The Master Algorithm – Book Review

Screen Shot 2015-12-21 at 4.41.43 PMOver the past few years we have witnessed an incredible explosion of interest and application of machine learning. Machine learning has become the predominant computational paradigm, and in short term it has gone from one successful application to another. However, machine learning is far from a unified field, and many different approaches and techniques are vying for primacy and dominance. Which raises an interesting question: is it possible to find a single all-purpose machine learning algorithm that can successfully tackle all protean problems that are currently being attacked from various angles. This search for this “Master Algorithm” in many respects has the flavor of the search for a unified field theory in Physics. From everything I know about Physics (which is a lot) and machine learning (quite a bit), this seems like a very elusive goal. Nonetheless, the very possibility of such an algorithm is very intriguing, both intellectually as well as in practical terms, and this book does a great job of giving an informative and readable overview of search for such an algorithm.

In this book, Pedro Domingos does an excellent job of describing the history and ideas behind various machine learning approaches and schools of thought. His language is fluent and entertaining, while the imagery he often resorts to makes an otherwise potentially dry subject matter appear vivid and accessible. This is a general birds-eye overview of the field, and as such it does not provide any workable examples. If you are a programmer or a data scientist looking for code that you can quickly implement in your own work, you will be disappointed (although a quick search through the given references will lead you to some code and examples.) Nonetheless, even though the book is aimed at the general audience, I feel that a decent amount of prior exposure to machine learning would be very useful for getting the most out of this book. For me the book was a great way of getting the “big picture” of the machine learning field.

One of the few shortcomings that I found with this book is that it is too sanguine about the prospect of what the out-of-hand machine learning and artificial intelligence could bring about. This topic has been garnering much more deserved attention recently, but unfortunately this book missed the opportunity to delve into it in a meaningful way.

The other issue I have with the whole notion of master algorithm is that even if we were to find it, it may not be the most practical algorithm for most of the everyday problems that we deal with. Taking the analogy from Physics once again, even the overarching unified theories that we have right now are often not used or implemented in practice. Most of the problems in, say, civil engineering can be handled with the relatively simple machinery of Newtonian mechanics.

Overall this is a very interesting book and well worth reading for anyone interested in data science, machine learning or artificial intelligence.


Bojan Tunguz

Bojan Tunguz was born in Bosnia and Herzegovina, which he and his family fled during the civil war for the neighboring Croatia. Over the past two decades he has studied, lived and worked in the United States. He is a theoretical physicist with degrees from Stanford and University of Illinois. Tunguz has taught physics at several prominent liberal arts colleges and has been writing about physics, science and technology for more than a decade. He also has a wide spectrum of interests, and reads and writes about current events, society, culture, religion and politics. Over the years he has reviewed many of the books that he has read, and posted his reviews on various online outlets. In 2011 he had become a top 10 reviewer on, where he continues to be very active. Aside from reading and writing, Tunguz enjoys traveling, digital photography, hiking, and fitness. He resides with his wife in Indiana. You can follow my review updates on the following pages as well: Facebook: Twitter: Google+:

Visit Website

There are no comments yet, add one below.

Leave a Comment

Your email address will not be published. Required fields are marked *