Hrothgar IE Overview

This document provides a short overview/rationale for the somewhat different approach to K-12 outreach underlying the Hrothgar Inquiry Environment effort. A much more thorough discussion is contained in the White Paper.

Hrothgar Inquiry Environment Home
Hrothgar Home

Assessing Business As Usual

The original Hrothgar Ising Model Unit is a fairly typical application of High Performance Computing (HPC) to K-12 education. It takes a computationally intensive real physics calculation and presents it in an "interactive" WWW-based form. The calculators and associated support pages do a reasonable job at exploring a number of interesting issues:

A physical model of ferromagnetism.
The role of computers and computer simulations in science.
Basic concepts and practices of applied statistics.
Modeling and "Thinking Like a Physicist".
The role of Quantum Mechanics in macroscopic phenomena.

However, attempts to incorporate this unit into physics classes at a number of local high schools have had limited success. There appear to be be two basic reasons for these failures:

The high-end science routinely done on HPC machines is not a good match to high school curricula, and versions of the code which have been simplified for a high school audience typically no longer require HPC. (In the Hrothgar case, interesting Ising Model calculations could easily be done on a single PC.)
High school juniors and seniors (and their teachers) are far too focused on college admissions and advanced placement exams, leaving little time for "innovative excursions".

The differences between research science at a level which necessitates HPC and the kind of "experiential" science needed in elementary and secondary education are non-trivial. "Big Science" is unlikely to be the best application of HPC for K-12 education.

Revising The K-12 Beowulf Strategy

Revised Premise: The relevant aspect of HPC for K-12 is not simplified versions of research science, but rather big simulations of simpler, K-12 concepts.

Having reached this conclusion, the challenge is now to give some definition to the term "Big K-12 Computing". A number of meetings between computational scientists and educators were held at CACR throughout summer, 1999, leading to the conclusion that some sort of large-scale, highly interactive simulation has significant educational potential. Specifically:

Students would directly control a number of "entities".
These entities would interact among themselves and with an appropriately lage set of background entities simulated on an HPC machine.
Entity interactions would be developed specifically to support inquiry-based learning within some broader curriculum element.

The HPC component lies in the scope of the underlying simulation, with the computers providing a large, heterogeneous environment of interacting entities. Most importantly, the simulated environment is simply a playing field. The only real outcomes and results in the simulations are consequences of student-initiated actions.

The actual subject matters for prototype simulations are still under discussion, but it seems certain that these will be drawn from the social or environmental sciences rather than the "hard sciences".

Connections With Contemporary Cognitive Science Research

Many HPC K-12 applications (including the Hrothgar 1998 work) are structured as excursions from the mainstream curriculum. This provides the simplest technology insertion strategy, but also tends to doom the efforts to the status of a nice but largely irrelevant sideshow.

The promise of a large-scale student-directed simulation lies in the potential for inquiry-based learning. To succeed, this inquiry must be coupled closely to a more comprehensive approach to learning/teaching - a significant change in educational practices, as advocated in the revent National Science Education Standards

The Diologic Inquiry approach to learning and knowledge, as advocated in various forms by Wells, Bereiter, and Scardamalia at the Ontario Institute for Studies in Education (University of Toronto) provides an excellent framework for the Hrothgar IE investigations. "Knowing" is achieved in a four step process:

Experience: An individual's social history defines the context within which new stimuli are to be encountered and processed.
Information: This is, in general, an "intrepretation of others" - an expression of meaning as presented from some external, often authoritative agent. It can come in a number of genres, including speech, written text, physical artifacts, and works of art.
Knowledge Building: In order to assimilate externally provided information, the learner must construct, use, and progressively improve various representational artifacts. Ideally, this produces a consistent, coherent `internalization' which is, however, individual and personalized.
Understanding: With time (and recurring use), the internal representations constructed during knowledge building become `second nature', and part of the learner's enhanced experience base. This transformation of `knowledge' into `understanding' is almost holistic.

The four-step cycle then repeats.

There would appear to be ample opportunities for incorporation of the Hrothgar IE simulation model within this picture. In particular, an appropriately flexible simulation would fit naturally into the knowledge building phase. This could enable construction and exploration of truly innovative knowledge models for disciplines in which experimentation would otherwise be difficult or constrained (e.g., studies of AIDS transmission mechanisms).

The Next Steps

The Hrothgar IE White Paper outlines a fairly broad strategy for implementation of a prototype "Inquiry Environment" for K-12 education. The HPC aspects are arguably straightforward, with a number of existing codes providing a strong working examples of the overall simulation framework. Three candidates are illustrated below, with the complexity of the basic "simulation entity" increasing as one moves from left to right.

The open issues in the educational realm are a bit more fundamental, going beyond the immediate concern of selecting a simulation scenario. An overall classroom unit incorporating the simulation engine should involve three key features:

Dialogic co-construction of knowledge.
A significant underlying activity in which knowing is embedded.
The construction and use of `artifacts' that mediate knowing.

A schematic for the basic building block of such a system is shown below.

In addition to an HPC (Beowulf) simulation engine and WWW-based user access, the system must contain a significant "user discourse" area, as indicated by the set of linked folders in the upper right portion of the diagram. The Knowledge Building methodology (and software) from OISE provides a strong initial candidate for this essential component of the overall system.

Extending The Community Of Collaborators

This project has depended on collaborations and interactions from the outset, and its continuing health will depend on extending these collaborations.

The participants in the initial summer 1999 discussions included:

From the research/administrative world: Janet Fisher-Hoult, Linda Polin, Billie Jean Knight, Nora Sabelli, John Cherniavsky, Lorna Miller, Hall Daily, and Jim Bower.
From the San Marino Unified School District: Andrea Mitchel, Jennifer Clark, Jerry and Kristen Koiles.
From the Pico Rivera Unified School District: Felicity Swerdlow, Melissa Manning, Steve Francis, and Yamileth Flores Orihuela.
From the broader community: Fin Cotton, Nancy Hunt, Wynn Wolfe, Chuck Champlin, Molly Feathers, and Mike Boughton.
The summer students on the Hrothgar Project were Will Cathcart, Tyler Hunt, and Nick Sun.

A number of contacts within the broader HPC and Cognitive Science communities have been initiated, including:

Boston College: Roscoe Giles and Raquell Holmes.
Concord Consortium: Paul Horowitz.
University of Toronto (OISE): Gordon Wells, Carl Bereiter, and Marlene Scardamalia.

Entending these contacts will be a major near-term effort of the Hrothgar IE project.