Bungee jumping off the ivory tower

Abstract

From the viewpoint of at least one corporate consumer of computer education, Lisp-based training confers many professional advantages, even if the student never goes on to use Lisp in real life. Among these advantages are a grasp of computer science fundamentals that comes from simply knowing the Lisp language (Whorf was right, in this domain at least) and good software engineering habits (incremental development, abstraction, modularity, object-orientation). These habits are widely believed to be advantages, but Lisp is not widely seen as helping students acquire them. This paper will corroborate the connection between Lisp-based training and these good habits. On the other hand, some of these habits may not be good at all (this is one reading of Gabriel's "worse is better" theory). This paper will argue that Lisp-based training is more flexible than training based on popular imperative languages at discouraging the bad in these habits. Then this paper will describe some true drawbacks of Lisp-based training. These drawbacks include stylistic jingoism, excessive delegation of performance issues to the Lisp implementor or to the machine, superstitious avoidance of the more complex (albeit more casually implemented) Lisp facilities, stubborn reluctance to distinguish between compilation, runtime, and delivery environments, and a tendency to make use of Lisp's extensibility to re-invent the wheel merely because the tread size is not quite right. Ironically, Lisp itself contains the cure for the diseases it spreads. This paper concludes by offering, from real-world experience, some suggestions for modification, not abandonment, of Lisp-based training.