The Garden with Insight garden simulator - Our pre-proposal to the National Science Foundation (spring 1997)

This page gives most of the text of a pre-proposal to the National Science Foundation (NSF) we (Kurtz-Fernhout Software) submitted in spring 1997 for a grant to improve the Garden with Insight garden simulator. The NSF's response was something like: "We don't understand why you don't want to sell the program commercially." We don't want to sell the program as proprietary software because 1) we would like the source code to be used and improved by others; and 2) we don't have the time and money to market the program. Our hope is that someday we may get to the version 2.0 modeling environment we speak of in this pre-proposal another way: through the help of many interested individuals.

Abstract

We have already invested six person-years in developing the Garden with Insight garden simulator. A public beta release is available at http://www.gardenwithinsight.com [or was when this pre-proposal was submitted; the version 1.0 product is available now]. This product for adults and children 12 and up can be used for both formal and informal science education. We would like to make this simulator available nationwide to promote scientific literacy by making it freely downloadable (with source) from our web site. We plan to enhance the product so users can develop their own models and modify existing models and collaborate with others by sharing models and data over the internet.

Tools for open-ended science exploration improve scientific literacy.
Many educational researchers have stressed the effectiveness of learning by doing, and particularly learning by designing. They call for quality science exploration tools that emphasize curiosity, creation, integration, and connection to personal experiences (e.g., Papert, Solloway, Reiber, Balestri, Ehrmann, Ferguson). Some excellent examples of science exploration tools are LOGO, The Incredible Machine, STELLA, the SimCity series, and Outpost. Many of these software programs are popular in the home market and reach millions of people.

Gardening is a popular activity.
Gardening is both widely popular and an effective introduction to science. It is the most popular recreational activity in the United States; and over half of the households in the US have a garden.

Gardening is an effective channel for scientific literacy.
Gardening is an excellent subject through which to improve public awareness about science, math and technology (SMT). Paying attention to where food comes from, learning how it is grown, and exploring the delicate balances that exist in the natural world all lead to an increased awareness of SMT and of the environmental consequences of everyday actions.

The study of gardening is integrative (it encompasses many areas of science including botany, earth science, chemistry, mathematics, environmental issues, and systems theory); it is experience-based; it is accessible (a simple house plant can make the connection); it is enjoyable; and it is life-long (most people who garden tend to do so for many years).

We have created a garden simulation.
We have developed a gardening simulation based on scientific models that combines the attraction of gardening with the exploration of an open-ended simulation. The public beta version of the Garden with Insight™ garden simulator is available at http://www.gardenwithinsight.com [or was when this pre-proposal was submitted; the version 1.0 product is available now].

Using a computer simulation to study science through gardening has unique advantages.
Studying gardening using a computer simulation is especially effective because of the space and time demands of real-life gardening. Users can make measurements and changes in a simulated garden that would be impractical, too expensive, or impossible in a real garden. Curiosity about the simulation can lead to interest in how scientific models are created through observation and hypothesis testing. Garden simulation is also a relatively open niche. Other programs exist that can be called garden simulations (SimFarm, Forever Growing Garden, and Lunar Greenhouse), but they are much simpler programs that do not emphasize scientific modeling.

Microworlds are great places to learn.
Microworlds have been recommended as learning environments especially suited to self-regulated learning through life-long play (Papert, 1980; Reiber, 1996). A microworld is a "little world," a learning environment characterized by: self-containment within a subject domain with a constrained universe of possible actions; relation of basic concepts to concrete realizations; intrinsic motivation; ease of use; interaction at several levels of complexity; an engaging environment rich in a variety of questions and hypotheses; and possibilities for open-ended design.

The Garden with Insight garden simulator is self-contained within the world of weather patterns, soil processes and plant growth. Basic concepts such as photosynthesis are related to concrete realizations such as green growing plants. The only goal is to grow beautiful plants and make bountiful harvests. The friendly, Kid-Pix-like interface makes the simulation attractive and enjoyable. Imaginative creation is encouraged -- users can create their own climates, soil types, soil amendments, plants, and garden backgrounds.

