转自amazon：Cybernetics

By kalm77 (New York)


Norbert Wiener, by his own admission, was a big fan of what today might be called "interdisciplinary studies", or "the no-man's land between established fields" as he describes it. In the process of describing his past accomplishments as a sort of cross-pollinator/evangelizer of cybernetics, he makes the case for a broad educational background, as a useful alternative to the darwinian academic trend of earning a living from increasing specialization.

Despite being an extremely sharp mathematician, and openly pooh-poohing the use of emipiricism and the application of maths to the social sciences, he nonetheless understood and even cared about the immense social impact of the technology ("technique" as he calls it), being developed by him and colleagues on the lay man of the future. He refers to the inevitable devaluation of future human beings, being forced to compete against computers, just as physical labor can not compete against automation ("...there is no rate of pay at which a US pick and shovel laborer can live which is low enough to profitably compete with a steam shovel as an excavator" serious aside: what does this say about the entire monetarist efforts of "the Fed" in tackling unemployment today! futility?).

Ironically, for a mathematician so deeply involved in the exposition of the advanced applications of the math of group theory, thermodynamics and statistical mechanics, to signal processing and control, his broad economic and societal comments (about the role of God, the undeniable technological impulse/drive of Man throughout history, Labor Unions, Capitalist "Hucksters" etc) are the most prescient and catchy part of his work! (maybe because I am not a mathematician myself).

But, these social themes are NOT what Cybernetics (2nd ed) is all about. Norbert has written many many other books solely musing over these social themes very profoundly! In Cybernetics, the preface and first chapter represent the full extent of his thoughts on this area, which he alludes to again in the second half of the book.

Cybernetics, is an eclectic work with several themes or threads. The eclecticism derives from his pure genius to see common patterns emanating from what one could call the study of "Signals and Systems" applied to biological evolution, development of instinct, learning, all the way to the more commonly explored information theory, computing machines, and feedback theory. Another significant overall theme is of the utility of modern statistical approaches in the study and analysis of non linear systems, to time's arrow, thermodynamics, and evolution itself. He certainly grounds all his observations (with many amazing nuggets to be picked up by an alert reader, along the way) by alluding to rigorous if not totally un-refutable mathematics.

The second chapter is the 'K2' summit of the maths to be encountered in the book (good luck). This chapter touches on topics which can be considered a good chunk of what qualifies today as standard rigorous graduate level course work in the field of signal processing in electrical engineering; for example, how fourier series are almost a second nature outcome of group theory and invariant transformations; how the ergodic theorems apply to Gibbs' statistical mechanics and indeed make noise filtering possible in communications. The next chapter goes into the development of theory of feedback and control. Then, in the rest of the book he leaves signal processing and he segues on to computation, artificial intelligence, gestalt and perception, and self replicating machines! One of the famous (infamous?) memes from this book is that "reality" is actually "computed" by humans, not observed or sensed.

Strangely, the blurbs/endorsements on the back cover of the second edition showcase reviews by a very diverse (to say the least) set of mass market publications (eg. NY Times, Electronics Magazine, Saturday Review of Literature), as if this book were a fun sunday afternoon's read: "lucid, direct....as readable by the layman as the trained scientist". Yeah. I highly doubt any of the reviewers mentioned in the blurbs really read the book (much less the even more mathematical first edition). It may be a fun read for the "layman" if the layman happened to be employed as a post-doctoral research fellow in signals and communications labs, certainly. But for the rest, may i humbly recommend a prerequisite of an undergraduate engineering degree to get the perspective and thrust of ideas Wiener wants so much to impart.

If you have an engineering degree somewhere, it can be a very profitable read. The only bones I have to pick are that this is an amazingly top-down book in 200 pages. Wiener tackles his subjects on from such a high conceptual level, that he can not go into the more pedestrian level which is also an extremely interesting discussion of the many themes he touches. (But then I feel no book should be more than 200 pages in today's time!) How can you drink from a fire hose? Verry carefully! But the fire hose has it's purpose too!

It is also important to read this book in the context that this author represents the culmination of the knowledge of computer theory at the very point of the introduction of computers into broader society. Prior to this, they were mathematical department curiosities. Certainly WW2 played a part. Then there is a strange leap to the "Whole Earth Catalog" level. So from a historical perspective and to understand historical milieu it is also a significant work.