ALLEGEDLY* is built in the shape of a gossip column and runs on the opposite of gossip. That contradiction is deliberate, and it is worth explaining properly, because it determines what you will and will not ever see in this game.
A story qualifies for the puzzle bank only if it clears one of these bars:
A criminal conviction or guilty plea. Not an arrest, not a charge, not an accusation — an outcome. Arrests get dropped, charges get reduced, and accusations sometimes turn out to be nothing at all. A conviction or a plea is a fact a court has established.
A civil judgment or formal settlement. Where a matter was resolved through the courts and the resolution is documented.
An official ruling by a governing body. A sporting federation stripping a title, a league issuing a suspension, a regulator handing down a fine. These bodies publish their decisions and stand behind them.
Something that happened live, on camera, in front of an audience. If millions of people watched it happen in real time and the footage still exists, there is no dispute about whether it occurred.
That is the entire list.
Everything sourced to an anonymous tip. Everything sourced to another outlet's blind item. Everything about someone's private relationships, health, sexuality, or family that they have not themselves made public. Anything where the most honest summary would be "people say." Anything involving a minor. And anything where the person was accused and never found responsible — that last one rules out stories that would make excellent puzzles, and it rules them out anyway.
Being interesting is not a qualification. Plenty of the best material in celebrity culture is unverifiable, and unverifiable material is exactly what this game does not run.
Three reasons, in ascending order of how much they matter.
The practical one: a game whose entire appeal is "guess the famous person" is worthless if players cannot trust the answer. If a case might be wrong, then getting it wrong is not a fair loss, and the streak you are defending means nothing. The public-record rule is what makes the scoring honest.
The legal one: repeating an unverified claim about a real person is repeating it, and republication is not a defence. A small independent project has no business making claims it cannot stand behind.
The one that actually matters: real people are on the other end of these puzzles. They have families who can read. A conviction is something a court decided after hearing evidence, and describing it accurately is fair comment on a public act. A rumour is something someone said, and turning it into a guessing game is doing damage to a person for entertainment. The first is a trivia question. The second is not a thing worth building.
Accuracy is necessary but not sufficient. A case can be perfectly sourced and still be cruel — stories about addiction, mental health crises, bereavement, or a person's worst private day are documented often enough, and they are not entertainment.
The rough test a case has to pass is whether it is about something a person did rather than something that happened to them, and whether the documented facts are already the subject of general public conversation. A world champion biting an opponent on pay-per-view is a public act everyone watched. A person's medical history is not.
About a third of the bank is not scandal at all. Same rules, same standard of proof, opposite valence: documented acts of generosity, absurd competence, or quiet decency that never got the coverage the scandals did.
They are also, consistently, the hardest puzzles in the game. Given a redacted story about a famous person, almost everyone's first instinct is to guess someone with a bad reputation. When the answer turns out to be a person who did something genuinely good, the guess history reads as an accidental confession about how we all think. That is the most interesting thing the game does, and it only works because the good stories are held to the same evidentiary bar as the bad ones.
Errors get fixed at the source, in the puzzle bank, so the correction reaches the archive and any future replay of that case. If you have found one, tell us. On a site whose entire premise is "this is on the record," a factual error is not a typo — it is the one failure that undermines everything else.