The Garden with Insight microworld has three levels (see picture on proposal cover [picture is of the program's windows]). The first level is a concrete representation of the garden in the familiar terms of garden tools, 3D plants, and soil surfaces. The tools look and act similar to their real-world counterparts. The second level of the simulation is shown in the browser. Using the browser, users can see the relation between numbers with units and pictorial representations of data. The third level of the simulation adds graphing measurements over time. Through graphing, users gain an understanding of trends over time and of relationships between variables. This most abstract level enables users to see the garden as an organized system of interacting parts.

Our target audience is young people and adults.

Our target audience is adults and children 12 and up:

• who are interested in gardening, especially those interested in the science of gardening -- who buy gardening books like The Soul of Soil (Gershuny and Smillie), and Botany for Gardeners (Capon), and who enjoy television programs such as The Victory Garden.
• who are interested in science in general -- who buy popular science books like Chaos (Gleick) and Powers of Ten (Morrison and Morrison), and who enjoy television programs such as Nova and Scientific American Frontiers.
• who use computers and like simulations -- who buy software such as SimAnt, The Incredible Machine, and Outpost.

st=st+rfv-qd

becomes

waterContent_mm := waterContent_mm + rainfallAndIrrigation_mm - totalRunoff_mm.

We want some traffic on the bridge.
The usual outcome of an effort such as ours is a commercially distributed software product with proprietary source code. However, research continues on modeling of soil processes and plant growth, and this product will soon fall out of step. A proprietary program may be a bridge, but no one can walk across it. By making the model source code available, we will bring scientists out from their side of the bridge to interact with the models, while at the other end of the bridge the general public will step out by changing the models themselves.

Open systems are the best places for interchange of ideas.
The careers of academic researchers depend on publication of their work. If we release the source code to this product, and an academic researcher adds a submodel for plant canopy shading or improves a water erosion equation, copyright and intellectual property issues will surface. Can the researcher publish our source code in an academic journal along with his or her changes? Can we use the researcher's changes in a future version of our product? The best way to ensure that communication between this product, the scientific community and the general public continues is to make the model source code freely available and to make it easy to integrate outside contributions to the models. We believe this product has the potential to become a vital link to which many could contribute and from which many could benefit. Other open systems such as Python and Linux are examples of this type of endeavor.

A common exchange framework for scientific models is needed.
Releasing the model source code freely is the first step in populating the bridge. The second step is to create a common ground for interaction. If users receive our current source code, they must purchase Borland Delphi, make changes to the source code, and recompile the entire project. There is a significant learning curve as well as financial cost to get to this point of interaction, and few will attempt it. Interchange of ideas in this situation is possible but difficult and frustrating.

We would like to facilitate communication by creating a version 2.0 product that includes a model development environment. In version 2.0, users will be able to transcend the existing levels of the simulation to make changes to the simulation models. Users will change interpreted model source code in an environment that incorporates debugging, version control, and flexible data storage. Users will be able to collaborate over the internet by exchanging source code and data in a self-contained system that requires no external tools.

Here is an example of what we envision for the program. Recently we saw a posting on a usenet newsgroup asking if anyone knew of a simulation program that modeled seed dispersal. If version 2.0 of Garden with Insight™ existed now, that person could research the scientific literature and add his own seed dispersal equations to the existing models. He could add those equations to the common bank of source code at the Garden with Insight™ web site, and others could download his code and examine it, review it, and incorporate it along with their own modifications. Discussions on seed dispersal would include experts in the field and interested non-professionals.

The product encourages curiosity and exploration.
The program draws users in to deeper understanding through curiosity about simulation results. For example, if two corn plants are grown in different soil types and one grows faster than the other, users will want to know why. Within the topic of tending and harvesting plants, users are free to ask many questions and create many scenarios. Many of the model processes can be enabled or disabled to ask specific questions. For example, one can turn off the percolation of water through a soil patch and see how plant growth in that soil patch changes.

