- Recommendation Letters. Should only be read by during the recruiting process and never by the candidate except as required by law. If someone other than the candidate asks you for a recommendation you shouldn't even mention to the candidate that you were asked.
- Referee Reports. While the reports themselves are furnished to the author, the identity of the referees must be kept secret. Any discussion between the referees and authors should go through the editor.
- Committees. Any discussion in a program committee (or any other kind of committee) should remain closed except as agreed upon in the committee. This allows the committee members to speak freely. In a PC you shouldn't even mention whether a paper was submitted.
- NSF Panels. You should not disclose any discussion during an NSF panel, or even the fact that you were a panelist.
- Salaries. You can announce your own salary but you shouldn't mention other people's salaries. Exceptions for surveys and states that require that the public have access to the salaries of all of their employees including state university professors. Update: I've just been told I am not allowed to publicly announce my salary as a University of Chicago employee. Apparently the employer can decide the appropriate policy for salary disclosure in the US.
- Personal Information. Disabilities, Illnesses physical and mental, Gender, Sexual Orientation, Marriage, Children, Religion, Race and other related issues except as necessary or as already publicly known.
- Email and Personal Discussions. You shouldn't reveal research or other discussions with someone else without their permission.
If you truly want things you say to remain private then don't say it. With many theorists having loose lips and the minimal security of email you cannot count on the fact that what you say that should not be spread will remain unspread.