Commentaires client les plus utiles
|
|
10 internautes sur 11 ont trouvé ce commentaire utile :
3.0étoiles sur 5
Reduction to nodes and links, Avril 28 2004
Albert Barabasi presents the lay reader with a stimulating description of the origins of network theory and recent applications. He describes random networks, small world and scalefree networks. In nonrandom networks the importance of hubs is emphasized. Small world networks are the ones with a well defined averge number of links, and in scalefree ones the density of links scales as a power law. For the many interesting examples discussed, I would like to have seen graphs showing scaling over at least three decades in order to be convinced of scaling. However, in practice, whether a network scales or not may not be so important. I liked best the discussions of terrorism, AIDS, and biology. If one could locate the hubs, then a small world network could be destroyed, but as the author points out there is no systematic method for locating the hubs. Also, destroyed hubs in a terror network might be replaced rather fast, whereas airline hubs could not be replaced so quickly. The book might be seen as indicating a starting point to try to develop a branch of mathematical sociology. For example, the maintainance of ethnic identity outside the Heimat is discussed in terms of networking. Now for a little criticism.I did not find the discussion of ‚the rich get richer' very helpful because network theory at this stage deals only with static geometry, not with empirically-based dynamics. In fact, the dynamics of financial markets have been described empirically accurately without using any notion of networking. In the text the phrase „economic stability" is used but stability is a dynamic idea, and there is no known empirical evidence from the analysis of real markets for any kind of stability. The absence of dynamics on networks means that complexity is not described at all: there is nothing complex about the geometry of a static network! Suggesting that cell biology can be described by networking is empty so long as dynamics are not deduced from empirics. Nonempirical models of dynamics will probably not be of much use for making advances in understanding or treating cancer, e.g. Everything we know about cell biology and cancer was discovered via reductionism, by isolating cause and effect the way that a good auto mechanic does in order to repair a car. Unfortunately, the author lets his enthusiasm get the best of him when he proclaims „laws of self-organization" and the need to go beyond reductionism. First, there are no known laws of „self-organization". The only known laws of nature are the laws of physics and consequences deduced from the laws, namely, chemistry and cell biology. Worse, every mathematical model that can be written down is a form of reductionism. Quantum theory reduces phenomena to (explains phenomena via) atoms and molecules. All of chemistry is about that. Cell biology attempts to reduce observed phenomena to DNA, proteins, and cells. Believers in self-organized criticality try to reduce the important features of nature to the equivalent of sandpiles. Network enthusiasts hope to reduce phenomena to nodes and links. In order to try to isolate cause and effect, there is no escape from reductionism of one form or another, holism being an empty illusion. So I did not at all like the assertion on pg. 200 that globalization (via deregulation and privatization) is inevitable, because there is no law that tells us that it is. Summarizng: there is no complexity without dynamics, there are no known „laws of self-organization", and reductionism is the only hope for doing science. Anyone who disagrees with this is welcome to explain to me and others the alternative (jmccauley@uh.edu).
Aidez d'autres clients à trouver les commentaires les plus utiles
Ce commentaire vous a-t-il été utile ?
