I attended a short series of lectures by Kalman years ago. As I recall it, he very strongly emphasised the virtue of working directly with observed data, avoiding the biases resulting from positing a model and trying to adapt it to the data. He was quite insistent on this point, citing Newton’s _Principia_ as a good example of this principle at work: Newton, he said, did not cast around for models that might explain Kepler’s laws. Instead, he _derived_ the inverse square law of gravitation from Kepler’s laws, largely using geometric arguments. He was an excellent speaker, and highly opinionated. And of course he did explain the idea behind the Kalman filter. However, since I never needed them myself in my work, I have long since forgotten the details.