The article linked below espouses a concept I’ve always found to be true but never fully appreciated as clearly as it is explained here.

The basic idea is “learn by doing” but I what I love most about this article is the way the author provides a systematic approach to this philosophy. I’ve been trying to learn Machine Learning lately and have noticed how easy it is to get bogged down in the details. Per’s advice is perfect here: start building something. You will initially be confused and unsure how to integrate what you learn into a broader framework. Go with it. Let go of that perfectionist impulse to understand every little detail. Right from the start you’ll have a goal, you’ll watch something come to life, and eventually, if you keep at it, the pieces will fall into place.

As an aside, the writer here, Per Harald Borgen, is one of the co-creators of an amazing new learning tool, which provides a seemingly magical experience: the ability to pause a running video and play with the code in progress. Check it out at

The Easy Way To Learn Hard Stuff - Learning New Stuff

The last couple of years I've spent a lot of time teaching myself web development and machine learning. While the subjects have differed a lot - from Javascript, Node and React to Python, Scikit Learn and neural networks - my approach to learning has stayed the same.