The site for the course is
http://nand2tetris.org.
The goal is to to educate yourself about how the modern computing stack works from the ground up - "from nand to Tetris." With that in mind, you need to choose, on your own, how best to learn this. For example, if you just read the book and do the projects, and never independently dig into topics like "flip flop," "memory mapped hardware" or "stack-based virtual machines," you would be doing yourself a disservice.
In other words, there is no "rule" about how to learn this stuff. As a "self-learner," you just need to answer to your own curiosity and questions.
The projects in the Coursera course are the same as in the book. The lectures on the Coursera site supplement the book, as do the lecture slides on the web site.
Here's what I did:
- Read Code, by Petzold, chapters 1 - 10
- Read Code chapter 11, and ECS all pages through chapter 1
- Complete project 1
- Read Code chapter 12, and ECS all pages through chapter 2
- Complete project 2
- etc...
And obviously using other course materials (lecture slides, Coursera vids) and Internet resources (including this forum!) as necessary, to answer one's own questions.