"Facebook jail" is a joke people make because the alternative is being angry about it. You tried to comment on a friend's post and got told you're restricted for three days. For what? A notice points at a post from last month, or at nothing. You click Disagree. Nothing happens.
Here's how the system works, as far as it's publicly known, and what we built instead.
The strike system
Facebook's enforcement runs on strikes. Content that violates Community Standards is removed, and the account gets a strike. Strikes accumulate, and each one past the first triggers a restriction of escalating length: hours, then days, then up to a month, with the restriction covering some or all of posting, commenting, reacting, messaging, advertising and going live.
Facebook publishes this structure. What it doesn't publish is how most strikes are decided, which is automatically, by classifiers that read your words without your context.
Why normal people end up in jail
The classifiers are built to catch the worst things at enormous scale, and they catch a lot of ordinary speech along the way.
- Sarcasm and quotes. Repeating something to criticize it reads the same as saying it.
- Friends talking to each other. Insults between people who've been friends for twenty years look like harassment.
- Words with two meanings. Entire hobbies have learned to misspell their own vocabulary to avoid filters.
- Mass reporting. Enough reports in a short time can trigger removal before any review.
- Old posts. Rules change; a post from years ago gets re-scanned and strikes you today.
The restriction arrives with a notice naming a rule category, rarely the specific text, and an appeal button that feeds back into the same system.
What to do
Click Disagree every time, even if nothing comes of it; reversed strikes stop counting. Don't try to work around a restriction with a second account, which is itself a violation and can cost you both. Screenshot the notice and the post it points to. Then wait it out. There's no reliable way to speed it up.
And back up your photos and posts with Download Your Information, because the escalation ladder ends with a disabled account.
The rules POINT uses
We wrote POINT's moderation around one principle: the person on the other end is a member, not a case number. In practice:
The rules are short and in English. No impersonation, harassment, hate, threats, sexual content involving minors, non-consensual intimate imagery, illegal goods, spam, scraping. They're in the Community Guidelines, and they're the whole list.
Every report is read by a person. A report doesn't trigger anything automatically. It puts the content in a queue with a snapshot and the reporter's note, and someone reads it.
The response fits the problem. Dismiss it, warn the member, remove the content, or suspend for a stated number of days. Removing a post and suspending an account are different decisions, made deliberately.
You're told what and why. A warning or suspension sends an email that names the content, the rule, and how long, with an address that reaches a person. Not a category. The actual thing.
Nothing is hidden. There's no ranking, so there's no reduced distribution, no silent limits, no jail that nobody tells you about. You're in good standing or you've received an email.
Your stuff is yours either way. Download everything from Settings any time. A suspension doesn't lock your data away from you.
The honest trade-off
Human review doesn't scale the way classifiers do, and we're a small network, so we can afford to read every report. If POINT grows, keeping that promise will get harder, and we'll say so publicly if it ever changes. But the starting point matters: we'd rather be slow and fair than instant and wrong.
If you're on day two of a three-day restriction for a comment you don't remember, you don't need another lecture. You need somewhere your friends can hear you. That's what the invite link is for.