Chaos by James Gleick

An August 1999 Wired magazine profile of James Gleick, popular science writer and former New York Times editor/columnist/reporter, describes the success of Chaos: Making a New Science (Chaos: MaNS) as playing a pivotal role in his life, earning him a seven-figure, two-book advance and his freedom as a writer (James Gleick's Survival Lessons). Chaos: MaNS spent twenty weeks on the NY Times Bestseller List and was popular enough that, in 1990, software maker Autodesk enlisted Gleick to work with three of its developers to produce a suite of DOS apps that generate visualizations of phenomena discussed in the book. Its apparent commercial and critical success notwithstanding, Chaos: MaNS is not a particularly good book.

Inside the book

Bad charts, worse explanations #

One of my biggest gripes is the complete lack of axis labels on graphs included in the book. Sometimes, if you're lucky, there are hints as to what each axis represents in either the captions, a brief aside, or the main text. In other cases, you can go to Wikipedia or Mathworld and search on the name of a researcher connected with the work associated with the graph or some other significant-seeming phrase from the caption or accompanying text and find a normal, labelled version of the same chart. Sometimes, however, you're simply out of luck. As an example, here's the first graphic you'll find in the book (p17):

That image is included in the first chapter and is meant to illustrate that the results of two different runs of Lorenz's primitive climate model varied increasingly as the runs progressed even though the initial conditions were the same. Here is Gleick's prose description of the significance of each of those lines:

So do the lines represent the latitude of a parcel of air? The latitude of some discrete point along the "airstream"? Pressure? What units? Presumably, the x axis represents time or a sequence of discrete calculation steps. It might even be ocean surface temperature since Gleick throws out the following line a little bit later: If a weather satellite can read ocean-surface temperature to within one part in a thousand, its operators consider themselves lucky. Wait a minute ... one part in a thousand? What does that even mean? Temperatures are measured in degrees (C/F) or Kelvin. Is he talking about a thousandth of a degree or a thousandth of a K or what?

The excerpt above is typical of the book. Insofar as the prose flows smoothly, it's a (mostly) well-written series of vignettes focusing on different threads in the development of chaos theory. Gleick tries to present some of the science and mathematics involved, but he's stripped away too many of the details for most of it to mean very much. Here are the images meant to illustrate the Lorenz attractor and the caption from the facing page (p28-29):

Would it have killed Gleick to include the 3 equations (simplified versions of the Navier-Stokes equations, which can be found online) that Lorentz was using to generate his model climate in either a box at the bottom of the page bearing the illustrations or in the caption - or anywhere else in the entire book? Images plus a garbled explanation plus a few equations wouldn't have been less meaningful than the images and garbled explanation alone. The fact that Chaos: MaNS did include some other equations (like the equation for a parabola) makes the absence of the Lorenz equations seem completely arbitrary.

Name that scientist #

Gleick's way of introducing researchers who figure prominently in later chapters can also be a trifle annoying. An example:

From the above, a reader might infer that, since the astronomer and engineer were being mentioned in passing, they were of little significance and likely wouldn't be mentioned again. Later on, though, the French astronomer studying galactic orbits pops up again is quite an important figure within the book. It's Michel Hénon, the discoverer of the Hénon attractor (see chapter 5). That unnamed Japanese engineer? It's none other than Yoshisuke Ueda. He resurfaces in the same chapter as Hénon. This happens again and again. The nameless population biologist at Princeton mentioned at one point is later revealed to be Robert May. Benoit Mandelbrot is introduced as a geometer at IBM and Libchaber as a French mathematical physicist.

Bad terminology #

There are a number of places where the definition or context for a new term lags after its initial appearance by several pages (e.g. "fractal" is first used in an image caption on p95 but not defined until 3 pages later). Gleick also seems to use the wrong terminology for a given phenomenon or concept on more than a few occasions. Below, for example, he captioned a graphic showing the process by which the Cantor set is created (repeatedly removing the middle thirds of a set of line segments) as the Cantor dust.

The explanation accompanying the caption blurs the distinction between the Cantor dust and Cantor set. The Cantor dust Wikipedia article has good visualizations of the Cantor dust in 2d and 3d.

Perspective shifts and disjoint narratives #

In places, the book really does read like an overly-long newspaper article. Quotes are sprinkled in here and there without much information on when or to whom they were given. The note at the back of the book for the following quote merely attributes it to Mandelbrot, but that was already fairly clear. When did Mandelbrot say this and to whom? Is it taken from one of his books or papers? Gleick never tells us.

When Gleick jumps between narrating the stories and first-person quote dumping, it can be a bit jarring. The last section of Chapter 6, Universality, seems to consist primarily of the better part of an interview with Mitchell Feigenbaum (conducted when? where?) dumped unceremoniously into the book as a series of quotes (about nine paragraphs-worth in all).

