Reinforcement Learning History
This batch includes Robonaissance/Substack material on the history of reinforcement learning, including a note about Christopher Watkins, Richard Sutton, Andrew Barto, and the origins of Q-learning.
Key note
One source recounts Watkins encountering Sutton and Barto’s work on learning systems, then later developing the algorithm that became Q-learning. Treat this as a narrative historical source rather than a fully verified chronology until corroborated by primary references.
Research interest
The interesting angle is historical contingency: Q-learning emerged from cross-pollination between animal learning, control, and expert-systems-era AI. For current RL-for-LLMs work, this page is useful as a reminder to distinguish enduring algorithmic ideas from the current training-stack fashion.