Customer Reviews

Fuzzy Thinking: The New Science of Fuzzy Logic

5 star:		(7)
4 star:		(15)
3 star:		(5)
2 star:		(4)
1 star:		(17)

 

 

Buddhist math? C'mon.

First, let me say that fuzzy logic and fuzzy arithmetic are great tools. They're valued parts of the 'soft logic' kit that includes probability, interval arithmetic, Bayesian and Markov networks, and lots of other good stuff. Fuzziness involves many of the formal techniques used in probability and elsewhere, and gives a useful, alternative view of the systems it addresses.

The basic fuzzy idea is that most descriptions involve shades of gray, that few systems really match the black/white, on/off, either/or duality of standard formal logic. That's fine, I can get along with that quite well.

My problem, though, is that Kosko presents the fuzzy world-view vs. the traditional or "scientific" in exactly the black and white terms that he rejects. He also argues that fuzziness describes the world more effectively than "scientific" terms, that the rules of arithmetic, probability, and calculus are just games. They are played for their internal consistency, not because differentiation or factorials occur in nature.

That's true, and as a heavy math user I know enough to distinguish my models from reality. Two facts remain, though. First, the models very often do describe reality in ways that can be checked easily enough: the bridge doesn't fall down and the TV receives its signal. Both happen because the bad old exact arithmetic has some kind of correspondence (no, I don't know what) to the real world, giving real ability to predict real results. Second, fuzzy logic and fuzzy arithmetic are themselves mathematical formalisms, games like all the others. Once you get past the gee-whiz stage, there is mathematical content as rigorous as in any other field of study. It's not either/or, it's very often a different way to interpret the same self-consistent games people have played for years. It adds interesting rules to the game.

The great thing is that you really can use the new interpretations and tools along with the old ones. Fuzziness doesn't demolish the old structures, it bolsters them and adds capacity.

And you can get all these benefits without shrink-wrapped, bite-sized pieces of Eastern philosophy.

//wiredweird

I was looking forward to a good popularization of Fuzzy Logic theory, but I am quite disappointed.

This is not a science book, nor is it a popularization of science. What science it does contain is unclear, is not presented in laymen's terms, and in some cases is factually wrong. The author asserts, pg. 107, that E = mc^2 is an approximation of a nonlinear equation, and attempts to prove it by pointing out where Einstein makes an approximation in his 1905 paper. It turns out that he was taking the Newtonian limit of an equation, and the limit implies the famous E=mc^2, but the E=mc^2 is not itself an approximation. This isn't a particularly serious error, and in fact he would have been technically correct until he said that the famous equation itself was an approximation. It is a sign of Kosko's attitude when he asserts that "[e]ven many physicists forget" that Einstein did this.

I do not doubt that fuzzy logic is useful, but this is not a useful primer from which to learn first principles. Rather it is a diatribe against "Aristotalean logic". Einstein's work, to use him as an example again, was impressive because it was it turned out to be correct to a high degree of precision, not because Newton's work was wrong. Einstein's attitude towards his work was "Hey look at what I found!" not "Can you believe this Newton guy? He got it all wrong!"

The author tries to unite philosophy and science and fuzzy logic and politics into one book, and fails utterly. His presentation of each of these topics detracts from the others. Douglas R. Hofstadter's classic "Godel, Escher, Bach : An Eternal Golden Braid" does this much better, in a far classier manner, and addresses quite clearly the "mapping problem" (as Kosko calls it) between logical systems and using them to describe the real world.

Fuzzy math is a clever mathematical device, and turns out to be very useful for engineering automatic computer controls for systems with several degrees of freedom, but you don't even learn this until the second half of the book, right before he dreams of applications of fuzzy logic to politics. Those who subscribe to the view that the "black-and-white" judgmentalism of science is a bad thing may enjoy the philosophical combativeness of this book, but even they will not learn anything new.

No, I do not recommend this book even in the slightest degree, except as an example of how not to write a popular science book.

Kosko struggles to articulate the upcoming paradigm shift from Aristotelian logic to multivalent Fuzzy Logic in the world of commercial technology. He does a good job of explaining the concepts behind the "new" field of fuzzy thinking, but has a hard time expressing the mathematics of it. To give him credit, though, he had to write a book understandable by the lay, and therefore couldn't get too in depth with his equations. But if you have any background in calculus, you'll have no trouble learning from the footnotes.

Aside from the introduction to fuzzy logic, Kosko dips into his personal life to talk about his contributions to the field. I found his digressions extremely interesting and his views enlightening. It's important to take a lot of what he says with a grain of salt, however (he likes to wax philosophical perhaps more than is appropriate), but he is an amiable character nonetheless.

