Boxer: a reconstructible computational medium

Reviewer: John L. Bennett

Boxer, a research tool for educational studies, is under development in the Laboratory for Computer Science at MIT, and the School of Education at the University of California, Berkeley. Prototypes have been implemented on Symbolics and Texas Instruments' LISP machines. The project explores what a programming language intended for laypeople might be like. Just as writing on paper is a widely available skill for both the everyday notes of people and the highly skilled creation of expert writers, could a medium be provided that would enable programming by a wide variety of people__?__ The term “reconstructable” refers to using the computer screen itself as an expressive medium. Images on the screen originally produced by experts (e.g., in a dynamic book) might be modified by people who are not experts but who know how to program “a little.” The authors acknowledge a similarity to Logo [1]. Their intended innovation begins by addressing the display screen as a medium that encompasses written language, hierarchical structures, databases, and interactive graphical tools. The box structure notion of containment is related to object-oriented programming [2]. Displayed boxes have computational semantics, and users have access to organization beneath the image on the screen. The boxes package local data and procedures, and computations are organized by sending messages to boxed procedures. The notion of “porting” provides an escape from the hierarchical structure by allowing a view on a box at some other place in the system to enable cross-referencing and sharing. Boxer makes use of a spatial metaphor by representing computational objects in boxes as regions of the screen that contain text, graphics, or boxes-within-boxes representing hierarchical structure. The language makes use of naive realism by showing the computational world in its entirety on the screen. The user sees computational objects, not just interfaces to them. Anything that appears on the screen should be available for the user to “run” or modify. The ultimate success of approaches such as that of Boxer will be dependent on the way in which the basic box structure is elaborated. Careful reading will reveal the tension between simplicity and triviality of function. What will be an effective way to support naive debugging__?__ What styles of use should be taught to enable the less capable student to grow in effective use of a language such as Boxer__?__ Will programs scale up from the examples shown in the text__?__ Since the user can modify everything, what kind of recovery mechanism needs to be provided__?__ Problems of scale and details aside, this is a well-written, thought-provoking challenge to provide a programming language usable “for personal construction by people at all levels of competence.” The authors appeal for research directed to the day when “programming will be a part of the everyday lives of many people who do not have expert programming skills.”