Books are amazing. Books are compact affordable ways to help Humanity extract itself from a naive state of Nature.

Yet... books are terrible. Books actually were amazing centuries ago. Books are symbols of knowledge in the sense that as we look at a book we imagine how it will helps us learn. Yet, the truth is far remote from it. Books can be terrible, with poorly written content or even arguably worst, beautifully written content is either factually wrong or deceiving.

Books were once the state of the art of conveing knowledge. That time is long gone, if it actually ever existed. Books are terrible because they give the sense of learning. They give the impression that because one has read about a topic, they are now knowledgeable about it. And yes, imagining that if one knows absolutely nothing about a topic, even the most modest book can improve the state of knowledge of that reader. Yet, is it actual knowledge of the topic or rather the impression of it? The only way to validate or invalidate that claim is to test against reality. The only way to insure that one did learn from a book is to check that newly acquired knowledge against the object of the topic itself. That means the reader must not just read but rather test. This can be relatively inconvenient, for example of the topic of the book is the temperature of the Sun the the reader would need a complex apartus, e.g a spaceship, to go and measure. This instead of often delegated to exercises, end of chapters questions with answers from the author. The reader instead of reading what the author wrote then have to temporarily let go of the book and use their own memory of the content of the book then try to see how that knowledge can help solve the challenge. This can be assimilated to a simulation, the reader tries to simulate the topic and solve. This already shows a very different way to interact with a book then "just" reading. Yet, this leaves much to be desire in the sense that the answer provided is often succint. The reader verifies that their answer matches the one of the author. If it is correct they assume they know. A great exercise will provide ways for the reader to actually verify on their own, like a mathematical proof done 2 different ways, that the result they find is indeed correct. This though entirely redefine both the consumption and creation of a book. At that point a book is not anymore a thing to read but rather simultaneously a thing to read and a thing to exercise with.

This is a delicate situation for everyone involved. Designing exercise that are genuinely bringing the person involved to a better understanding without the ability to correct on the way is not the same skill as writing. Also having the confidence in launching oneself in exercises is vastly more demanging that reading a sequence of words and assuming they are indeed interpreted in a way that the writer would find correct. That means a traditional book to read is fundamentally different from what is usualliy refered to as a textbook. Yet, the very fact that expensive textbooks are the basis of classes, the one place and moment in time dedicated to learning, is not random. Over time the consensus has been that a book itself is not sufficient, rather it is a text intertwined with checkpoints that can validate or at least invalidate the acquisition of that knowledge that is superior. Most textbooks though are not consumed outside of the classroom. This begs the question of why. How come, if a textbook is generally regarded as superior, it is limited to a classroom whereas anybody at anytime could use it?

The hypothesis here is that both designing and actually learning from a textbook is more demanding than solely reading from a book. Consequently the classroom provide support in terms of direct help from the teacher and also motivation from a broader curriculum with social markers like a diploma. Yet, textbook in or outside a classroom themseves are also relics of the past. For decades now the computer provides a new way to both design and consume textbook. Namely that a textbook can now provide not just an intellectual environment to run exercises inside of but rather a computational environment.

A modern text provides the text, the exercises but also the computational environment to complete exercises. This sounds like a minor technical improvement but it is a radical difference because that environment becomes reality to the reader. The reader now has a place, even though an imperfect one in the sense of being simplified, where they can test their knowledge. This is a fundamental difference because the reader is not bounded anymore but the challenging yet very limited space offered by exercises and their solution. Instead the reader can complete carefully crafted exercises but also everything in between. Exercises become ways to efficiently navigate through concepts the author believe as essential but nothing more. The environment provided is of incredible value to the reader.

So yes, a book is an amazing device. It has tremendously helped us to progress due to compactness and now affordability. Today though a book is not sufficient anymore except for the pleasure of reading itself. As a device to improve knowledge the book is outdated. The book should instead become computational notebooks providing environments to explore, to learn from the reality of the topic.

Finally, if that is truly the case, how come computational notebooks are not prevalent in every field? A simple answer would be that progress takes time and that author of books might not have the skills needed to design computational notebooks. If so, time will hopefully solve that issue. A more subtle challenge though might be that the challenge of accepting to be challenged through exercises is intelectually and emotionally challenging. It requires one to be humble to let reality, even in the form of a simulated one, to push back. It always feels easier to assume one know versus discovering that no, truly, one does not. This form of interactivity can be seen as a spectrum. From consuming passively a medium, being a book to a movie, to consuming it actively while annotating it individually or socially, a form of hermeneutics, to finally interacting with the medium itself. That spectrum of interactivity might not be solely correlated to the depth of knowledge acquire but also the decision fatigue one must go through in order to complete such challenges.

If computational notebooks should replicate books as the new medium to acquire knowledge, we must remain aware of how both designing and consumming them is genuinely more demanding to everyone. Hard fun remains hard but the agency it brings to both is a truly beautiful prospect for a learning society.