First Quarter Changes

Initially Agar, like many of the CAA systems discussed in 3, tried to enforce only binary grading decisions in the automated testing process. It was hoped initially that Agar's automated testing framework could identify those things that didn't require any human intervention (those that met the spec perfectly), and initially give no credit for any test that did not have a positive outcome. During the first quarter of deployment, this was quickly identified as one of the worst assumptions. Simple things like docking a number of points per day late could not be easily expressed with such a system, and so support for non-binary grading was introduced. In light of constructivism in Computer Science pedagogy, this seems quite obvious. It is far better to support methods that can encourage additional intuitive leaps than it is to immediately punish any understanding that is not wholly correct.