Yorum is for discussion that leaves everyone a little less wrong. We'd rather have fewer, better conversations than many loud ones.
What we ask
- Argue with ideas, not people. Blunt disagreement is welcome; insults, sneering and personal attacks are not.
- Show your reasoning. Say why you believe something and what would change your mind.
- Hard topics are fine; harmful content isn't. You can discuss violence, sexuality, drugs, suicide or extremism. You can't post content that itself harasses, threatens, sexually gratifies, spreads hatred, encourages self-harm, or gives step-by-step help with serious harm.
- No explicit or 18+ material, ever. Non-graphic discussion of sexuality is allowed behind a content note.
- Respect privacy. Don't post private details about people.
- No spam or self-promotion without real substance.
How every contribution is reviewed
Nothing is published until it has been checked. Here is the whole pipeline, so you know exactly what happens to what you write:
- Instant checks. A fast filter looks for a small set of patterns (including common tricks like spaced-out or look-alike letters) and for spam signals such as link floods.
- An AI moderator reads it in context. A language model assesses the post or comment against these guidelines, distinguishing discussing a topic from doing harm. It also files posts under topics and picks frontpage candidates. Only the text you wrote is sent to it, never your account details.
- Clear rules decide. The model's assessment is turned into an outcome by fixed, published rules: publish, publish behind a content note, hold for a human, or reject. When in doubt, the rules hold content for a person rather than guess.
- People have the last word. Held content goes to human moderators. Anyone can flag something they think was missed; flagged content is re-reviewed with a stricter bar, and enough flags hide it until a moderator decides.
Automatic translation
Yorum supports English and Turkish language preferences. When a post or comment is created, a language model asynchronously translates and persists it in the other language. Content written in a different language is served automatically in your preferred language with a clear "Automatically translated from …" label, and you can switch to the original text at any time.
Content notes
Some allowed posts deal with distressing themes, such as suicide, abuse or violence. These are collapsed behind a short content note so you can choose whether to read them.
Trust
New accounts can post a little less each day, and their contributions wait for a verdict if the AI moderator is ever unavailable. Limits relax as your contributions are accepted. Rejected contributions count as strikes; after three, an account's contributions go to human review until a moderator lifts the restriction.
If something you wrote was held or rejected
You'll see why on the post itself. Mistakes happen in both directions; moderators can see every automated decision and overturn it, and overturned rejections don't count against you.
Deleting what you wrote
You can delete your own posts and comments at any time. If others have already replied, the replies stay so their contributions aren't lost: your text is replaced by "deleted by author" and the discussion is kept, faded and closed to new comments. Moderators can also remove content; the page then says it was removed by a moderator and why. Removal is not a strike, and moderators can undo it.
If you're struggling
If you're having thoughts of harming yourself, please reach out to someone. You can find free, confidential support in your country at findahelpline.com.