Platform ban waves: a system change or a decision about you?

Enforcement behavior changes without any rule changing. Thresholds get adjusted, a category gets a sweep after an incident, a market gets a different setting, a detection system is retrained. From the outside, all of that looks like a decision about you.

Telling the two apart changes what you should do next, and it is one of the few things you can work out from your own records.

What to take away

What actually changes

A threshold moves. The written rule is untouched and the point at which the system acts shifts. This produces the largest number of surprised accounts and generates no announcement anywhere, because internally nobody thinks a rule changed.

A category gets a temporary sweep. After an incident or a news cycle, a category is policed more aggressively for a period. Enforcement then returns to roughly where it was, which is why some accounts see an action reversed weeks later with no explanation.

A detection system is retrained. New behavior, new errors, and frequently new false positives concentrated in a category nobody predicted.

A market gets a different setting. Two users doing identical things in different countries get different outcomes. Frequently this is a service implementing an obligation in one market, which is covered in the routes a rule takes to reach an operator.

The outcome attached to a rule changes. The prohibition is identical and the consequence is not: removal becomes reduced distribution, or a warning becomes an account restriction. Rarely announced as a change at all.

The published rule changes. The least common of the six in terms of how many accounts it affects, and the only one with a document behind it.

Distinguishing a wave from a decision about you

The left column is not a diagnosis and there is no way to confirm it from outside. What it changes is your expectation. A well-argued appeal is worth writing either way, but in a wave what most often resolves it is time and a bulk correction rather than your submission.

What to do in the first days

The steps are the same as for any action, and they are worth doing before you decide which situation you are in.

Watching for changes before they reach you

There is no reliable early warning, and three habits get you the closest available thing.

Keep your own record of what happened to your material and when. Over time this is the only evidence anyone outside the service has about the enforced rule as opposed to the written one, and the gap between the two is the subject of how a moderation system actually works.

Archive the policy page when you make a decision that relies on it. Pages are revised without a changelog, and the previous version often stops existing publicly.

If a decision concerns something you published in March, the rule in force in March is the operative one. Proving that requires a copy you took at the time. The reasoning behind that, and what the agreement lets a service do, is in the terms you accepted.

Watch the categories rather than the announcements. Changes to advertising, monetization and commerce rules arrive on separate pages from the general policy, and those are the ones most likely to affect anyone operating commercially.

Writing an appeal into a moving system

Short, specific, factual. Which item, which rule was cited, why it does not apply, and what supports that, with dates. An appeal is an input to a process with very little attention to give it. Length, indignation, and argument about the policy in general travel badly.

The reason a decision has to be explicable enough to answer at all is set out in a general account of notice, and the reason review by somebody uninvolved matters is set out in an explanation of appeal. Neither describes an obligation on a private service; both explain why the structure exists where it does.

One point is worth making explicitly in a wave: if your item was published under an earlier version of the rule and the decision cites the current one, say so, and attach the dated copy. That is a concrete point an internal reviewer can act on, and it is the main reason the archive is worth keeping.

Common questions

How long do these periods last?

There is no general answer, and anyone offering one is guessing. What is knowable is your own appeal window, which is published, and which is the thing not to miss while waiting.

Should I change what I publish while a category is being swept?

That is a business judgment rather than a compliance one. What is worth avoiding is deleting your own archive in a panic, since that removes the evidence of what you published and when.

A competitor is doing the same thing and nothing happened to them. Is that useful?

Not as an argument. Enforcement is automated and sampled, so the absence of action against someone else shows that they have not been actioned, and nothing more.

Is there anywhere to see whether a wave is happening?

Not authoritatively, but the EU comes closest. Article 24(5) of the Digital Services Act requires online platforms to send all their statements of reasons to the Commission's DSA Transparency Database, which is publicly accessible. Its search lets you look for statements of reasons from specific platforms or timeframes, which is the nearest thing to checking whether many accounts were actioned in the same window. Elsewhere, aggregate reports are published on a delay and in categories chosen by the service, and the caveats on reading them are set out in the discussion of transparency reporting in the moderation pillar.