Very nice and useful post. Although not a book, I would also recommend the recent review by Schölkopf: https://arxiv.org/abs/1911....

Amazing flowchart Brady! For people who don’t like graphs, I’d also add “Observation and Experiment” by Rosenbaum, which is a very gentle read and guide into the potential outcomes methodology. See:
https://www.amazon.com/Obse...

Also I should mention that the Morgan and Winship book (that you mention) covers both graphs and potential outcomes, and really is able to outline the best of both (and no, it’s not quite true that graphs subsume potential outcomes, as a lot of the econometrics methodological tools can’t be expressed with graphs.)

Thanks, Nicolas! This suggestion actually prompted me to add an "Other Books to Consider" to section to the post.

I'm interested to understand what can be expressed in potential outcomes that cannot be expressed using SCMs. I understood the basic potential outcome to be a derived quantity in the SCM framework. I'll make sure to read the Morgan & Winship book some time in the next 6 months (the econometrics books are next on my list).

This paper, by Guido Imbens, discusses some standard methodological estimation practices in econometrics that would lead to extra-graphical information in a standard SCM framework: https://arxiv.org/abs/1907....

Great Recommendations! I went over "Elements of Causal Inference" recently to see how I should integrate causal discovery with neural networks. I found that the book contains tons of details and I'm not sure I understood the book material completely. I truly enjoyed the causal course you put on YouTube based on your own book, and I understand that the course was meant to be an introduction to causality. I know it might be too much to ask from you, but were you consider offering an intermediate level causality course, could it be possible for you to use "Elements of Causal Inference" as the textbook? I would really love to hear your version of the book material. Thanks again for the amazing course and for book recommendation.

@bradyneal your causal inference course looks fantastic. May I suggest you take a look at www.linkthebook.com? After inventing transfer learning and writing the original book with Sebastian Thrun, I left academics and was very surprised to observe how little machine learning was used for the largest and most complex decisions out there. This started a journey that led to the creation of decision intelligence which is summarized in this book. It may look weird to you because it is not from an academic, but from a commercial application perspective, to help to bridge this gap. Decision intelligence depends critically on a subset of causal inference that bridges from the actions that people can take to the outcomes that these actions achieve through a causal pathway. I think from that point of view it is a good complement to the rest of your course and I would be honored if you would consider recommending this book as well.