First Principles Thinking Explained with Examples

elon musk first principles thinking

In this post, you will learn about the concepts of First Principles thinking with the help of examples. The following topics will be covered in this post:

  • What are first principles?
  • What is first principles thinking?
  • Examples of first principles thinking

The first principles thinking can be used to solve any type of problem including real-life problem, product related problems, science related problems etc. 

What are First Principles?

As per Wikipedia, the first principle is a basic proposition or assumption that can not be derived from any other proposition or assumption. From Mathematics perspective, first principle can be thought of as Axioms. Axioms are propositions which are regarded as accepted or self-evidently true. For example, 2 x 2 = 4 or any number multiplied by zero is zero. From Physics perspective, the theoretical work is said to be from the first principles if it starts from the established laws of Physics. For Philosophy perspective, first principles are considered as the prior established facts based on which conclusions are derived.

Simply speaking, the first principles can be considered as the basis or origins from which a thing can be understood in a better manner.

What is First Principles thinking?

First principles thinking is about breaking down a problem statement into the basic elements by asking questions and creating solutions by combining basic elements in an innovative manner. First principles thinking removes all assumptions when looking for the solution of the problem.

There can be different ways in which first principles thinking can be applied:

  • The five whys: Start breaking down the problem in sub problems by asking “why”. The popular “why” technique could result in the statement which can not be deduced further.
  • Cartesian doubt: Cartesian doubt is a method adopted by Descartes. The method recommends doubting everything one could possibly doubt until one is left with what can be seen as purely indubitable truths.
  • Socratic questioning: Socratic questioning helps in¬†arriving at the first principles in a systematic manner. The following are some of techniques used in Socratic questioning:
    • Examining the ideas origin
    • Challenging assumptions
    • Looking for evidences
    • Consideration for alternative approaches
    • Examining consequences and implications

The first principles thinking has recently become very popular because of Elon Musk. He openly has made the statement in different forums that he applies the first principles thinking to solve several problems.

Examples of First Principles Thinking

Let’s try and understand how first principles thinking can be used to solve a machine learning classification problem. Lets say there is a problem of classifying an image as cat or dog. Let’s start by breaking down the problem statement into sub-problems. Here could be some of the sub-problems:

  • How do we classify an image?
  • How do we differentiate a cat and dog image?

Let’s break each of the sub-problems to another set of sub-problems. We will take up the problem statement – how do we classify an image:

  • By color (self-evident truth)
  • By objects present on the image
  • By how different objects are laid out on the image

What are objects on the image made of?

The objects on the image are built of pixels.

What are pixels made of?

The pixels are made of different combinations of red, green and blue. And, pixels can be represented as numbers. For two colors black and white, 1 bit is enough. Two bits can be used to represent four colors. Four bits can be used to represent 16 colors and so on and so forth.

Thus, it can be said that the images can be classified based on how pixels are laid out on the image.

From machine learning perspective, if we can convert the image into pixels (numbers) and feed into machine learning algorithms, we can classify the image.

Ajitesh Kumar
Follow me

Ajitesh Kumar

I have been recently working in the area of Data Science and Machine Learning / Deep Learning. In addition, I am also passionate about various different technologies including programming languages such as Java/JEE, Javascript, Python, R, Julia etc and technologies such as Blockchain, mobile computing, cloud-native technologies, application security, cloud computing platforms, big data etc. I would love to connect with you on Linkedin.
Posted in Reasoning. Tagged with .

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload the CAPTCHA.