Think Complexity: Exploring Complexity Science with Python - 2e

Reviewed by Stephen Davies, Associate Professor, University of Mary Washington on 5/13/19

Comprehensiveness rating: 5 see less

It's difficult to be truly "comprehensive" in the field of Complexity, since it is so broad and so (relatively) new. An introductory text would do best to give the flavor of the field and enough interesting applications to pique the reader's interest, which this book does.

This text's references to approachable works in related fields will be of great value to some readers; there aren't too many of them, and those I recognized are indeed accessible to the audiences who would read this book.

Content Accuracy rating: 5

The content is accurate, error-free, and unbiased.

Relevance/Longevity rating: 3

Coverage of the main topic itself ("Complexity") is general enough to stand the (immediate) test of time.

The example code is unfortunately in Python 2, which is not forwards-compatible with Python 3 and which is (slowly) becoming obsolete. The book would benefit from upgrading to Python 3.

Clarity rating: 4

Overall, the presentation is exceptionally clear and accessible to an introductory audience. The author does a terrific job in making a potentially esoteric subject approachable and interesting.

The text does assume a good deal of background knowledge, in a broad variety of subjects. To cite just one example, chapter 5 makes statements about "long tailed" and "short tailed" distributions without defining those terms. It seems unlikely to this reviewer that most computer science students would have previously encountered those terms, and remember them so well that not even a refresher is necessary. As another example, it is doubtful that many computer science students without a signal processing background could follow the arguments in chapter 9 (self-organized criticality) with just the information presented.

In a few places, the text is a bit too terse. Having taught Dijkstra's shortest-path algorithm in a Discrete Math course, I doubt that the brief description of it in section 4.5 would be enough for most students to successfully implement it as the exercises call for. Similarly, few readers, no matter their background, could actually implement Schelling's segregation model (called for in exercise 10.1) with only the information given on p.91. Thus instructors should expect to provide their classes with significant additional explanation in resources in many cases.

Chapter 5, which presents the concept of statistical distributions, is lacking figures that, in my mind, are crucial for understanding the difference between Gaussian, exponential, Pareto, etc. families. The exercises call for the reader to implement these from scratch and plot them, but even if most readers would actually follow-through on this, the text itself would be enormously enhanced by simply depicting a probability density function for each of the major distributions discussed.

One might argue that casting Conway's Game of Life in terms of "convolution functions" obscures what is otherwise a natural and easy concept.

Consistency rating: 5

The text is internally consistent in terms of terminology and framework.

Modularity rating: 5

The chapters are very well-designed and modular. Instructors will have a good deal of flexibility in which chapters they choose to cover and in what order.

All of the case studies at the end of the book could be adapted to make promising self-contained projects for students. For most of them, this will require some investment on the part of the instructor, as these appendices do a better job describing existing studies than they do giving guided instructions for writing one's own simulation. They are all quite well-written, however, and present intriguing problem areas.

Organization/Structure/Flow rating: 5

Overall, the organization is quite strong. The text does occasionally wander into programming concepts that, while they would be interesting in a book about programming languages or Python idiosyncrasies, are a bit of a distraction from this book's stated context and purpose. The foray into iterators and generators in chapter 2 is an example, as are hashtables and list comprehensions from chapter 3.

In places, the author calls for the reader to read something external (often a Wikipedia article; sometimes a seminal research paper) and then answer questions about it. In some cases, future sections of the text depend on the knowledge gained from this process. This makes the book somewhat less self-contained and perhaps less lucid. (If instructors actually assign these readings for credit, however, this could be considered a strength of the text rather than a weakness.)

Interface rating: 5

The text is well-formatted, indexed, and easy to navigate. The only weakness here is the aforementioned lack of figures in certain areas.

Grammatical Errors rating: 5

I found no grammatical errors in the text.

Cultural Relevance rating: 5

The author has a fascinating way of putting complex systems into the overall context of the history of science. The various sidebars about contributors to the field (Erdős, Dijkstra, etc.) are amusing and well-situated. They give additional flavor to an already readable text, and help connect very technical concepts to human concerns.

This is a very well-written and entertaining book on a somewhat unusual combination of topics. It does a great job making the theory of complexity "relatable" to a college audience by posing intriguing, yet approachable, questions raised by the computational models. Chapters 6 and 7 in particular are outstanding in this regard.

Especially well-done are the philosophical questions the book raises about what kinds of models do in fact lead to reliable knowledge about the systems they represent, and which do not. Far too few texts in this and other fields even mention this issue, let alone demonstrate its importance and point the reader to how it might be answered. The discussion of holism vs. reductionism in chapter 9 is one of several sections that stand out in this way.

Although the subject it describes is multi-disciplinary, this book is probably best-suited to a computer science audience. At times it divagates into somewhat off-topic concerns that would nevertheless be of interest to computer science students (examples: how iterators and generators work in Python; how to use the pyplot package; a summary of algorithmic complexity; object-oriented API design).

Excellent comparison of "classical models" (based on physical laws and equations) with complex system models.