Buddy, can you spare some empirical data?

Posted by mouthyb | Posted in | Posted on 2:03 AM


I hear this is the way it goes: as a student, you read a theory or a set of results, and it is as if a key turns in your head.

Like tumblers, the ideas unite a mass of inchoate or partially congealed thoughts and the world becomes, if only for a moment, clearer. The view through the door is clean, elegant and fascinating.

The first time I read about Complexity Science, that is pretty much exactly how I felt. It united a series of observations I had made privately about the nature of human systems, and a set of reservations I have about the subjects I most love.

I love theory that deals with humans and human behavior, but the ongoing necessity of making exemptions in top-down theories of behavior motivation really bothered me. If conflict, for instance, is our only motivator, what is to be done with negotiating strategies which do not employ threat? Do we make a blanket pronouncement that all negotiation is implicitly threatening?

Is that really a place a theorist has any business going, categorical statements built on revisioning the definitions of the terms involved?

Many times, this causes theorists to go back over and over and redefine their terms. Statistically speaking, at least in terms of statistical inference, it is bad damn design to go back and tweak bits of the experiment and terms until it fits. All that tweaking confuses the issue and obscures the relationship you're examining. In this case, conflict starts to become a blanket term, as does any central term in a theory, until it covers so much human behavior that anything is conflict. At that point, you can't test it because there's nothing it doesn't cover.

It can't be wrong, by its definition. And for that reason, it can't be right, either.

To be science, it has to be able to be wrong. I love huge theories, but if they can't be wrong, I have to put them down like I would any shiny toy when I have work to do.

This is precisely why I love network theories. Complexity science, as applied to humans, starts with the assumption that pockets of order are created from disorder, by the application of energy. If, in the case of conflict, I am interested in studying an instance of conflict, I must define it in terms of actions which can be mapped, because I will be looking for the application of energy which caused a pocket of order (the situation of conflict) or node of intersection to form.

As always, I have to apply some assumption to the situation which I can test. I may assume, for instance, that many conflicts result from disrupted habits. To define it in terms of energy, I may say that my hypothesis is that when someone is jarred from their regular patterns, it causes them to have to expend energy they have not budgeted for, which causes their mood to be poor.

If I wish to keep applying science to that theory, I'd better also test if there is no effect on the mood. a separate test.

And to test, I need actions which I can measure.

I like network theory precisely because it does not start at the top and redefine until something emerges, like Frankenstein, to lumber into the light. It forces the testing of the null and the alternate hypothesis, and forces me (very junior scientist that I currently am) and other scientists to start with what can be observed in order to explain human behavior.

The world I can observe is quite complicated enough.

Comments (0)

Post a Comment