Reports, penalties and how they lift
The goal is a community that needs almost no moderating. So every penalty has a way back that the system checks on its own, and only two things still need a person: a clean picture with heavy reports, and a legal takedown.
Reports
Every picture has report this image with a reason. Per person: at most five a day, and none at all below a reputation of 0.50, so a suspected ring can't file. A report weighs what the reporter's vote weighs.
A picture is held, back behind the gate rather than deleted, when the open reports against it add up past a weight of 3.0. That's one trusted paying member, or six anonymous handles. With the scanner on, a reported picture is also re-scanned first. Moderators see the heaviest reports at the top of their queue.
What happens to authors
| Situation | Effect |
|---|---|
| A picture of yours is removed by a moderator | Counts against you; the People page shows it |
| Three removals in 30 days | Paused: you can still vote, but not submit or comment |
| Six removals in 90 days | Suspended for 30 days by the system |
| Reputation below 0.50 | Can't report, can't submit |
| A comment of yours is heavily downvoted | It collapses; readers can still open it |
A picture retired as "community rejected", meaning it sat at −5 or worse from five or more people for 48 hours, is not a removal against you. Losing a vote is normal. Having a moderator take something down is the thing that counts.
How every penalty lifts
| Penalty | How it lifts |
|---|---|
| Trust below 0.50 (can't report or submit) | Two votes that later match a human outcome, +0.25 each |
| Paused after 3 removals in 30 days | The oldest removal ages past 30 days, or 3 votes cast after the third removal later match an outcome, or a moderator clears it. Your account page shows the date and your 0-of-3 progress |
| Ring cap at 0.25 | Only mutual upvotes in the last 60 days count. Stop, and the flag is gone at the next scoring run, within six hours |
| A picture held by a downvote pile-up | Still at −5 or worse from 5+ people after 48 hours: retired by the system as community rejected, and the downvoters are paid |
| An open report, scanner on | Scan unsafe: removed now. Scan clean and the open reports weigh under 3.0 after 24 hours: cleared. Clean but heavy waits for a person |
| Suspension | 30 days, lifted by the system when served. A suspension a person imposed is never lifted automatically |
What still needs a person
- A picture that scanned clean but carries heavy reports.
- A legal takedown.
- Anything a person chose to decide, and any decision a person wants to overturn.
Every automated action goes into the same audit trail as a human one, with no actor and a printable reason, so a moderator can find it and reverse it. No AI sits in this loop. Being right is the evidence, the rule is the judge, and the reason can be explained to anyone who asks.
Safety is not a vote
A vote can rank a picture. It can never admit one. Every picture passes a safety scan before anyone sees it, and a picture that fails the scan is out regardless of how many people liked it. That rule is inherited from the app and it does not bend.