Exactly, it’s clearly an edge case.
Once upon a time I made a game called Forumwarz that was, quite literally, a troll simulator. This game itself had very active forums and moderating them was quite a challenge. My co-founder there even did a talk at GDC about dealing with trolls, and it got a lot of coverage.
Trolls thrive on engagement. There’s a reason why Don’t feed the troll is practically universal knowledge for those who frequent online discussions.
On Forumwarz, we deleted posts liberally and ruthlessly. There was no automatic notification about why their post was removed, we just took it out of the stream to let other people discuss things. Sometimes we’d attach a temporary ban with a reason, but that was up to the moderator’s discretion.
If we’d had an automatic notification, it would engage the troll. You might as well pat them on the head and feed them a treat! They’d run back into the topic and cry censorship, or bring up why their post got removed or all sorts of other manipulative techniques to derail or create a whirlwind of drama that focused on them.
So, sorry guys, but I agree with Jeff here. It’s the moderator’s discretion if they want to engage someone once they delete their content. Our software does nothing to prevent follow up conversations and in fact makes sending a PM related to a post quite easy. It’s a couple of extra clicks. (As an aside, it’s funny how often heated arguments about software boil down to 2-3 extra clicks.)
I will offer this though: Many of you WTF members are programmers; how about a plugin that creates the behaviour you guys want? If you can encourage many people to install and use said plugin, I’d certainly be more willing to change the default behaviour.
Most of us do not try to cultivate / manage a population of trolls. That might even be an edgier case than Jeff’s.
Not as funny as people not understanding the power of defaults.
Trolls shouldn’t be considered the default scenario in “a Civilized Place for Public Discussion”.
Shouldn’t we assume good intentions, even when those intentions lead to something that requires a deleted item, and build software for those mainstream scenarios?
No, to be clear the average user is not a troll.
But generally when you delete a post that’s the reason. Someone is doing something they shouldn’t, posting something they shouldn’t, etc.
Maybe I’m mistaken - can you give me some examples of mainstream scenarios where posts are deleted when not by a troll?
Newbie with good intentions didn’t read the rules before posting.
He’s not out there to cause grief but he posted something he shouldn’t have.
We don’t have many trolls in our community because we’re all trying to play nice together to build stuff to help people, but we do have people that make mistakes (the following were taken from 10 years of mailing list experiences because we’re new to Discourse):
- Posting personal information inadvertently or without understanding cultural notions of privacy.
- Someone that doesn’t understand how a system works and sends a personal/direct message to someone that wasn’t intended for public viewing.
- Account is hacked for spam purposes (far less likely for Discourse I’d assume).
- Someone blows up with personal attacks.
All of the above scenarios are real and pretty much the only scenarios in we’ve removed posts from our mail archives over 10 years. In all those scenarios, we want to help educate.
It is interesting that you haven’t had to delete much other stuff, although part of me wonders if some of the other controls Discourse has over a mailing list (such as new user education, rate limits, etc) change the characteristics of bad posts.
I’ve personally not deleted any posts on meta here that fit your 1-3 points, and #4 is what I’d consider trolling (or at least bad behavior).
It seems to me that this all comes down to a difference of experience and preferred workflow. Out of all the outstanding issues with Discourse saving some moderators 2-3 clicks is not high on my list. But I was serious about what I said about a plugin - if we could see with evidence (number of pluigin instances) that people drastically prefer the message prompt I’d be willing to merge it into master.
I’ve had at least a couple of posts deleted here that I did not believe were trolls (and perhaps one that was a semi-troll).
I find it fascinating that the mindset here is as follows; whenever an admin is forced to delete an inappropriate post that the OP must be an irredeemable troll, therefore we shouldn’t even bother acknowledging them or the post itself.
Why bother deleting the post in this situation? The troll will only make more. You’ve not done anything to address their trolling. They will just continue to troll, and you will just continue to clean up after them (until you ban them).
In this hypothetical situation, you are dealing with a bonafide troll who doesn’t contribute anything valuable to your forum. Just nuke them from the system completely and have done with it. Deleting posts is a waste of time in this particular edge case.
Allow me to pre-empt the counterpoint here:
But what if it is someone who has made some valuable posts? We only want to delete this one post, not everything they’ve ever posted.
This is your 99% common use case! If it’s someone who has made a valuable contribution to the forum, then they absolutely deserve some feedback, not a silent delete. This is not your pathological edge case of “feeding the troll”, trolls don’t make valuable contributions.
I’m not sure it’s really about saving clicks, except that making things harder makes those things less likely to happen (see @Dan_G and the power of defaults).
Again, I personally see silent deletions as the opposite of promoting civilized discourse. Perhaps they’re appropriate for legitimate trolls, and if that’s what you’re mostly dealing with, then maybe that makes sense as your default. But I don’t think that’s the common case (WAG), and if it is, new users / suspensions / deactivation becomes a more useful tool.
We are just repeating ourselves at this point.
I’ve offered up a fair path to get this integrated into core: make a plugin, and show us that it’s better. I’ll gladly offer help to anyone who gets stuck / has questions implementing this.
Otherwise I think this topic should be closed, it’s just cyclical at this point.
What would show you that it’s better?
The core team is pretty damned stubborn about what’s better.
I already said this, but if lots of people install it that would go a long way to convince us. Much more than posting in this topic.
Also, I think some of the previous points I referred to were deleted, so wanted to get it into the record.
Your opinion (and the others who contributed) have been duly noted.
As you can imagine we have a list about a mile long of features/tweaks/bug fixes we want to implement and it’s much harder to get onto that list than the plugin approach, especially if there is any amount of disagreement on our side as to its value.