Consider moderating first posts from new users

Site Help & Feedback
,
1

On the Ansible Discourse instance, we enabled the feature that requires moderation to approve the first post of newly registered users and expanded the moderation team to account for the added workload to combat a spam wave (see Let’s talk about spam - Forum Guide & Feedback - Ansible). I think this could help prevent AI slop and new users coming in and writing abusive posts (see the recent Change Proposal discussion) in Fedora. Is this something the Fedora mod team has considered?

EDIT: And maybe we can exempt users who are already CLA + 1 from this if that’s technically possible.

2

I think it would more disturb then help. There are so many new users signing up everyday, it would also slow down the response time of serious meant requests.

About the example you mention, I think it is on the persons responsibility, which creates the change request, keeping an eye on excessive posters. Feel free to respond and link to the docs which explains the purpose of a change request.

To stopping a chatty person in a discussion, it would be best, just move the comments into a separate topic, probably best in the The Water Cooler section (off topic).

@decathorpe as a Fesco member has access to the Team Workflows > Moderator Coordination section and can ask for assistance when he needs help.

P.S.
To separate a topic a user needs at least to be on TL4 in Discourse.

3

While it is technically possible to require all first posts from new users to go through moderation, it is (or would be) a logistical nightmare for the moderators.

Even if the moderating staff was increased, you would be spending more time clicking Approve Post than actually working towards a healthier community.

Here are my recommendations:

  • Actively encourage the community to participate in monitoring and flagging posts that are either spam, AI, abusive or in any other way unhealthy for the community
  • Offer badges or other little treats as a reward for keeping things neat and tidy
  • If not already in place, consider enabling Discourse’s AI spam detection
  • Consider using the Watched Words feature (accessible to both admins and moderators) to help flag content that requires a more watchful eye

Even though I only joined recently I have only seen one very obvious spam post (which I instantly flagged), so the question is how problematic is it here?

Personally, I think staying on top of topics and keeping them focused and making sure users abide by the CoC is more important than a few spam or AI slop posts slipping through, especially if the community catches them quickly enough.

Anyway, just wanted to share some perspectives on this subject which is not an easy decision whichever angle you look at it from.

At the end of the day, it is about finding a good balance.

4

FWIW, by the time I got back from touching grass away from my desk, at least one post had already been flagged for moderation, so I thought I didn’t need to ask for even more attention on it. :slight_smile:

I think Maxwell was referring a different (not PURL) discussion thread though (since that one didn’t have any first-time posters)?

5

The situation in the PURL discussion actually did prompt me to file this ticket. Technically, it wasn’t the user’s first post, but their first post in the AI Desktop thread a week ago was also… a bit much. I was thinking that maybe some kind of proactive moderation/guidance about forum guidelines could help in cases like this. In past topics that actually are controversial, there have been multiple cases where completely new user accounts do come in and post… nonconstructive comments as well.

Do we have any metrics on how many new users are joining and posting per day? If there are enough moderators spread across timezones relative to the number of new users’ posts, this burden should hopefully be manageable. To that end, I did volunteer to be a moderator.

I don’t think this is a good model. For one thing, most Change Owners are not already moderators, and in the cases where they are, when someone is attacking you and violating rules, it really needs to be a third person telling that someone to stop.

Anyway, I am not super partial to this moderate-first-posts idea for Fedora, but it was very helpful for us on the Ansible Forum. I was just irritated by the user being disruptive/abusive and flooding the other thread, and was wondering if there wasn’t something more systemic we could do. I agree that it’s important to strike a balance and that moderation is difficult.

6

If you catch them at it quickly enough, use the option to silence them.

Choose a timeframe and reason.

7

I guess you misunderstood my proposal. You not have to debate in public whit such users. First step which everyone can do is, do not answer directly on arguments from such users.
Second, If the user not get the message that they disturb the change process, then let us know. There are several ways to do that:

  1. In the case of the mentioned topic, user said it selves that it was a provocation to accelerate the discussion. That is not what the COC says, we practice a civilized and respectful discussions here! In case of a change request, discussions already have been made before the release of the request and the objective of posting this change requests is, to make the different WG, SIG’s and other contributors aware of it, so that in the end FESCO can vote about it.
    It is not the objective to inform users which are not experienced in this topic, to debate with them. The user clearly stated:

    I’m not a packager, so I have limited knowledge on this subject.