The product enables self-paced learning.
The program can be used at many levels of understanding. For example, one user can plant two tomato plants, water only one of them, and see how well they grow by looking at size and color of the 3D plants. Another user can plant two tomato plants, water one of them, and watch how their water stress factors diverge over time in the browser as they demand more water. Another user can plant two tomato plants, water one of them, graph their biomass and water stress, see how the two are correlated, and come up with a hypothesis as to how water stress affects plant growth.

The product helps users understand scientific concepts, topics, and processes.
Many of the concepts used in gardening are important for understanding science. Units, for example, are prominent in the simulation; users must learn what it means when an amount of material in kilograms per square meter, for example, is converted to a concentration in grams per kilogram of soil. Graphing simulation variables and examining correlations between them is also stressed. Many other important concepts are explored.

The product helps users develop scientific thinking.
Since a simulation is a little world in which users can freely explore, it is a perfect environment in which to safely test the scientific method. Here is one of many examples. Why did these two pea plants grow differently? Let's look at the soil they are in. The first pea plant is in a sandy soil and the second pea plant is in a loam soil. We will hypothesize that the first pea plant is not getting enough water because the water is draining through the sandy soil very quickly after it rains. To test the hypothesis, we can water the pea plant in the sandy soil every day and water the pea plant in the loam soil once per week. If the two pea plants grow the same amount, we will have proved our hypothesis. But is this an adequate proof? What can we do to validate our finding? And so on.

The product involves girls in science, math and technology.
This topic is of particular interest to one of the principal investigators (Cynthia Kurtz). One of the reasons she got involved in the project was because of her frustration at finding so little software that interested women and girls. Both of us have found motivation in the effort to produce software that is non-violent and that comes out of the creative and beautiful aspects of gardening. The cooperative aspect of the product is also appealing; girls are often drawn to situations such as this where competitive pressure is absent and where they can create without extrinsic goals. In our testing so far, girls and boys have shown fairly equal interest in the program.

For example, one of the people who downloaded the beta version told us: "My 15 year old daughter learned to manipulate many of the tools in about 4 minutes and had plants growing in about ten minutes....By the end of 12 minutes she discovered how to turn the music on, did so and left it on with pleasure....She fumbled across the Browser, but was not really interested in any sort of explanation about what it could do, as of yet. She spent another 30 minutes in the garden harvesting, reseeding, starting new soil patches, composting, double digging and more or less just seeing what happened....When I asked her if this would be a program she would like to use again she said, 'Yes.'"

The product encourages parents in the support of their children's SMT endeavors.
Gardening is a cross-generational activity; grandparents, parents and children often enjoy gardening together. Exploring science through gardening is a wonderful way for parents and children to learn together, because 1) it emphasizes a positive activity most parents are happy to encourage; 2) many parents are interested in learning about gardening for themselves (it's more interesting than compound fractions); and 3) parents and children can carry out gardening experiments such as growing seedlings on a windowsill at home with very little cost or time.

The public beta version (version 0.9) of the Garden with Insight garden simulator was released on the internet (at http://www.gardenwithinsight.com) in January of 1997. (Since its release we have seen about 2-3 downloads per day with no advertising except listing in several search engines and making a few usenet posts.) Version 1.0 of the product will be available in July 1997 [it was released in July].

The modeling framework of version 2.0 will support the development of models ranging from very simple to complex. Users will choose a base model, then add their own equations and methods or modify existing ones. The model development environment will include debugging, version control, and flexible data storage. Sharing of model code and data will be supported so that collaborative efforts will be possible over the internet.

