I've been arguing this a lot with my friend Hilan.
As I see it, atomism is the only real model for any theory of change we can have.
In other words,
Q: how or why does X change?
A: We see X as being composed of sub-components, which are capable of changing relative to each other. (Changing their positions, their dispositions, or whatever) And our explanation for the change in X is to point to those sub-components
and their changes.
It's the "atoms in te void" explanation that goes back to the pre-socratics.
The purpose of holism is allegedly to shift our focus away from analysing wholes into components and to focus on understanding the connections between things. But as far as I can see, it's not holism but atomism which actually allows this.
It's only if we understand something as being composed of parts, that we can focus on the relation or connection between those parts. True holism, which should ask you to ignore the parts, should also ask you to ignore the interaction between the parts. It leads to an ineffable, black box view of the system you want to understand.
- Holist : "It goes from state A to state B"
- Me : "How?"
- Holist : "I dunno! Gradually."
But even a mathematical model, which describes a system as a set of variables and equations is a form of analysis into atoms. These atoms are the terms in the the equations, they may not map to conventional objects, but the theory of change this models, as X slides against Y, is still a population of individuals, moving past each other in the void.
Oli usefully demarcated HowAndWhyQuestions (which I've moved to that page), and continues :
Indeed, the interesting point is that usually when you have multiple explanations like this you are in the situation where none of the explanations can be considered complete - and the explanations live in sufficiently different domains that they cannot be combined into one mega-explanation.
Indeed, Berkan has done some good thinking on this - and we've had some great conversations about it ;)
So, Phil, the question then is whether it's 'useful' to declare that all variables in models can be considered as 'atoms' of the explanation - and thereby that all model based explanations are atomist?
I'd be very interested in Berkan's theories, but remember he is still banned from ThoughtStorms, pending sending me two chapters of his thesis to read ;-) Can you give a resumé?
On the is this useful question. Maybe you can just say I'm defining holism too harshly; and of course it really just means, "look at the wider context". If it does, I accept, it's a good heuristic. It's probably only as useful to worry about the similarity as it is to worry that explanations are wrong because they are atomistic.
See also TypesOfExplanation