(Of course they didn't know it at the time!)
The students in my introductory genetics pilot class have had weekly Peerwise assignments all term. In alternate weeks they either had to either create at least one multiple-choice genetics problem suitable for an open-book exam (i.e. not based on memorization) or to answer and critique at least two problems previously created by other students.
I had used Peerwise a couple of years ago in a first-year class. Each student only had to create one question and critique four. I had found their questions to be full of confusions, poorly written and lacking important information, but the critiques were quite good.
This time, my students have told me that they don't mind answering the problems other students have posed but find creating their own to be quite difficult. The first problems I had looked at hadn't been very good. But last week I started going through their more recent problems, looking for ideas I might use in creating problems for the upcoming midterm, and I was pleasantly surprised by the high quality of many of the problems they had developed.
I was so impressed by the students' questions that I decided to use them for the midterm, rather than writing my own. I downloaded a range of questions to consider - some of them were very good but much too difficult for a 45-minute midterm. I had thought that the good questions might have all been written by one or a few students, but when I checked their usernames I found that 16 of the 17 questions had been written by different students; this means that most and maybe all students are creating good problems!
I used 14 questions for the exam - most of them I was able to leave unchanged, but some required minor editing for clarity. The students must have recognized their own questions, but I haven't seen anything on the course discussion board to suggest that they've realized that they collectively created all the questions on the exam. I'll tell them on Monday.