    The implicitly accusatory language was a strategy I imposed. I’m not sure if it worked – it may very well have. Making a discussion somewhat more heated, will likely increase the “velocity” of the discussion, so to speak. If the discussion is heated, people are also more likely to seek answers to any concerns that they may have, and actually post them on the forum.

    If the new/inexperienced users need to discuss, we can not forbid that. However we do have the possibility to move such discussions into the water-cooler (tech-talk). The system is creating a link to this topic and users can follow if needed and the change request process[1] can continue as planned.

    1. How to let us know when someone is disturbing in a Change Request Discussion?
    • You simply can talk with others about it. While using Matrix or other tools, to confirm with others if they also get disturbed about the comments. A TL3/+ user can “whisper” in a topic and create a answer which is just visible for TL3/+ and ask to move the “intruder” into a separate topic[2].
    • An other possibility is to Flag the comment you dislike. This will hide the comment if more than one user is flagging it. If you flag you can choose the last option (other reasons) and mention that the commentator is disturbing the change process, and you would like that someone is separate the comments into a new topic.
    1. A farther possibility to avoid new users “intruding” is, having the discussions in first place in the discussion Section under the WG/SIG which plans the change. If they appear then in this discussions and disturb, we can proceed the same way as described above with moving or flagging.

Using FPLA +1 here will probably not work well. However we could restrict the Change Request and also the Discussion category to the TL1+ users. TL1 you get very quickly when participating here. And if we move the discussions from the mailing lists into the discussion section, this users get very fast TL1 and can also propose changes :slight_smile:

P.S.

I do not know this option. Must be for Moderators only?!

  1. Manual Change Process ↩︎

  2. Moving requests into a separate topic can be done by a TL4/+ ↩︎

8

Correct.

Admins and moderators have a few tools they can use with, let’s say, difficult users.

  • PM reminding a user of the forum guidelines/CoC
  • Official warnings via PM
  • Silencing, means a user can log in and view but not reply or start topics/posts
  • Suspend, means a user also cannot log in to the system or reply/start posts

In most cases, except pure spam or hateful posting, the flow would be as above with suspension being the final resort.

9

Thank you for clarifying. I see that others flagged the comments before me, but I will try to do so in the future.

To be clear, I am not trying to imply that new people should be excluded from Change discussions (and I don’t think you were implying that I was implying that either :slight_smile: ). It’s okay to ask questions. It’s not okay to flood discussions with multiple posts in quick succession and ask questions in bad faith and use accusatory language on purpose and then argue with the moderators. I realized that the user was new to Fedora packaging and didn’t understand what was being purposed (it makes sense that a non-packager wouldn’t know what RPM Provides or Generators are, which were the subject of this Change), and I would’ve been happy to chime in and answer questions if they hadn’t been so hostile.

10

Of course, my comment was before we give the Moderation & TL4 more work, I would propose to take some rights away from new users. As I wrote, even if a user which not used discourse till now, they will be able within a week to get TL1 if they daily read here and give some comments ore make some questions.

:grin:

11

I’m not a moderator but I would absolutely hate having to vet new users first posts.

Other things to consider would be if we have enough coverage for whatever the most popular languages are here. Don’t have the metrics on that so can’t say how popular other languages are bar English.

12

Ask.Fedora is very active, day and night. This would overburden moderation (processing flags would be no longer a case of 3 minutes), or end up in a denial of service for users. A bad thing especially for those with more immediate need for help.

That said, the mix of our plugins and our very active user base, filter most spam and such long before it becomes a problem. I have to say the system flagging real users as spam is more of a burden at the moment than actual AI or spam cases (at least for those with urgent issues), but at the moment we can guarantee that every day at least one mod, usually more than 1 mod (and usually more than once a day:), is here and clicks through the flags.

“AI attacks” are seldom, but occur. While most are tackled the same way as spam, there are some that ain’t. But those that ain’t be tackled that way have a realistic chance to pass a “first post approval” anyway, as they have become hard to identify, and sometimes we need many posts going along with even a discussion to be get the details straight to be sure if it is AI. A moderator would need to invest a lot of time to exclude these cases from new users’ posts, if possible with certainty at all.

