(Somewhat in reference to the other answer and pursuant comments...) Even though I do privately think that my job is to give the students the best education I can, it seems both "fair" and "anxiety-reducing" to make quasi-contractual promises in advance... all the while realizing that the students make no such promises to me, etc. But it's a very asymmetrical situation.
So basically what I try to promise is to assure the anxious, primarily about the grading system, gradelines, and how much material will be "assessed-upon". In effect, such an approach does give openings to those seeking to "game" the system, but I reject a role of "policeman" for one thing, and, for another, see it as undesirable that the earnest (often anxious) students get put-upon for the sake of trying (which is often unsuccessful anyway) to keep at bay the game-the-system people.
The aspect of the syllabus that specifies the extent of the content of the course can be made fairly independent of the grading scheme specification, I think, even if this amounts to declaring some "extra topics" outside the bounds of "what will be tested-on". (In upper-division undergrad courses I'd regularly include will-not-be-on-the-test information, with a few catty remarks about people who only care "what's on the final"... and certainly in graduate courses only talking about what's reasonably testable does a great dis-service to the students, since they'll have a "textbook" picture of something (with the pursuant testability constraints, non-interaction with other topics/ideas, and so on), rather than the "live" version.)
I really do think the "promises about grading system" and "outline of what will be covered (and tested upon?)" are two very distinct things. The grading system should be a bit contractual, I think, given the power discrepancy, and potential for harm/anxiety. The content promises are rather different, since there's not necessarily any adversarial relationship.