In this presentation, we reexamine the concept of general purpose computing. Our argument is that general purpose approaches in workstations or web browsers have already failed in that they cannot deliver their intended functionality within the threshold of complexity of their intended users.

We argue that to provide access to the intended functionality while reducing complexity to the user is through the adoption of an approach to design that is based on task/location/user specific designs.

The claim made is that the form of the device itself, in this approach, becomes the interface. By limiting the functionality of the tool we gain in two respects. First, like a camera, the device can have embedded in it a knowledge of the basic task morphology. Thereby, much complexity can be off-loaded from the user to the device. (Compare photography today to that of two decades ago.)

Second, the specificity of the device means that the intended task can be represented in an appropriate form, or notation. The claim here is that notation is a tool of thought, and that one of the most important attributes of this "ecological" approach is that task/location/user specific design will reduce the complexity of task performance. (Compare the "cost" of doing long division using Arabic versus Roman numerals, for example.)

Just as it has been argued that diminutive applications, or "applettes" result in a reduction in complexity, we claim the same benefit for diminutive appliances, or "appliancettes."

Finally, while purpose-designed "appliancettes" may reduce the complexity of performing a particular task, are we not just transferring the cognitive load from the individual task to the coordination of the larger set of tasks? With traditional technologies, we would agree that that was the case, and that the designer is left with a choice between weak general versus strong specific tools. However, if the new class of information appliancettes are networked, and can communicate regarding state and context, then the claim is made that we may, for the first time, have the capacity to design - as a suite - a set of strong general tools.

To summarize, the enemy is not generality, but rather generality at the expense of complexity and strength - the natural consequence of the current approach to design. Our objective is to argue that there is a way to overcome this enemy and achieve the true potential of this technology. Finally.