Marvin Minsky’s “Computation” (Book Review)

“Computation: Finite and Infinite Machines” by Marvin Minsky, published in 1967, is meant to be a sort of textbook about basic theoretical topics in computer science (Minsky used the book himself in teaching a course at MIT).  Minsky takes the study of “computation” to be an exploration of what machines can and can’t do, and how to best study and conceptualize “machines”.

In so doing, Minsky covers much of the classic ground of what is traditionally studied as “computation” – Turing machines, finite state machines, simple neural network models, the “halting problem”, and so forth.  There are many books which cover these topics, but where I think this book stands out is that it doesn’t just dive into the details.  Instead, throughout the book Minsky steps back and tries to address the bigger picture.  He asks questions such as “What is a machine?”  “What is computation?”  “What could be meant by a description (in the context of ‘describable numbers’)?”

I think this is important because we haven’t really settled these more fundamental questions yet.  For example, we haven’t yet settled the question of what the Church-Turing thesis might mean about what can and can’t happen in the physical world (see a previous post for more explanation of this).  This book is good introduction to some of these questions.  Thinking about what computation means and what its limitations are is also becoming increasingly important, as we model the mind more and more with computational models.

As a side note, Minsky has also authored a number of other books, including another of my favorites, “Society of Mind” (1986) in which he lays out his basic thoughts about the nature of the mind.

References

Minsky, M. (1967). Computation: Finite and infinite machines. Englewood Cliffs, NJ: Prentice-Hall.

Minsky, M. (1986). Society of mind. New York: Simon and Schuster.