I found https://www.lesswrong.com/posts/no5jDTut5Byjqb4j5/six-and-a-half-intuitions-for-kl-divergence https://www.lesswrong.com/posts/no5jDTut5Byjqb4j5/six-and-

I found https://www.lesswrong.com/posts/no5jDTut5Byjqb4j5/six-and-a-half-intuitions-for-kl-divergence https://www.lesswrong.com/posts/no5jDTut5Byjqb4j5/six-and-a-... very helpful for getting intuition for what the K-L divergence is and why it's useful. The six intuitions: 1. Expected surprise 2. Hypothesis testing 3. MLEs 4. Suboptimal coding 5a. Gambling games -- beating the house 5b. Gambling games -- gaming the lottery 6. Bregman divergence