There has been growing support for component-oriented architectures for educational software. For example, diSessa [4] has described and advocated ``Open Toolsets'' such as Boxer; these are flexible and malleable collections of components which can be combined to create ``microworlds". These microworlds offer students an opportunity to explore the interaction of elements they have constructed themselves. Similarly, the SimCalc project [14] is currently developing educational components using ``open'' architectures which can be extended, customized, and integrated. See also [9] for a description of component-oriented exploratory software for Mathematics.
There are technical, social and pedagogical motivations for investigating the potential of component-oriented architectures. Their primary benefit is that sets of specific- or general-purpose tools can be built rapidly and relatively cheaply; they then can be easily modified, extended and combined to yield more tools. Moreover, tools built by other development groups can be integrated and customized to meet the needs of a wide community of learners.
So, instead of relying on professional software developers, teachers and learners are given a toolbox with which they can create their own interactive, dynamic resources. As diSessa points out, this might encourage a wider range of groups, including teachers and students - as well as programmers - to become involved in the design of tools and activities. It is envisioned that a large and powerful toolset might eventually emerge as a consequence.
Since these tools and resources can be distributed over the World-Wide Web as Java applets and services, they are not platform dependent and do not require the purchase of expensive specialized software or hardware. This ensures that the learning opportunities afforded by these technologies are available to the widest range of learners as possible.
Students working with a component-based toolkit will have an authentic integrated learning environment. They will have access to a broad spectrum of computational, symbolic, and visual tools which they can combine to test relationships and discover patterns. Students will have the opportunity to construct knowledge in response to a problem at hand - one which they will investigate by visualizing, transforming and simulating the mathematical concepts involved. This mode of learning reflects the ``constructionist'' approach spearheaded by Papert [12] and underlying much of the LOGO based environments, including StarLogo [13].