Ninety Percent of Moderation Is Boring and the Arguments Are All About the Other Ten

Public argument about content moderation is entirely about hard cases: the ambiguous political post, the satire that might be harassment, the line between rudeness and abuse. Those cases are real and they are perhaps one in a hundred of what a moderator actually does. Here is what the other ninety-nine are.

Spam, in volume. Duplicate posts from people who did not realise the first one submitted. Questions in the wrong place. Someone posting a phone number they will regret. Two people who have been arguing for four days and need to be told to stop, both of whom will accept it. A person having a bad night who has posted something they should be allowed to delete. None of this is contentious. Most of it takes under fifteen seconds and requires no judgement beyond noticing.

The reason this matters is that the discourse about moderation, which is entirely about the hard cases, produces bad intuitions about the job. People imagine a role that is primarily about adjudicating speech. It is primarily about maintenance, in the janitorial sense, and the skill it most requires is tolerance for repetition.

It also means the failures that actually damage a community are usually not the controversial decisions. They are queue length. When our backlog went above about eighteen hours, the effects were immediate and measurable: spam stayed visible long enough to be seen by everyone, arguments that would have been stopped early became threads, and members’ reports stopped coming in because reporting had visibly stopped working. Every serious deterioration in that place traced back to moderator capacity, not to any decision anyone made.

The second most damaging thing is inconsistency, and it is nearly always a capacity problem in disguise. The same post treated differently on two occasions does more harm to a community’s sense of fairness than a wrong-but-consistent rule. And the cause is almost never a change of mind; it is that a different tired person handled it at a different hour.

What I would tell anyone starting a community: your moderation problem will not be philosophical. It will be that there is more of it than you expected, that it arrives disproportionately at night, and that the people willing to do it will burn out in about eighteen months unless the load is shared. We lost four good moderators before I understood that, and in each case I had interpreted their departure as a loss of interest.

The one thing that helped most was writing down decisions on the boring cases, so that new moderators inherited a hundred small precedents rather than a philosophy. Not a policy document — a list of “we do X in this situation.” It made the work faster and the consistency much better, and it turned out that nearly all of the hard-case anxiety I had been carrying was actually the accumulated friction of re-deciding trivial things forty times a day.