Overall, this is a good read. It's informative not only scientifically, but philosophically as well. Kosko's style is engaging, although too heroic in some passages, and thorough. Because of its controversial nature, some readers may find it offensive. But if you don't mind entertaining alternative opinions and don't mind suspending tendencies for dismissing foreign ideas (which a lot of the people who gave this book bad reviews couldn't apparently handle), you'll enjoy this book.

Let me start with a little context. I'm a scientist/engineer, and I enjoy reading these popular science books as a primer to fields I'm not well versed in, but would like to know more about. For me, this book was less about philosophy and more about the history and underpinnings of fuzzy logic. From my viewpoint, this book fulfilled my objective. I now feel I have a sufficient introduction that I can dig deeper into this interesting topic. Will I? Absolutely. I believe there is much I can do with fuzzy logic in my day-to-day research, and I'm grateful to the author for making me aware of it.

What about the philosophical aspects of the book? The author clearly feels that western culture is hostile to the notion of fuzzy logic. I'll extend on that a little further. I think western science is hostile to new ideas in general, not just fuzzy logic. But, I will agree that it is for the same reason he cites. Western science tends to see every new idea as true or false. Since new ideas tend to contradict old ideas which have achieved the status of true, then new ideas by their very nature are instantly regarded as false. Only with the passage of much time or with overwhelming evidence does an old paradigm die. Eastern culture, by contrast, is not so quick to reject a new idea since "shades of gray" is woven into the culture. No wonder, then, that fuzzy logic has emerged in the east in real products while we still debate whether fuzzy logic exists.

Who should read this book? Definitely this is worthwhile for scientists and engineers of all stripes, but should be considered a must read for those involved in system engineering. Laypersons with a scientific interest will find this book to be accessible and will be rewarded for reading it. Others without a scientific interest may also find this book worthwhile since fuzzy logic is likely to become a technological force in the near future.

Rating is 4 stars. Good read, but I save 5 stars only for great literature.

Well, I got about 1/2 way through the book and then gave up - in disgust. This is one of the worst popular science books I have come across, excluding obvious whacko books such as by the likes of Velikovsky.

First of all, I want to say that I have no doubt that fuzzy logic is a very useful branch of maths with important applications to engineering etc and I have no doubt that Kosko has made important additions to the subject through his research.

The serious objection I have is the claim that fuzzy logic is somehow breaking ties with Aristotelian logic. It is not. At the center of fuzzy logic is the concept of a fuzzy set. Instead of an element having {0,1} membership - either you're in the set or you're not - the rules of the game are that you can have degrees of membership lying in the real line segment [0,1]. Simple concept. Now develop a theory based on this... If you read any mathematical text on fuzzy set theory you will see definitions, theorems and proofs just as in any other branch of maths. Until a hole is found in a proof of a conjecture it is a theorem ie 100% true. If a conjecture is discredited by a counter-example or proof by contradiction (techniques that are established in all branches of mathematics) then until a hole is found the conjecture is 100% false. The point is that any well-formed conjecture in any axiomatic mathematical system is either true or false although finding the answer can be extremely difficult (eg the 4-colour problem - solved using a computer - or the Goldbach conjecture - unsolved to this day). This is the same for theorems about fuzzy sets. The object (a fuzzy set) may be more exotic, but the principles of reasoning (eg modus ponens) for asserting truths about these objects has not changed from the times of Aristotle.

Unfortunately, I suspect that a lot of people without a comprehensive and advanced background in maths will regard fuzzy logic with a sense of awe and mystery as if fuzzy logic somehow represents a distinct paradigm shift. Remember folks, fuzzy logic is just another branch of maths just like its trendy cousin chaos theory and just like considerably more obscure branches of maths such as K-theory, universal algebra and differential geometry which are likely to receive blank stares at dinner parties.

Well, as to the other problems with this book... I suspect that for all the travels to Japan that Kosko has done his grasp of Eastern thought is at best tentative and naive... I'm really dubious about his connections between fuzzy sets and yin-yang concepts.

And to cap it off, I found the book to be very repetitive - in the same vein as Men are from Mars, Women are from Venus - so maybe if you liked that book you'll love Fuzzy Thinking.

Finally a word about grey. Grey is one of my favourite colours (I'm serious!). Kosko goes on about grey a lot. And I mean an awful lot. I mean on the scale of "a lot-ness" he probably gets about 9/10. Well, I, just like you, dear reader don't like to be pigeon-holed as a 100% "commie" or a 0% "tall" person. Seriously, it's kind of ridiculous how survey agencies like to put us into neat parcels such as "white" or "employed" or "happy" or whatever. Life is complex and full of shades of grey and I certainly applaud any attempt to acknowledge that fact.