Steps in development of version 2.0 will be:

• Choice of a development environment. We are strongly considering VisualWorks Smalltalk, a multi-platform environment with which we have considerable experience. Other candidates are IBM VisualAge Smalltalk, Squeak (an open Smalltalk system), Borland Delphi, C++, Franz Common Lisp, GNU Common Lisp, and Java.
• Creation of the model code interpreter, possibly building on an existing interpreter system, which allows flexible data storage (possibly in an object-oriented database).
• Creation of the modeling environment framework, with code editors, browsers, debugger, and version control, and conversion of the current models for use in the modeling environment.
• Creation of simple and intermediate models leading up to the complex models already in the simulation, so that users can choose the level of complexity at which they want to work.
• Porting of the existing user interface to combine with the model development environment.
• Complete documentation of the modeling environment as well as of the base models.
• Improvements to the web site for communication between users, including a discussion system and a file upload/download area for exchanging model source code and data.
• Creation of a parent/teacher guide with experiments, plans and ideas.

Timeline for project accomplishment

• July 1997: Release of version 1.0 with software and model source code freely available.
• July 1998: Release of first intermediate version 1.5 of garden simulator with modeling environment system and model source code on web site.
• quarterly 1998-2000: New intermediate versions of garden simulator with modeling environment system released quarterly, with incremental progress towards final version.
• July 2000: Release of final version 2.0 with complete modeling environment.

Our plans for evaluation consist mainly in making the program widely available to the public and the scientific community as it evolves. Along with quarterly releases of new system versions we will make annual reports to the NSF on our progress. Reports will include statistics on web site visits and product downloads, and feedback from program users.

All versions of the Garden with Insight garden simulator developed during the period of the grant will be freely available over the internet and freely redistributable.

We are positioning this educational product primarily for the home market because schools are only slowly adapting open-ended learning tools. We would love to see this product used in both formal and informal education, but we think starting in the home is our best strategy. Making the product freely available to schools will help the product move into formal education more quickly.

Within formal education, this product could supplement existing curricula that use gardening to teach science. For example, the Fast Plants and GrowLab systems, which grow plants "from seed to seed" in about one month, could be adapted to work with this simulation. There are also possibilities to integrate the simulation with sensing technology, including using weather data from a local weather station and using data collected by students with sensor probes. Funding could be obtained in the future for these adaptations.

We have spoken to some educators about the possibility of holding a teachers' conference to create teaching materials for the software. At this conference teachers would learn about the product, use it, and work together to create curriculum materials. The materials created would be freely available to all teachers and home users. We have no definite plans in this area, but hope to pursue the idea when time and finances permit. Funding would be pursued by a joint grant application with educators.

Making the software and the model source code freely available as well as creating a modeling environment that will encourage exchange of ideas will give the project a life of its own. We may also apply for additional grants in the future, and other organizations may become involved.

The two principal investigators, Paul D. Fernhout and Cynthia F. Kurtz, are a husband-wife team. Both received their M.A. degrees from the Ecology program at SUNY Stony Brook.

Paul's B.A. in Psychology is from Princeton University, and he has done some Ph.D. graduate work there in Operations Research and Statistics. He has been programming for over seventeen years. He has worked in many computer languages including Smalltalk, C++, Pascal, Lisp, and Forth. His expertise lies mainly in object-oriented systems design, interpreter and language design, project management, and interface development. He has developed three nationally marketed software products: PetTeach CAI (a computer-assisted instruction product), a video game, and RAMAS/age Macintosh, an ecological modeling software package.

Cynthia's B.S. in Biology is from Clarion University of Pennsylvania where she also studied art and writing. She has been programming for eight years in Pascal, Visual Basic, C++, FoxBase, Paradox, and Smalltalk. She has nine years of academic training in biology and two years of post-secondary teaching experience. Her expertise lies mainly in her interdisciplinary skills that combine science, education, programming, visual design, and technical and creative writing.

Our commitment to the project is warranted by the fact that we have already put nearly six person-years of effort into it at great expense and financial risk. Our long-term goal over the next ten years is to develop a series of scientific simulations like the Garden with Insight garden simulator, all of which are open-ended and encourage exploration and creativity in science. Our mission is to improve understanding of nature, technology and society.