Elastic Windows: Evaluations of Multi-Window Operations

Eser Kandogen and Ben Shneiderman

This paper outlines a new metaphor for presenting information in hierarchically-organized windows. The basic idea is that lower-level windows are contained in higher-level windows, grouped according to topic, and that all lower-level windows together exactly fill their parent window by stretching and shrinking proportionally.

The other major idea in this paper is the idea of role management. This is not the main thrust of the paper, but it is (in my opinion) one of the reasons this approach works. Role management is the philosophy that a person's work is divided up by the roles he/she plays. E.g., a professor may have work associated with being a teacher: classwork, grades files, and so on; and also work associated with being a researcher; etc.

Implementation and Experiments

Elastic windows was implemented (in C, I think, running under SunOS / OpenWindows). The authors then ran some nice experiments that we should definitely look at the design of.

The experiments consisted of training on the Elastic windows system, and the more standard twm window manager, followed by experimentation. The subjects were given three tasks of increasing complexity. As the results show, the users were significantly faster at both setting up their work enviroment and completing each task when using the Elastic Windows system.

Users were also much faster when switching task environments. This is pretty clear from the design; EW allows switching tasks simply by navigating to the appropriate parent window for each task, and as most windows are usually visible, this consists of acquiring the window visually, and simply selecting it, in comparison to all the window movement and focusing needed in twm.

My thoughts: I really like this idea. It wouldn't be impossible to implement this in Java, though integration with the operating system would be nearly impossible. I am interested to try it out; however, the authors have not elected to share the system they implemented.

I think what makes this work is twofold: (a) people don't mind thinking hierarchically, which allows them to organize effectively if they think that way (messy people, however, will still be messy; there is no "lamp" room); also (b) the tasks selected, while somewhat representative of work done by users, still play to the strengths of the EW system. This might serve for the incredible advantages shown in efficacity of the EW system over twm.