answer is Not Sure, then Akismet is not trained either way and the Mod is given option to Delete, Silence or take no action on the profile.
I wonder how many sites silence spammers instead of deleting them. How common is this really? Would love input here from more sites on their work flow when dealing with spam signups.
I would keep these as configurable options vs forced changes in core function.
With PMs they can only respond to PMs they have received and cannot create new ones. Most users in my XP don’t realize this possible loop hole and generally can be quite harmless
I do think a silenced user should be prompted to message Staff if they want the silence reviewed/make attempt to resolve as not all sites change system user to attach to human mod.
That often comes down to what someone thinks is spamming. While much spam meets the Justice Potter Stewart standard (I know it when I see it), I’ve seen Akismet flag posts that were perfectly acceptable for no discernible reason. I’ve also been a part of communities where almost anything off-topic, especially if it has a ‘commercial’ feel to it, is considered spam even if Akismet lets it through.
And I’ve had more than a few users who got hacked and were being used as a spam conduit without their knowledge.
So in any given community, there may be mitigating circumstances surrounding labeling someone a spammer, especially if it involves taking actions that are not easily reversed or without any type of due process.
And sometimes what’s really needed is explaining to the ‘spammer’ what is and what is not acceptable. I’ve found most people will pay attention to a reasonable explanation of a community’s standards, though there are people who just want to be troublemakers or don’t believe rules should apply to them.
I mean something very specific: a spammer is someone whose sole purpose on the forum is to drive traffic to some commercial, scam, or malware site. They are not acting under a real identity that is meaningfully attached to a real human being — not in the sense of “real name policy”, but actually. The account is entirely scripted or run by some low-rent clickshop.
I get what you mean about edge cases, where someone is a real person but keeps posting links to the same site, or advertises services where that’s not allowed. To me, that’s a different problem from what I’m describing, but I recognize that some people and communities do label that behavior “spam” as well.
The default settings would have the first several posts from new users held for staff review. I don’t know if Akismet sees them before or after the staff review.
I like the idea of having Akismet (or other spam detectors) silencing spammers much more than having Akismet delete them, because IMHO deleting a bad actor account should be a staff function. There needs to be a way for a silenced user to contact staff if only for due process purposes.
Some of the bots are getting quite sophisticated and could easily fool someone into thinking they were a person rather than an automated script. That’s only going to get worse with AI. Will Discourse develop AI tools to combat AI bad actors?
Yet another possibility would be to create a separate status (like silenced) for ‘quarantined spammer’, which might have different options than ones used for silencing unruly users. This might be helpful in a large community in which there are more than a few silenced users at any point in time, because it could be searched for separately. But that would likely be a more involved project.
