Discourse Trust Levels

Hello forum;

I thought I’d raise the question whether we should relax the requirements for the level 3 trust level on the Snapcraft forums. There’s a link here about trust levels that describe how they function.

https://blog.discourse.org/2018/06/understanding-discourse-trust-levels/

My main concern is that there’s a fair amount of activity in these forums that are (likely) bot accounts. These bot accounts tend to come in, respond to messages in an automated fashion for the purpose of then embedding a link to raise traffic to malware/phishing/etc, and while in a lot of cases the flag functionality works as intended to remove posts; it’s not always the case. Sometimes a post that was originally fine will be edited weeks after the fact in a post with hundreds of replies; never going noticed by most. In some extreme cases, I’ve seen posts that pretended to be store admins, parroting responses word for word from previous store-requests!

Discourse Trust Level 3 gives a fair amount of control for community moderation, in that any T3 user can instantly remove the posts of a T0 user, which would cover the most common attempts of fake-posts. However, it’s incredibly hard to actually achieve T3 status in our forums, as it’s possible to lose T3 and revert back to T2 if the forum considers you “inactive” (or lesser active, at the very least).

I’d propose that we lower the requirements for T3 trust. Comparing myself for example:

I’m here a lot!

Like… a lot.

image

Did I mention… a lot?

image

And yet, I’ve held T3 status for about 2 weeks over a period of 4 years, because there’s some default criteria that makes it fairly hard to stay in that bracket. Presumably, posting at least 10 times in the past 3 months, recieving 20 likes in the past 3 months, and giving 30 likes in the past 3 months.

Personally, I’d feel that for the traffic in these forums and considering the nature of what goes on, it’s actually quite hard for any non-employed individual to sustain tier 3 trust; primarily I’d assume it’s the likes requirement. E.G, I’ll only like (as in literally pushing the button) 1-2 posts a week, I’d never hit the threshold, and I imagine a fair amount of people would be similar.

So to cut it short, would it be possible to review lowering the requirements on L3 trust to a level that allows for more community moderation? I’d personally feel the like requirements should be removed entirely; putting 10 hours of effort into helping a user might only get you one like. A funny comment might get you one hundred, I don’t personally feel that this metric has any value in Snapcraft’s instance.

There might also be a case for lowering the amount of threads read, but I’m less sure of what to propose here. A lot of threads will be store-requests for example, and while people will skim them to see what’s going on in the community, there’s not much most people who aren’t store admins can actually participate with for those kinds of requests in the ideal case, where a request is made, 2 votes are granted, and a decision is made being the optimum situation. For those threads, most people wouldn’t have much value to contribute.

So, I’m not entirely sure where the limits should be, but I feel there’s certainly a case to be made for making them lower than they are now.

Please let me know your thoughts and considerations.

2 Likes

I’m happy to give you a like to boost your standing. But I agree, these metrics don’t really correspond with anything useful in allocating responsibility.

For the authority to remove posts I think a better metric would be the proportion of post flagged that resulted in removal by an admin: Getting 3/3 in 3 months unlocks the power.

PS I’ve no idea whether I’ve achieved T3 status (and never had a reason to be interested in it).

1 Like

Thanks for the heart Alan, could I buy them on subscription :)?

It surprises me that you’ve never hit T3, again because you’re literally in the badge above for consecutive visits; you contribute to a lot of the Canonical projects and although I’ve never asked, I’m assuming you’re a Canonical employee!

Ultimately, most the T3 stuff isn’t very useful. Being able to comment and like more is great, but I’ve never hit those limits anyway. Meanwhile, you have posts like this:

This is the only thing I want T3 for. The post was edited 4 weeks ago to include malicious links and hasn’t had enough activity with flags to actually get removed, even though the user is T0 and any T3+ user would have been capable of doing it.

(There’s probably something to be said about making Canonical-ers T4 which is permament unlike T3, but I’m focused on the bots in this case and would rather not propose a too radical shakeup because I don’t think it’s needed).

3/3 valid reports in 3 months sounds like a great suggestion, I’d be unsure if it can be implemented in DIscourse since it’s presumably more than just lowering the existing defaults, but it’s absolutely more focused than the current scenario; being actually atainable and heck relevant to the one bit of functionality I see it useful for.

1 Like

I flagged it as spam now…

2 Likes

I’m not an admin for these parts but a moderator. I agree we could use one or two more moderators who can ninja posts in these situations, but that’s above my status to implement.

@admins ^

1 Like

I just checked my own trust level 3 requirements and yours @James-Carroll

This is mine:

The only thing you don’t meet is the number of likes. Get liking more threads, my man! :smiley:

1 Like

I’ve already liked every thread in this post ;), so I can imagine the number has jumped up 2 just now :slight_smile:

It might be a personality quirk on my own here, but, amongst the Linux community in particular I think one that’s not too uncommon. I simply don’t vibe with the concept of “likes” in most platforms as they’re an avenue for data mining and advertisement more than a genuine attempt at social interaction. E.G back when I’d use Facebook years ago, I’d literally avoid liking anything at most opportunities. The exceptions would be close friends where I’d struggle to justify “Oh sorry I don’t want to like your post because it’ll feed into an AI that is profiling my behaviours and relationships” and the friendship is more valuable than the data; but, that’s what I personally apply to everywhere. 1200 games on Steam and not one review. Youtube subscriptions? I already know who I like to watch and will visit them when I feel like it.

Obviously in the context of these forums it’s different, but having re-read what I wrote a few months back, ultimately I’d still agree with myself on this:

Seeing the rest of the requirements plain in your screenshot; they don’t actually look unreasonable. But I don’t want to be liking posts saying “Snap sucks use $ALTERNATIVE” just to hit an arbitrary threshold (even if I understand their point and “agree” with it), and on the other hand, there’d be posts I’d like more than once if I could :).

2 Likes

Understood. On forums I see “like” as a replacement for “nice” or “me too” responses. Also good for showing support when someone responds to a question. But that’s just my interpretation.

1 Like

It’s an interesting one from a psychosocial point of view. Maybe we could borrow the Super-Like idea from Tinder, or split up the likes into moods like Facebook or IM’s started doing. As I say, personally I’d avoid liking a comment I agreed with if the impression were “it’s effecting me too” if it isn’t, but then maybe if there were thumbs up vs love hearts and more nuance to it, I’d be more likely to.

Ultimately this is sounding like it might be mostly a problem with myself though. Would you happen to know how many T3+ users we have? T4 users would also be interesting. I’d mentioned above it’d make sense for Canonical employees to be given T4 by default since it never reverts back, informally I think people are aware who the “regulars” are even if the system itself would disagree, but I’d wouldn’t be surprised to find reviewers for example not being given T4+ even though they hold a massive amount of trust in other aspects.

I’m about to pop off to enjoy the weekend, but since we’re tangeantially on the topic of AI and I’m trying to find some new hobbies in both IT and out, I’ve asked the computer to create an icon symbolising a superlike for us, so here you are.

Safe to say I need to practise my prompt engineering skills :wink:

1 Like