I've never met Bart Kosko, but reading his book I feel as if I may have. It has a very freely associative (almost conversational) style that leads me to think Kosko lecture material might've made a substantial contribution to this work. Possibly Kosko's many discussions of the Buddha vs. Aristotle reflect the dualistic nature of Kosko himself- part parodoxical sage, part analytical rationalist- both personality poles seem to be represented in many of the Koskoian connections put forth (what Kosko might term 'the part in the whole'). Other mentionables:

1) This book is pretty light on the math, even for pop-sci material. References citied are more than adequate for in-depth investigations of this material, however.
2) Kosko's extraverted enthusiasim for his subject matter is evident and may come across as overly-excited and unstructured rambling.
3) Explanatory style is broadly associative; examples provided draw from a variety of disciplines (which is somewhat expected of someone with degrees in philosophy/econ/EE/math), although not necessarily with equal rigour. Some philosophical and historic parallels might be contended; these need not detract from larger ideas of the book as there are plenty of solid explanations to choose from.

After my first foray into this subject material it seems that as a tool, fuzzy logic posseses tremendous utility for computer scientists and engineers. As a new paradigm of thought in mathematics or physics, fuzzy logic's importance is definately more... well, fuzzy. As with any new tool, the development of new practitional paradigms might be necessary for portions of its potential user community; this book should be up to that task.

As mentioned I've never met Kosko, but would certainly like to after reading his book; he's obviously an intelligent and interesting guy. The entertainment value of its ideas and engaging presentational style earn this book a five... had it rested solely on it's educational merits and coherency would've probably garnered it a three.

There really doesn't seem to be much doubt that the book wanders across the map. It trails through Buddhism, cryonics, methods of meditation, parking lots, and apples on its way towards attempting to present its theory of fuzzy logic. I can understand why somebody quite familiar with the theories involved might find it quite frustrating as discipline and rigor didn't seem to be two of its strong points.

For me, however, it was a useful read. I have been curious about Data Mining and Neural Networks and the book was recommended to me as a place to begin with the concepts and what I found was that once I got the hang of the loopiness, it did a remarkably good job of bringing home some of the basic ideas to a math challenged person (me).

I tend to be a suspicious person, so I wasn't convinced by some of his more grandious claims and I think I like Aristotle rather better than he does, but these points didn't prevent me from feeling like I got something out of the book.

At the very least, I felt like I finished Fuzzy Thinking a bit better armed to dig further into the subject matter, which was really what I was trying to accomplish in the first place.

This book won't win any medals as a textbook, but Bart Kosko clearly did not intend for it to do so. He writes clearly, presenting the basic idea and "philosophy" of fuzzy logic in laymans terms with just enough math to get the average person interested without feeling overwhelmed. An excellect introduction to burying Aristotlian logic and thinking closer to the world experience. I highly recommend it.

I read this about 7 years ago and it totally changed my life and my view of the world. (NOTE: It changed for the better)

It's kind of a disorganized mess of a book but you'll get used to it. Read at least the first few chapters.

Basically taught me to let go and think of everything in a "fuzzy" way. I think the people of the world would be much happier if they were able to toss the idea of black-white/true-false answers.

This book incorporates a general discussion of the fuzzy logic paradigm,which was started in 1965 with an article by L Zadeh.Kosko's contributions in this field are incorporated within a framework of bibliographic sketches that bring a personal note to the book.It incorporates a philosophical emphasis on the vagueness of reality and the decision making context faced by the individual.Philosophers Bertand Russell and Max Black ,for instance ,provide support for the epistemological view of fuzzy logic.Unfortunately,Zadeh and Kosko appear to be ignorant of the fact that their general approach is not original.Certainly they have presented a very specific variation on a theme.However,it was George Boole(Laws of Thought,1854,chapters 16-21),John Maynard Keynes(A Treatise on Probability,1921,chapters 3, 15,17,20,and 22 ),and Theodore Hailperin,who demonstrated that the interval estimate approaches of Boole(upper-lower limits) and Keynes(indeterminate probabilities,non numerical probabilities,approximate probabilities) could be translated into a linear programming approach in 1965.Books on this topic followed in 1976,1986,and 1996..One must also mention the ambiguous probabilites of Daniel Ellsberg in his seminal 1961 article in the Quarterly Journal of Economics.Kosko's discussion of his view of an uncertainty principle could have benefited greatly from an understanding of Keynes's own approach to uncertainty,which underlies his The General Theory of Employment,Interest and Money(1936).