The Book of Why 9.3分
读书笔记 Intro: Mind over Data

# The Time of Statistics

Causal inference (WHY?) is an innate ability of humans.

Some tens of thousands of years ago, humans began to realize that certain things cause other things and that tinkering with the former can change the latter.

We are lacking the tools and principles (and languages) to deal with casual quesitons in the complex situations.

Scientifc tools are designed to meet scientific needs.

However, when modern "statistics" developed, statisticans failed to answer the questions about "heredity" and turned to develop a thriving, causality-free enterprise --- "Statistics".

When you prohibit speech, you prohibit thought and stifle principles, methods, and tools.

Causal vocabulary was virtually prohibited for more than half a century.

(Statistics) tells us that correlation is not causation, but it does not tell us what causation is.

Data Science cannot tell "WHY".


# Causal Revolution

The calculus of causation:

1) Causal diagrams: What we know;

2) Symbolic language (Resembling algebra): What we want to know;

"do(X)" vs. "x"

P(L | do(D)) do-operator signifies that we are dealing with an intervention rather than a pssive observation. (would wash away preexisting differences and provide a valid comparison) P(L | D) standard conditional probability is a passive observation (let people decide what to do, affected by various factors) A world devoid of P(L | do(D)) and governed solely by P(L | D) would result in paradoxes.

"Counterfactual" -- a retrospective thinking

We can emulate human retrospective thinking with an algorithm that takes what we know about the observed world and produces an answer about the counterfactual world (what-ifs).
Language shapes our thoughts. You cannot answer a question that you cannot ask, and you cannot ask a question that you have no words.
A causal reasoning module will give machines (AI) the ability to reflect on their mistakes, to pinpoint weakness in their software, to function as moral entities, and to converse naturally with humans about their own choices and intentions.


# Causal Model

Why we need this complex model?

Our causal intuition alone is usually sufficient for handling the kind of uncertainty we find in household routines or even in our professional lives. But if we want to teach a dumb robot to think causally, or if we are pushing the frontiers of scientific knowledge edge, where we do not have intuition to guide us, then a carefully structured procedure like this is mandatory.

Data is dumb, and “data-driven” has bad “adaptability” .

You are smarted than your data.
《The Book of Why》的全部笔记 46篇
免费下载 iOS / Android 版客户端