On the subject of grading: I used Perusall last semester for my survey of classical Japanese literature. It’s a 200-level double Gen Ed course (Humanities and International Perspectives). My algorithm settings were the same as this semester. The thing that hurt students’ grades the most was not that Perusall was a “tough grader,” but that students often simply didn’t do the assignment. Because I do not drop the lowest X number of assignments, each skipped assignment counts in the average. I think we instructors tend to think of our grading in terms of the A through E range, i.e., 100%-60%. But the Perusall grade, even with a point spread of 4-7-10 (which is what I use) is 100%-0%. If a student doesn't do the assignment, he/she scores 0%, not 60%.