The storytelling in the book also simply falls apart in places and, earlier in the same chapter, we can find a prime example of the sort of disjointed narrative that plagues Chaos: MaNS. Beginning on p159, Gleick meanders from a biographical sketch of a young Feigenbaum to a description of the challenge Peter Ambler Carruthers (once head of the theoretical division at Los Alamos), introduced here only as "Carruthers", faced in managing physicist Kenneth Wilson and then to the work of other physicists (Leo Kadanoff and Michael Fisher) ... in the space of just two pages with no perceptible transitions. There may be connections between these passages but Gleick never makes them. It's as though he was rushing to make a deadline and had intended to come back to this section to fix it up but forgot.

Gleick: "I never liked the word 'popularizations.'" #

Chaos: MaNS was not an easy book to finish. As I worked my way through its pages, the urge to chuck it into the trash gripped me numerous times (whenever I encountered yet another graph without any axis labels, for instance), but I'm glad that I resisted acting on those impulses. The book introduced me, indirectly, to some elements of chaos (aka nonlinear dynamical systems) theory. I didn't pick up very much from the book itself but googling to figure out what Gleick was trying to describe led me to better explanations of the same topics.

While I was cataloging the flaws in the book, a question dawned on me. What had Gleick set out to accomplish when he began writing Chaos: MaNS? Perhaps, I thought to myself, he had succeeded by his own standards. Thanks to Google, I stumbled across an paper by Danette Paul, Spreading Chaos, that includes several quotes from interviews conducted with Gleick in 1997. I coudn't find any place where he explicitly laid out his goals for the book, but there were a few passages that seemed relevant:

He's saying that doesn't think of Chaos: MaNS as science popularization but, rather, some sort of quasi-historical-slash-news-story depiction of the development of chaos theory. I suppose that as such, from his point of view, he would have been under no obligation to get the science and math bits lucid or even correct.

Unfortunately, that doesn't really hold up. If he hadn't intended to write a popularization of chaos theory but rather just a narrative of the human side of the development of the field, he could have omitted the attempts at explanation and all of the graphs altogether (but kept the pretty color insert in the middle of the book) and just restricted himself to the he-said-she-said parts. That's not the route he took. He did try to explain a wide array of phenomena and did include numerous charts - he just managed to write muddled explanations and abuse the charts he chose to include in such a way that someone who didn't actually look at the graphics too closely or think very much about what they'd read might come away with the impression that they'd learned a great deal when in fact they'd learned very little indeed.

See also

• Chapter summaries
• Computers mentioned in the book

Notes

1. If you search the Pulitzer Prize Board's website on Gleick's name, you'll find Chaos: MaNS listed as a 1988 Pulitzer Prize finalist, along with two of his other books (both biographies): Genius: The Life and Science of Richard Feynman (1993) and Isaac Newton (2004). In a sidebar on his web site, around.com, Gleick lists all three books as national bestsellers. I'm not a Times Select subscriber, so I can't access any but the most recent New York Times Best Seller List and the lists available online at hawes.com seem to be restricted to hardcovers (judging by the prices shown), so I have no way of verifying their bestseller status or checking to see how long they stayed on the lists or the highest ranking that they attained. I emailed hawes.com to ask specificaly about Chaos: MaNS and Arlene Hawes Petersen very graciously replied with the following information:

   Chaos first appeared at #11 on 1/24/88. It stayed on the list for 20 weeks, last appearing on 6/12/88. The highest it ranked during that time was #7 on 2/14/88, 3/27/88 and 4/3/88.

   Arlene Hawes Petersen, email, September 14th, 2006

2. In 1992, Autodesk handed the copyright for the software back to Gleick. Subsequently, he and the programmers involved in its development made it available for download (source available as well). The programs run (and the results look pretty neat) under Windows XP. Interestingly, one of the Autodesk team members (the one currently hosting the downloads) was none other than mathematician/author/futurist/hacker Rudy Rucker.
3. I highly recommend Danette Paul's Spreading Chaos to anyone interested in a thoughtful and well-researched look at science popularizations. I've only used some of the quotes from her interviews with Gleick here, but her paper is well worth reading on its own merits.
4. I bought Chaos: MaNS over the counter here in Hong Kong and so got the UK edition (Vintage, ISBN: 0749386061). To my eye, the UK edition cover is more attractive than the US edition's, which is why I'm displaying it here. The Amazon link is to the US version, however, since most of you reading this would likely be ordering from Amazon.com. Don't freak out.
5. Better books covering some of the same territory (more rigorously):