While reading Franklin's autobiography, I was completely struck by the systematic, almost algorithmic way he taught himself to write. He designed a process of compressing, reconstructing, and comparing his work against a "gold standard" to find and correct his "faults."
The more I looked at his 4-step loop, the more it resembled the modern training loop for a machine learning model (data collection, forward pass, loss function, backpropagation/gradient descent).
I wrote this piece to explore that parallel, how Franklin basically reverse engineered the principles of deliberate practice and error minimization 250 years before we had the formal language of AI to describe it. It's a fascinating piece of history, but also a practical model for how anyone can learn a complex skill.
spf13•4mo ago
While reading Franklin's autobiography, I was completely struck by the systematic, almost algorithmic way he taught himself to write. He designed a process of compressing, reconstructing, and comparing his work against a "gold standard" to find and correct his "faults."
The more I looked at his 4-step loop, the more it resembled the modern training loop for a machine learning model (data collection, forward pass, loss function, backpropagation/gradient descent).
I wrote this piece to explore that parallel, how Franklin basically reverse engineered the principles of deliberate practice and error minimization 250 years before we had the formal language of AI to describe it. It's a fascinating piece of history, but also a practical model for how anyone can learn a complex skill.
Happy to answer any questions!