I have to say, at least in the current context, I prefer 10 undiscovered AI accounts that make users talk rather than blocking 1 real user who needs help or who wants to contribute (us blocking others can be also construed in problematic ways → we have seen a lot…).

Imho, in our specific Discourse context, the negative impact of such a function would exceed the negative impact of the cases we had.

However, I could imagine in other Discourse instances things behave different. My experience is limited in our Discourse instances to be honest, but I think most have not such an active user base as ours: many tl4, even more tl3, and a battalion of tl2 :classic_smiley: many of them are active in flagging, and many are established in the community and have learned how things are and what triggers some ai or spam cases have. Average users have become an integral part actually (here I think ask.fedora makes a difference between our and other discourses).

The interaction of the trust level user base flagging and mods being limited to process the flags in such cases works quite well (of course mods also flag if they see something:).

Sorry that I have not the time to read through the whole thread, but wanted to add my perspective to elaborate my “why” and ensure at least one mod gives some feedback :classic_smiley:

Such things are hard in our case. The “occurring” FAS and Discourse accounts are actually quite separate (despite a common login), which leads to some special behaviors in moderation. Elaborating the behavior between FAS and Discourse would exceed this topic, but the short answer is: we can use the login of FAS and transfer badges (which actually is transferring the badges out of FAS into the Discourse account by a script triggered by a magical topic → not “one account with its badges” from the systems’ PoW), but usually that’s it, at least at the moment :wink:

Given your interest in how moderation works here, another interesting hint: when we block a spam user, we have to delete their Discourse account, and then disable their FAS account → two actions. Not optimal, but not the highest priority at the moment, as we have established processes that work well to mitigate that. Our new guidelines that hopefully are to be published soon will elaborate these processes in detail, reviewable by everybody

13

I think we need to have a separate conversion for specifically Change discussions.

They should be moderated differently than other forum categories, so rather than setting up global rules, we need to look into the tooling which is available per category. And we can be more creative but also more restrictive with the rules here.

If someone would feel unnecessary limited in theChange discussion they would still be able to appeal in other areas of the forum, and we will be able to tweak the rules further.

Thus couple of questions:

  1. Can we enable the slow mode for the category by default? So that it applies to all changes without us waiting till someone abuses the discussion?

We can even have a Slow Category, with Change discussions and Council discussion being subcategories of it, so that it is more obvious for newcomers from the start that these areas of the forum have a different set of expectations, and are not meant for casual chat.

  1. Can we consider something like “initial feedback, then full feedback” rounds for a Change? Basically publish the change openly, but for the first day or two restrict it to TL1+, and then after two days open it further for all? This is essentially the same slow mode, just even slower.
14

Interesting points, but I think this should be put in a ticket in the repo, as I expect we will need to find out with the site admins, and then it’s good to have it documented for the future too. The tickets in the repo aim to involve those who can find out, those who are involved with daily moderation and “in contact” with the questions and ideas that come up in discourse, and consolidate and knowledge over time, so that we do not need to find out the same things regularly again (which so far is a real problem ^^) :classic_smiley:

→ https://forge.fedoraproject.org/discussion/tickets/

That said, my expectation would be that the answer to the first will be yes, but I have to say I don’t see how it solves the issue you want to solve. We have been always waited with slow mode until it was necessary. At the moment, I do not fully understand the exact problem they could solve? (Maybe I didn’t fully understand your point)

Also, keep in mind that some change discussions might belong to the tickets at all (not fully sure I understand which you have in mind in specific). Some types are in mod area, as the noise compared to the value of feedback of users who are here for a day or two and add their ideas in masses have in the past also shown to be a problem → the tickets in forge are also intended to mitigate that problem by being somewhere else (away from the one-day-users) but still public.

Concerning the second, I do not understand: can you describe the technical measure you have in mind in more detail? And the exact goal it shall achieve? (Sorry :slight_smile: )

15

Another interesting experiment we could try is to have a Matrix chat associated with the change linked in the Change announcement.

The chat would be a buffer zone for quick questions or comments like “oh noooo”, or “oh yeees”, and some random thoughts. But with the expectation that FESCo won’t really read it. A peanut gallery of sorts. :slight_smile:

And then the main thread will be slowed down and reserved for more properly outlined thoughts and suggestions.

