Customer Reviews

The Laws of the Web: Patterns in the Ecology of Information

5 star:		(3)
4 star:		(3)
3 star:		(0)
2 star:		(0)
1 star:		(2)

 

 

The title promises much. One had hoped that with so few pages a concise outline would be the product. Alas no. One has 95 pages of vaguery, allusion to supposed meaningful research which is never explained, and trite examinations of the substantial observations that have been borowed from other authors. His reference to the power law does not result in anything applicable to understanding the Web. His reference "tragedy of the commons" a la Peter Senge, suggests he undestands neither the metaphor nor its relationship to the Web or the information that exists there. Unfortunately this takes up one of the five pages of anything containing potential substance. The discussion of nodes begins vaguely and ends with no law. Another page down. The power law suggests an upper level of tolerance, but in its lack of conclusion loses another page. Social dilemma leaves the reader with the abiding question: So? With the final page ostensibly dealing with a critical number of clicks the reader is left to infer that reading this book is too many clicks (pages turned) and with the end we are left with no code, no guidelines, no greater understanding of the growth of the Web, and appreciation that while the reader is left no wiser, at least the book was short. There is great pretension here, but no delivery.

I had high hopes for this book, but ended up being disappointed by it. First, it's very short (a hundred tiny pages), and feels more like a long survey article than a real book. A big chunk of the book is taken up by vague, trite observations about how the web is changing various things; these are familiar to everyone with the slightest interest in technology. The core of the book has very little detail. For example, one part deals with power laws in various statistics for the web. I knew the book was aimed at a non-technical audience, so I didn't expect any mathematical or scientific detail in this part of it, but I was hoping for a broad, motivated survey, that would discuss the history of power laws, where they appear and what they mean, how they were discovered here, etc. Instead, we get a couple of paragraphs that vaguely mention earthquake sizes and barely touch on the history (e.g., Zipf, Mandelbrot). Huberman does go into more detail in some aspects of power laws, like a discussion of toy models of the internet and which ones exhibit power laws (vs. log-normal distributions, say). However, I really don't see what audience he is aiming at. A technical audience could read something much more sophisticated, and I'm not sure why a non-technical audience would care about which toy models show which distributions, if the history and meaning hasn't been explained to them. Overall, I don't think most people will get much out of reading this book (although it does have the merit that it can easily be read in one sitting). It's kind of sad that some readers will probably be very pleased by Huberman's breathless descriptions of how interesting and important various things are, and will never realize that, if he had written a longer, more detailed book, they could have achieved a much deeper and more satisfying understanding. (Telling someone how great something is is never as good as showing them.)

One final comment is that much of the work in this book is Huberman's own. I tend to think he wrote it as a short advertisement for his papers. His work is indeed important, but lots of other people have done related things. Perhaps someday someone will write a broader book that puts the whole field in proper perspective.

In this book, the author introduces results of research that shows there are surprising strong global regularities present on the web resulting from the local browsing behavior of agents. He explains in simple terms and well-chosen familiar examples, key ideas to understand how these regularities come about. The ideas and the regularities described in every chapter are backed by refereed papers from the author and his associates that have appeared over the years (in Science, Nature, ...) and that I would recommend the technically inclined reader to look into. As the author takes the reader through the different chapters, he introduces in simple terms the methodology of study and analysis borrowed from the physical sciences (to study the dynamics of large number of interacting particles) which in my case it was very helpful as I am trained in computer science where we do not get exposed to those techniques. The regularities are explained by way of interesting models (e.g., social dilemmas, six-degrees of separation, Brownian motion, etc.) that make for a refreshing reading.

The author goes further than just presenting and explaining the results as he gives very practical applications where knowing these regularities can help the design of better algorithms, web sites and systems. Among some of the results presented are: a law that can predict how far users will go on clicking on pages of a given site, the existence of `internet storms' where the net becomes very slow even though there is no obvious event that caused it (like when sometimes in a highway you slowdown to a halt even though there does not seem to be any accident), a law that predicts the distribution of the sizes (in pages) of web sites and several other regularities. Among one of the clever applications described is an algorithm that figures out when to wait or request again for a web page so that the user on average downloads it faster.

Huberman deals with very large bodies of data, such as thousands of AOL logs, surveys of large audiences, aggregations of thousands of web pages, and, in addition, conducts his own experiments to determine latency and other functions of the Internet. He then analyzes his data with mathematical approachs drawn from a wide range of fields, and when necessary calls upon economics and game theory to further understand the data. He is widely published. He can also be said to write well for a lay audience, though his arguments sometimes requires repeated readings to fully understand.

These are just a few of the questions that Huberman answers (more thoroughly in some cases than others) and his approach to them is always based on a variety of useful insights. This is not an easy book, but the information and understanding to be gained from it more than repay the effort.

For a full review see Interface, Volume 3, Issue 3.

Huberman's book is a remarkably novel way of looking at the
Web. With simple and well developed examples, Huberman
provides a clear description of hidden structures existing
in Web. It's a book that is useful for both curious readers
and researchers. Readers interested in understanding the
use of information by society, including ways of searching, organizing, and interacting
with large information systems must read this book.

Although dated, this book still has useful information. It is an expose' on social aspects of the web and the author's opinion of why, how and when things occur. Also, the author discusses how to benefit.

This is a really good book and is underappreciated in the community. It is one of the first - before Barabasi - that noted the commonality of laws guiding the organization of large emergent systems.

brief book is very interesting... full of intersting obervations about human-computer interaction and information seeking via the web.