From Opasnet
Jump to: navigation, search

Class → set?

How to read discussions

Fact discussion: .
Opening statement: A collection of variables, based on some similarity of content of the variables, should be rather called 'set' instead of 'class'.

Closing statement: a 'set' is a group of objects (items) that belong to the set. This can be an arbitrary collection of objects. In contrast, a 'class' is a collection of objects that can be unambiguously defined by a property that all its members share. Thus, class is the word that is used for shared properties in the pyrkilo method.

(A closing statement, when resolved, should be updated to the main page.)


←--1: . In some ontological discussions, which are (or at least should be) also influencing the development of Intarese theory and method, the word class has a specific meaning, other than presented on this page. In order to avoid confusion, I suggest the word class in this particular meaning presented on this page to be changed to set. In the four-category ontology by Lowe (Lowe E.J. Recent Advances in Metaphysics, Facta Philosophica 5 (2003) 3-24), the word class has the meaning of kind of universal object. This four-category ontology, and thus also the meaning of the word class, is included in the PSSP ontology by Pohjola (Pohjola V.J. Formalizing Waste Management in Gabbar H. (Ed.) Modern Formal Methods and Applications, Springer Netherlands (2006) 47-82), which has been reflected upon in the development work of the pyrkilo method and thereafter also the development of the Intarese method. --Mikko 15:16, 14 August 2007 (EEST) (type: truth; paradigms: science: defence)

⇤--3: . Argument 2 is based on the terms used in the fundamentals of mathematics. Therefore, argument 2 is stronger than 1. --Jouni 12:20, 23 August 2007 (EEST) (type: truth; paradigms: science: attack)
⇤--2: . According to the set theory, a set is a group of objects (items) that belong to the set. This can be an arbitrary collection of objects. In contrast, a class is a collection of sets (or sometimes other mathematical objects) that can be unambiguously defined by a property that all its members share. Therefore, class is what we mean when we talk about generalizable objects in the pyrkilo theory. --Jouni 12:20, 23 August 2007 (EEST) (type: truth; paradigms: science: attack)

Argumentation about variables themselves (not their content) 18.4.2007

The current set of rules says that "However, a new variable must not be created, if there already exists a variable with the same scope."


  • If the scope is changed by one unimportant word, it is not the same. Thus, this rule does not actually prevent formation of overlapping variables.
  • World cannot be divided into a set of exclusive and mutually exhaustive variables (variables that together cover everything but do not operlap) that would make sense in general.
  • There was no rule to deal with partially overlapping variables. (Although this need was recognised.)
  • If creation of variables is prevented, we restrict the process to things that happened to be out there first.


  • We need rules about argumentation about the existence (not only about content) of variables. We should think what these rules are.
  • It should be possible to invalidate a variable based on some criteria and argumentation. This, not rules about creating new variables, is the way of getting rid of poor variables. The criteria could be something like these:
    • Variable is irrelevant: no causal connections to anything, and therefore it cannot be a part of any risk assessment.
    • Variable is conceptually defect: the scope does not form a coherent entity. (Not even the oldest man in the world understands what the variable is about.)
    • Variable does not have clear boundaries: it is not clear what belongs to the variable and what not.
    • Variable is not measurable: A clairvoyant cannot give an answer to the result. (Note that this does not mean that the result must be numerical! Excellent and red are valid results.)
    • Variable is inefficient. There is another way of expressing the variable, and it is more practical (e.g. less time consuming to update, or data is better available). This is always compared with another variable; at least one variable must remain valid. This may also be a case when a variable is merged to another one that has a wider scope.
  • To make this work, we need an efficient categorisation of variables. This can be an extension of the issue framing method.
  • The discussion about variables cannot be held within the variable itself (at least not always: inefficiency is always about several variables). Therefore, there should be a place for this argumentation. I suggest that this is held at the class level. This means that classes become an inseparable part of the method.

Generic variable attributes

Generic variables form the elements of the scoping diagram. Similar to (specific) variables they are objects that have attributes (e.g. name). However, the main aim of generic variables is to help find the relevant pathways in a whole net of possible pathways in the causal chain. Therefore most of the attributes of (specific) variables do not apply to generic variables. Instead they have at least the following attributes:

  • name
  • predecessor
  • successor
  • subgroups
  • and other attributes to define the relationship to other parts of the causal chain.

To create a new generic variable, the user can open a template and fill in all the attributes. Then, the generic variable can be used equally to the predefined generic variables. If the user cannot fill in all relevant information the generic variable has only restricted impacts on other parts of the causal chain.

How to read discussions

Fact discussion: .
Opening statement: Class criteria

Closing statement: Resolution not yet found.

(A closing statement, when resolved, should be updated to the main page.)


Alex: What's the difference between scope and definition?

Jouni: Scope is the list of the common properties. Definition is the inclusion criteria. E.g. you may have a class with scope "PM2.5 yearly average is between 5 and 100 ug/m3". The inclusion criteria can be that all variables that describe areas > 10 km2 and are outside cities larger than 1000000 inhabitants fulfill the criterion. So, you are not allowed not to apply the class for PM2.5 variables unless you are able to show that

  • your variable is about <10 km2 or a city >1000000, or
  • the criteria is not valid

Alex: Well, I thought before that the criterion reflects the scope and thus would be „all variables which refer to a PM 2.5 yearly average concentration between 5 and 100 µg/m3 are in this class”.

In a world full of variables it would not be obvious to check for PM2.5 conc. when designing an area variable. Maybe you are not interested in PM2.5 at all because you look at amount of nematodes per m2 soil. So then you need to bother to find out your PM 2.5 concentration and either accept the class application or need to prove that the class criterion is wrong? Nobody would do this! --Alexandra Kuhn 10:51, 10 June 2008 (EEST)

Jouni: You are right, but the problem is not that the idea of class is wrong or unfeasible (because it is not) but that my criterion was stupid. Of course we must add "all variables about ambient PM2.5 concentration" to the inclusion criterion.