16

Ah, right, the famous “what we are actually trying to achieve” question..

From my perspective, the set problems we need to solve are:

  1. We do not have a default place for a casual chat about changes, without strict rules applied.

  2. We do not make it clear for newcomers what are the requirements for good discussion of a Change. What is a good comment, what is a bad comment? How do we teach people to do it right?

  3. We need to set the expectation that the forum discussion here is different from a regular chat in other places to which people are used to. That saying something in the thread is not just sharing whatever thought you are sharing, like dumping a thing at social networks, but rather every comment on a Change thread has an associated cost to the project.

Slow mode is partially a solution here, as it makes people value the opportunity to say something. You would not waste it on random off-topic comments, you would try to bring your thoughts in order before posting.

  1. We want to prevent the waves of love or hate comments happening when people see the announcement of the change but have not spend time to actually read and understand the change beyond the title.
17

I lack time today unfortunately, so I cannot elaborate the whole problems, except mention that some have proven impacftul imho, but I would avoid to add further redundancy, especially as long as we have many audiences for whom it is not yet easy to understand where to post what, and blurs where to add and search for what type of documentation and discussion. Also, the moderation team is generally reluctant to engage in Matrix. E.g., a dedicated matrix channel for us was rejected, among others, to avoid more overhead. There are several more reasons for that (maybe someone from mod has time to elaborate). Sorry for the half-response :frowning:

Actually, the site feedback category is the casual place. So far, I find the rules ok. I think we never needed to use slow mode in this category, did we? (not 100% sure)

I think we should wait for the (self-)moderation guidelines to be published. They might clarify a lot and avoid some repetition of discussions → they aim to add a lot of transparency! We work on getting this done asap. I yesterday found some time to at least get the raw data online → https://forge.fedoraproject.org/discussion/self-moderation-guidelines-and-rules/src/branch/main/modules/ROOT/pages/index.adoc (this is not yet formatted! but the content is the approved and finalized guidelines that are to come)

I think indeed we need to more clear help and document to differentiate what is done in site feedback, moderation area and forge repo. That is where I would start. For most users, the question would be only when to use the casual site feedback, and when to file tickets. For people involved in FESCo, Council, etc. , some more information when the mod area is involved might be useful too.

When approving the new guidelines, the other discussion was still ongoing too. So they do not yet consider details about that. But that might be worth to be added to the guidelines → when who shall approach what (mod area, site feedback, forge repo).

We need to work towards achieving consistency and avoid redundancy, but also get stuff documented and available to those who need it at the moment when they need it. I think we are on a good way, but making transparent when to use which of the three channels remains a key issue to tackle soon. I would love to have a discussion at FLOCK about this! To exchange some thoughts in a more efficient manner.

18

Yes, there is a way to do this. Create a new category and then under category Settings >> Moderation Enable "Slow Mode" for new topics in this category

Two things to be aware of: Slow Mode is not retroactive, so it would be best to create a new category with the specific setting.

Secondly, Slow Mode only works to help keep posts more focused after someone has had their say. In other words, let’s say I post a comment, with Slow Mode engaged I would then have to wait x amount of time before commenting again.

However, and here is where moderators need to keep a watchful eye, Slow Mode does not prevent hateful, abusive one-off fire-and-forget posts.

19

Sorry, I think I haven’t made the context clear.

When I said Change, I specifically meant the “Fedora Change” as that official thing which FESCo votes upon. Not the change as in change on the forum.

My point is that the “Fedora Change Process™” can and probably should be done via much more strict rules then generic forum conversation. And all the points I made I meant to apply to specifically only the type of discussion as in Change Proposals - Fedora Discussion

20

A general reminder from my side: as much of what has been put forward has been put forward many times in the past and also discussed many times, I will not further respond to cases that involve site administration. This costs too much time and repeats too often in too many channels. This costs over time too much of the time of mods (our contribution time is limited too :wink: ).

Stuff that needs site administration (changes of categories and defaults etc), shall be put into a ticket at the forge repo, to be documented once and forever :classic_smiley: I hope everybody understands :classic_smiley: Of course you can discuss it here, but don’t expect much response from mods or so.