If you want a model that is useful, you need to leave some things out. You can!t have a
map that is 1:1, it would be impractical. It is the same in the art of making models.
In Search of an Analytical Knife:
-Every model by deﬁnition omits certain dimensions of reality (but are we leaving things
out that are important? How do we decide?)
-Every model therefore includes implicit fundamental assumptions on what matters
-There is always more than one way to cut the "cake!, there are always going to be
competing explanations (Using a geological map vs. a population density map)
Correlation vs. Causation. A population density matches a frog population map. Why?
Well, because the stork brings babies, and storks eat frogs, right? Water perhaps?
Where will most people see frogs to make these maps… in places they live? Climate?
-But they are only useful with we can falsify them
Look at a bipolar system. We use this term because we feel that it explains something…
(Cold war and why it mattered). But there is a lot of theorizing that goes into the idea. It
assumes the presence of a system, and goes on from there.
-Theories need to be falsiﬁable to be useful -> empirical evidence
-What do we do when a theory is falsiﬁed?
-Modify theory and start again (avoid ad hoc assumptions)
-When should we abandon a theory?
-Problem: probabilistic hypotheses (democracies never ﬁght each other vs. joint
democracy reduces the possibility of war)
Example: Golf causes world peace!
Deﬁne golﬁng/non-golﬁng nations:
#1 course per million of population
Check the war involvement of golﬁng nations: never on opposite sides of a war?
1) Britain vs. Argentina (prob. fewer golf courses in 1982) and Northern Ireland
(Basques in Spain?)
2) South Africa (how do we deﬁne war? Domestic or international?)
In search of a causal mechanism: Why does golf stop warfare?
-Golf occurs mainly in wealthy countries!
-Golf is correlated with wealth