# 01 - Why Model?
{%hackmd G-uuuRi2RyKS_IyjBJS3Kw %}

###### tags: `Model Thinking` `Courses`

## What are models?
Models are abstractions and simplifications that help us understand reality.

>_... all models are approximations. Essentially, all models are wrong, but some are useful. However, the approximate nature of the model must always be borne in mind...._ [George Box](https://en.wikipedia.org/wiki/George_E._P._Box)

It takes experience to distinguish the models that will be of use. One prerequisite though, is having multiple models and developing the ability to speak across them. If you only have one model, you'll often be wrong.

The inaccuracy of models obliges two responses:

1. We must constantly refine and improve them.

- We must open dialogues between models and reality. By identifying when a model fails, we learn more about the conditions necessary for it to work.

2. We must accumulate collections of models.

- Models don't have flaws so much as they have limited scope. Models are simplifications. Each focuses on only a part of the whole. By possessing a collection of models, an individual or a collection of people can see the limitations of each singular approach and gain a more comprehensive understanding.

We can sometimes combine two or three models to learn about interactions between variables, but we cannot construct a grand model containing every possible causal factor. If we did, and that's assuming we had enough data to disentangle all of the interaction terms, we would wind up with something as difficult to comprehend as the real world.

__Recommended Book:__ [[Bibliography#Expert Political Judgment|Expert Political Judgment]]

## Key element to have in mind when working with models:
### Parts of a model:

* Assumptions
* Results
* Applications

### Technical details:

* Measures and Proofs
* Practice Problems

### Fertility

* Where else does it work?
* New and not obvious applications.
## Why Model?

Models make us clearer thinkers.

### How models make us clearer thinkers?

Models and narrative ideas are quite similar. We can map the parts of a model into the building blocks of sentences, nouns, adjectives, verbs, adverbs, and so on. If models just capture common intuitions, why then go to the trouble of constructing a disease model? Why not just give a verbal description?

Here's why: _to get the logic correct._

In writing a narrative or drawing an analogy, we can be looser. We leave vague some of the particulars. Sure, we can omit details in a model too, but we have to say what we're putting in and what we're leaving out.

Anyone who ventures a projection, or imagines how a social dynamic would unfold is running some model. But typically, it is an implicit model in which the assumptions are hidden, their internal consistency is untested, their logical consequences are unknown, and their relation to data is unknown.

### How to build a model?

1. Name the parts.
2. Identify relationships.
3. Work through logic.
4. Inductively explore.
5. Understand classes of outcomes.
6. Identify logical boundaries.
7. Communicate.

### Models and Data:
Models lead us to data

1. Understand patterns.
2. Predict points
3. Produce bounds
4. [Retrodiction](https://en.wikipedia.org/wiki/Retrodiction)
5. Predict other data.
6. Inform data collection.
7. Estimate hidden parameters.
8. Calibrate.