On my new Reciprocity website , I've added a feature which is a potential security vulnerability. This market resolves to Yes if either of these occur:

• The feature is exploited successfully to lead to a privacy failure.

• I remove this feature or substantially limit it because I'm afraid of it being exploited, or because it's too painful for me to monitor it to prevent exploitation. (I'll probably do this only if I believe that the feature is more than 20% likely to lead to a privacy failure.)

By "privacy failure", I mean that any user learns a fact about the checking preferences of any other user that they shouldn't have learned according to the normal rules of the site.

If there's controversy over how to resolve the market, @RyanGreenblatt will adjudicate.

• Update 2025-06-30 (PST) (AI summary of creator comment): A privacy failure will be considered to have occurred if any user who was using the site sincerely is misled into making unintended selections that result in a match.

  • A user acting sincerely is one who is not trying to cause this market to resolve Yes.

  • This condition holds even if only a small number of users are affected.

• Update 2025-06-30 (PST) (AI summary of creator comment): If a user is tricked, the market will resolve Yes if the creator believes it would be reasonable for the affected user to be annoyed at him for the feature that led to the incident.