BAGS: Basser Automatic Grading Scheme

The Basser Automatic Grading Scheme (BAGS) developed at the University of Sydney in the mid 1960s represented a significant improvement in the state of CAA applied to programming exercises. For the first time, grading of assignments on a batch-processing system did not require special actions or knowledge on the part of the operator. The BAGS system was ``simply part of the standard operating system and its exercises [were] run as normal, batch-processed jobs.''[15]

In the paper describing the Bags system, J. B. Hext and J. W. Winings specifically enumerated their requirements for a CAA system:

1. It should handle exercises in ALGOL, in MINIGOL (a subset of ALGOL), and in the KDF 9 Assembly Code.
2. It should not place any additional burden on the operators.
3. It should record every attempt at an exercise, with sufficient data for calculating a mark.
4. It should provide summaries on request for specified classes and exercises over a given period.

[15] also gives concrete examples of the types of programming problems that were being graded in this context: well-specified mathematical operations with zero margin for error. The grading scheme used is well suited for this type of problem. However, in terms of modern Computer Science pedagogy, there is really no way to view these problems as anything but toy problems: easily solvable in less than 5 minutes and 20 lines of code in a modern programming language. While there was a relative flurry of papers published and work done on CAA approaches to programming assignments during the 1960s, the work is simply too far removed from the modern realities of computing to be anything other than a historical interest piece. To get a better notion of what CAA means, we must advance through the literature several decades.