The reasoning provided here is exactly why we often explicitly told students that they were not graded on a normal distribution. I was told that, historically,

The reasoning provided here is exactly why we often explicitly told students that they were not graded on a normal distribution. I was told that, historically, when students were graded on a normal distribution, it typically led to bimodal distributions of grades - a few students continued to try, but most were convinced that if they all did poorly then they would all still do fine, and so they just tried to be "good enough" and chance it. By the time I was teaching it was simply departmental policy not to grade this way. Even without this historical evidence, however, it does seem unfair to grade this way for exactly the reasons in BugsJustFindMe's response. It might make sense in larger first-year classes, but in the more advanced smaller classes (say, 30 students) it's likely that you're already down to a set of students who try really hard and potentially all deserve an A. There's a lot of correlation between (perceived) difficulty of the course material and the students who take the class, so there's no reason to expect that you have the same distribution of students from those beginning first-year courses in your "Advanced Stochastic Processes" course.