The purpose of Heande is to understand complex environmental and health issues, find solutions to them, and offer guidance to societal decision-making related to these issues. It is especially designed to handle very complex issues such as climate change, outdoor air pollution, or global environmental taxation. To really understand these issues, a large group of experts and other people must work together to bring in all information necessary for such an effort. Heande is not only a website for reading about these issues. Heande is also a place for openly working together, collecting the information necessary, and synthesising it into a useful form. Indeed, Heande is a description of a huge model (well, not so huge yet, but hopefully in the future) that helps to quantitatively estimate important parts of these issues and analyse the relationships between different parts. The collaborative work is directed to developing this model.
In addition to this wiki site, Heande contains parts of the actual model as computer code that can be run on different platforms. The summaries of these results can then be brought back to the wiki pages for others to read and utilise. And third, Heande also has a database that contains the actual, detailed results from different parts of the model. These results can be utilised as inputs for other parts of the model.
The format of information that is brought into this system is very flexible. Basically, the information can be anything in any format that is relevant for the issue and is either a) scientific information about the issue or b) information about value judgements about the issue.
Heande is about collecting, manipulating, and synthesising scientific information and value judgements. This is done in practice by building standardised blocks for the model. (These blocks are also called formally structured objects.) Each block has the same basic structure. Each block describes a particular part of reality. The blocks are usually but not always described quantitatively as probability distributions. The blocks are connected to each other based on their causal relations. The causalities are described as conditional probabilities such as P(C|A,B) where A, B, and C are blocks (random variables) and A and B are causes of C. Together all these blocks form a causal network which is also called a Bayesian belief network. Value judgements can be connected to these blocks for describing, which outcomes are good or more preferable than others. When value judgements are added to a Bayesian belief network, we have an extended causal diagram. In brief, the work done on Heande is about developing extended causal diagrams by utilising open participation. The aim (as mentioned above) is to help societal decision-making about environmental and health issues.
The outputs of the procedure are formally structured objects that can directly be used as building blocks for models. These are called variables. In addition, some outputs are actual assessments for a specified decision need. These are called assessments, and they consist of a set of variables relevant for that particular decision. One variable can be used in several assessments.
Heande is based on several scientific theories. The idea of Heande is to borrow and apply the best and most efficient methods and techniques in combination. Some of the most important ones are briefly mentioned here. There are more extensive pages describing these theories elsewhere.
- Falsification: Science consists of statements (theories) that can be falsified. Science is an evolutionary process where poor theories are falsified. The current knowledge consists of those theories that have not (yet) been falsified
- Bayes' theorem: A posterior probability given new data can be calculated from a prior and the likelihood of the data
- Decision theory: Decision analysis is a rational method for making decisions.
- Quality of an estimate: The quality of a quantitative estimate (probability distribution) can be evaluated against a golden standard using informativeness and calibration
- Bayesian belief networks: They describe the reality by using conditional probabilities.
- Vines in Bayesian belief network: Probability distributions can have any form and they can still be solved analytically, if vines are used.
- Argumentation: Disputes can be solved by using formal argumentation that consists of attacks and defends of a specified statement
- PSSP: A system can effectively be described using two kinds of objects: processes and products that are produced by these processes. Each object has attributes purpose, structure, state, and performance.
- Wisdom of crowds: A group of people is likely to outperform an individual expert in many situations, if they can use individual knowledge, act independently and in a decentralized way, and their opinions are effectively aggregated.
- Mass collaboration: A large group of unorganised people are able to produce complex artefacts, if the product is information or culture, the work can be chopped into bite-size pieces, and the pieces can be effectively synthesised.