 
|
|
|
|
|
|
5 internautes sur 5 ont trouvé ce commentaire utile :
5.0étoiles sur 5
Book's Audience: Who should be linked to this book., Juil 23 2003
I have focused this review on the audience of the book, since other reviews have quite adequately summarized the material.There have been a lot of books recently that have been published on the new science of networks. Network theory and how it applies to many different fields from technology, marketing, biology, social science, terrorism, disease control etc. (Six Degrees by Duncan Watts, Nexus - Mark Buchanan, Smart Mobs - Howard Rheingold, Tipping Point - Malcolm Gladwell etc..). Barabasi's is a welcome addition to the field and has a nice niche, which isn't filled by the other books. As some other reviewers have pointed this book is a popular science book, which means it covers scientific and mathematical theories at a very high level and makes these theories accessible to a wide audience. The niche lies somewhere between Gladwell's Tipping Point and Watt's Six Degrees. It is very well written and draws you in with stories that explore the theories. Some of the other reviewers have complained that Barabasi has done a disservice to the theories that he explains by making them too simplistic. I disagree, I actually found this book to be very rewarding, and a quick read, which is a sign of a well-written book. I have never been a fan of scientific and academic books that pride themselves on being totally incomprehensible. Richard Feynman, the Nobel Prize winning physicist, once said that if someone truly understands a subject they should be able to explain it to a general audience without resorting to technical jargon (Feynman's Lectures on Physics Vol 1,2,3 are a perfect example). To be able to explain a complex subject you need to resort analogies, examples and stories. Stories give a framework for the general reader to absorb the complex material. Barabasi has managed to explain the science of networks using all three. I am not sure how this can be seen as a bad thing. This exposes a wider audience to a very interesting subject; this has to be good thing. Summary:
Anybody who loved Gladwell's Tipping Point and was looking for a book that explains some of the theories behind the phenomena will love this book. It's a little bit more technical than Gladwell's book, but it is well written and it will appeal to a wide audience. As popular science books go, this is definitely on par with Ed Regis's Nano and Steven Levy's Artificial Life, but not quite at the level of Gleick's Chaos. If you are looking for a technical book, you should look at Duncan Watt's Six Degrees, or Small Worlds.
Aidez d'autres clients à trouver les commentaires les plus utiles
Ce commentaire vous a-t-il été utile ?
|
|
|
|
|
|
2 internautes sur 2 ont trouvé ce commentaire utile :
2.0étoiles sur 5
A report from a confusing area of research, Déc 7 2002
If you read this, you will read about Barabasi's exciting work and the work of his friends. You will read about the risks he and his colleagues take with their careers. You will read about the incredible inertia in academia. But, you won't find much insight into the principles of network dynamics? I'm not sure the book delivers. We get a 'report from the field', but not much detail or general understanding. It's all too confusing and new, if I caught Barabasi's drift. But, is this a good 'introduction' to network dynamics? Based on the reviews here, it seems clear the prose appeals to many readers. If this inspires people to read more, then great. I am afraid they are attracted by the comforting tone and soothing outlook, though. We get too much of Barabasi, the expert grant writer. Barabasi foresees network dynamics leading us to Kurzweil's happy 'Age of Spiritual Machines'. A more down to earth view suggests networks bring us Osama Bin Laden. Barabasi is quite thrilled to find small world dynamics in his network research, but never connects them to the 'small world dynamics' of drug lords and suicide bombers. I'm a bit puzzled by Barabasi's problems with the details. For example, he does a poor job of explaining exactly what a 'power-law' distribution might be, though he uses the term over and over, again. How does one 'find' a power-law in experimental data? Most people have probably gone through much of their lives never seeing a single one! If you find one, will anyone agree with you? Offering a few examples that one could work with at home would go a long way. For instance, Barabasi talks about the way wealth approximates a power-law distribution. If you try to work with published data on this subjects, there won't be much that looks like a power-law. In fact, the whole idea is rather controversial. It confounds our intuitions and sense of what is right. A power-law distribution of wealth has a few rich, a few more at the 'middle income' level and huge masses in the 'poor' bracket. We would rather have income distributed according to a 'bell curve', a few rich, a few poor and most 'middle class.' If you want to claim 1) natural is 'good', 2) power-laws are 'natural', and 3) wealth has a power-law distribution, why complain about a vanishing middle class? A big middle class is unnatural! These and other conundrums of the network await the reader's next journey into the subject matter.
Aidez d'autres clients à trouver les commentaires les plus utiles
Ce commentaire vous a-t-il été utile ?
|
|
|
|
|
|
Commentaires client les plus récents
|
de Duncan Watts
de Lorenz
de Mark Buchanan
de Steven H Strogatz
