model_thinking.1

Model Thinking
==============

https://www.coursera.org/modelthinking/class/
http://ccl.northwestern.edu/netlogo/

http://modelpractice.wordpress.com/

http://openstudy.com/study#/groups/Model%20Thinking%20%28Coursera%29
http://www.geazle.com/group/21


Section 1 - Introduction: Why Model?
====================================


Intelligent citizens of the world
---------------------------------

All models are wrong, but some are useful

models are the new lingua franca - business, politics, non-profit

models clarify logic 
	give us the conditions under which certain things hold true
		two heads are better than one, versus
		too many cooks spoil the broth
		opposite proverbs website

models are better than we are

	y - discrimination
	x - calibration

	hedgehogs use a single model
		terrible at predicting
	hedge-foxes
	fox-hogs
	foxes - use lots of loose models in their heads
		better at predicting
	mindless competition
	formal models


models are fertile
	they can be applied to multiple, distinct domains

models make us humble
	we're much worse at predicting than we think we are
	they make us see the full dimensionality of the problem

Many-model thinkers know best
	foxes did better than hedgehogs
	formal models did better than foxes
	lots of formal models did better than individual formal models


Thinking more clearly
---------------------

developing a model
	name the RELEVANT parts
	identify the relationships (between the parts)

Choosing a restaurant
	name the parts
		restaurant
		people going
			how much money they have
			how much time
			preferences
				cuisine

models allow us to 
	inductively explore

	understanding class of outcome
		equilibrium
		cycle
		random
		complex

	identify logical boundaries
		the conditions under which one thing holds, and another doesn't
			when is it the case that two heads are better than one?
			when is it the case that too many cooks spoil the broth?

	communicate


Using and Understanding Data
----------------------------

Why you model

1) understand patterns
	flat line, sloped line, wave cycle, spikes, etc
2) predict points
	if houses cost $100/sqft, a 2000 sqft house is $200,000
3) produce bounds
	inflation 10 years from now will be between 0-3%
4) retrodiction
	
5) predict other stuff
	derive unknown variables based on known variables
	predict previously unknown things
6) inform data collection
	name the parts
7) estimate hidden parameters in the model
	
8) calibrate

model refinement
	using the model to tell us more about the data
	using the data to tell us more about the model


Using Models to Decide, Strategize and Design
---------------------------------------------

1) real time decision aids - help figure out when to intervene (or not)
2) comparative statics - what's likely to happen if we make this choice?
	moving from one equilibrium to another
3) counterfactuals - what would have happened if we hadn't chosen that course of action?
4) identify and rank levers - which entity/factor among many has outsized impact
5) experimental design - 
6) institutional design - should we have a market? a democracy? a bureaucracy?
7) help choose among policies and institutions