If we would guide by the light of reason, we must let our minds be bold.—Louis Brandeis

The primary goal of forum moderation is to clear away barriers to healthy discussion. In particular, moderation addresses harmful human behaviors that every community sees from time to time. Beyond enforcing rules, moderators cultivate a society from people gathered together in an online space.

That's the high-minded way of saying it. In practice, moderation is largely about noticing which people are causing harm to the community and doing something about it. Different platforms have different tools for moderators to do their work. I'm most familiar with Stack Overflow moderation and I'm learning how it works on Vanilla Forums. But today I'd like to share my research about Discourse moderation tools. There is an official moderation guide, which is helpful and describes tools I don't touch on here. This is more of a quick survey of what Discourse offers that other forums generally don't.

Trust Levels

The Trust system operates on a simple, but often-ignored principle: joining a group takes time and effort. Large and mature organizations observe rules, customs and rituals that make them distinct from other groups. It's natural to feel a little uncomfortable at the start of a new relationship. All good things demand some effort to obtain and that's certainly true of groups.1

Unfortunately, few communities do a good job of initiating new users to community standards. Quite a few conflicts occur when veterans interact with people who haven't taken the time to educate themselves on the customs of the site. It's like walking into a stranger's house and not noticing that everyone else has taken off their shoes. So Discourse assumes everyone new to a site will need to take time learning the group before they can be trusted with various actions.

Site administrators can adjust what new users can or cannot do, but the general idea is the damages they can do to a site is limited. Spammers can't post dozens of links. Trolls can't send private messages. Unlike other forum platforms, Discourse doesn't let users escape these restrictions by posting a lot. Instead, new users earn basic trust by reading the site. This doesn't mean they've learned all the rules, but it does mean they've made the effort.

Trust levels provide other, less obvious, benefits to a community. The user interface can be simplified for new users, for instance. And the investment required to establish an account means moderator actions have more bite. Each time a user gains a level, the system congratulates them on their accomplishment. Perhaps most importantly, however, trust levels allow the system to delegate straightforward moderation and content curation tasks to the most active members of a community.

Editing content

Traditionally only moderators can edit other people's posts in forums.2 Discourse grants that ability to more people in various ways so that the moderators aren't overburdened. I'm describing the default system, but almost all of the details can be adjusted if desired by the administrators of the site.

Lengthening the grace period

Members who have earned the second level of trust can edit their own posts for up to 30 days. That can help them avoid the occasional embarrassment of a typo or correct a misunderstanding. All edits are tracked by the system so that moderators can take action when someone make an inappropriate edit.

Wiki posts

At Trust Level 3 and above, regular users can create wiki posts that can be edited by people who are at the basic trust level. This can be useful for creating crowd-sourced documents that stay up to date. Again, these edits are recorded, so that anyone can see that they happened and what changed.

Fixing titles and categorization

Also at the third level, users also can rename threads or move them to other categories, which is useful since newer users tend to make mistakes in that department. Using more descriptive titles and putting threads in the right category goes a long way to making content easier to find.

Acknowledging leaders

One of the more unique concepts Discourse offers is Leader status. When staff recognize particularly helpful members of the community, they can grant a designation just short of moderator. These individuals have authority to edit other people's posts at any time and make other important changes to content on the site. Leaders don't have all the responsibility of moderation, which frees them to focus on improving the site.

Flags

Moderators can't possibly read every post on a site, which is why members of the community can flag posts. One of the great features of Discourse flags is they can be effective without even requiring a moderator to act. Depending on the circumstance, community flags can hide posts, close topics and even silence spammers.

More complicated flags must be handled by moderators. In addition to giving them an opportunity to intervene, flags create a record of behavior for the flagged user. The system itself prevents users with too many flags from gaining Trust Level 3. When a person continually causes problems on the site, there comes a time when a moderator should step in and take action.

Private messages and official warnings

Before taking more drastic actions, moderators usually ought to contact a user directly.3 Often resolving disagreements with words prevents a small problem from becoming a bigger problem. There's a cycle of dissatisfaction with a community that can usually be short-circuited by explaining the problem in a nonconfrontational and calm manner. When people feel they have been fairly treated, even if they don't get the outcome they are looking for, they tend to respect the process.

Moderators have the option to turn a message into an official warning. The difference is that moderators can see a count of official warnings when they look at a user's profile. It goes on their "permanent record". Importantly, however, only the user being warned and other moderators can see the warning. The feature accomplishes two goals:

  1. Moderators have a historical record which helps them decide next steps if problems arise in the future.
  2. People aren't publicly shamed when they make a mistake or misunderstand a situation.

Silencing and suspending

If a user continues in behavior that harms the community and quality if conversation after being warned4, moderators have the option to silence or suspend for a fixed period of time. The immediate goal of both actions is removing the user for a period of time. Silenced users can't post, reply, flag or start private messages. They can reply to messages and they can like and bookmark posts. When moderators silence a user, they can give a specific reason which only the recipient and other moderators can see. Silencing someone stops them from doing anything meaningful on the site.

For more serious situations, it's helpful to suspend a user. Suspended users can't log into the site and the reason for the suspension is displayed on their user page. Obviously this signals to the community that the user ran afoul of the communities standards of behavior. Once the suspension is over, their public profile5 is returned to normal. This encourages rehabilitation.

So when do you silence and when do you suspend? Simply put, suspend if there's a pattern of negative behavior and silence if it seems like a temporary situation. History of bullying other users? Suspension. Uncharacteristically got caught up in a fiery argument? Silence. Realistically, suspensions will be more common than moderators6 silencing accounts because most people tend to behave according to their character most of the time.

As a rule, the time period for these restrictions should start small (overnight, for instance) and increase if the behavior continues: days, weeks, months and up to a year. After that, moderators can suspend a user forever, but a year usually more than suffices. I've found that users can return to a site after a year-long suspension having matured into a productive member of the community. I don't like to give up on people.

Summary

Discourse provides integrated systems that work together to help active communities to moderate themselves. At the core is the Trust Level system which creates an on-boarding framework and formalizes a hierarchy that often arises naturally on forums. Moderation tools allow for immediate removal of negative behavior and a path toward rehabilitation. And all of these systems can be configured to fit with existing moderation styles.

Footnotes:

  1. I moved across the country from Virginia to attend college at UCLA. I arrived early because I played tuba in the band and we had a couple of weeks of bandcamp before instruction started. On my first weekend, I went to the student union to get some dinner. After paying for my food, I looked around for a place to sit. A guy sitting at a table called me over. We started eating together, struck up a conversation and he asked if I had a church. I didn't. I couldn't believe my luck and we arraigned to go together that Sunday. Only after I discovered he'd invited me to a high-pressure recruiting cult did I figure out why this had all been so easy. ↩

  2. Often people can't even edit their own posts after a short grace period, which is intended to prevent abuse. ↩

  3. As always, there are exceptions. Obvious spammers won't respond to message no matter how well composed. ↩

  4. Again, moderators might skip the warning stage if necessary to stop particularly destructive behavior or if they estimate someone has no respect for the community. ↩

  5. Moderators can see a record of suspensions, however. ↩

  6. The system silences suspected spammers, so the absolute rate could be the other way around. ↩