Three-Toed Sloth

Books to Read While the Algae Grow in Your Fur, January 2007

Langston Hughes, A Negro Looks at Soviet Central Asia

This new edition (ISBN 9967-23-555-1) was printed for David Mikosz in Bishkek in 2006; it reprints the first edition (Leningrad: Foreign Worker's Cooperative Publishing House, 1934), with a preface by Mikosz, and end-notes giving the hand-written corrections in Hughes's personal copy, now at Yale. (These are almost all improvements.) This is by no means Hughes's best writing (even with the corrections), and it is painfully clear that the author was naive about the Soviet system. (The happy collective farmers! The fraternal solidarity of the Russian and Turkmen peoples!) But it also conveys why an American in 1934 might have wanted to believe in the Soviets --- especially a black American. Only recommended to those with a special interest in Hughes, the history of Central Asia, or, pardon the phrase, fellow travelers. §

Steven Johnson, The Ghost Map: The Story of London's Most Terrifying Epidemic --- and How It Changed Science, Cities, and the Modern World

Well-told rendition of the story of John Snow, the Broad Street pump, and the discovery that cholera is transmitted by fecal contamination of water. The distinctive feature of Johnson's version is his setting it in the larger story of the history of cities, waste disposal, and the evolution of disease, which are all more fascinating than you might expect. The last chapter, however, in which Johnson jumps from the 19th century to the 21st, has only the most tentative connection to any of the foregoing, and with a little editing could have been printed on its own in Wired or something of that ilk; for all I know it was. For some grousing about this book by a historian of medicine, see here. §

Paul Embrechts and Makoto Maejima, Selfsimilar Processes

Compact (99 pp. + references), highly selective primer on self-similar stochastic processes, their characterization, uses, and a little bit of their statistical inference. Stochastic processes are self-similar when time and space can be re-scaled in ways which leave the distribution invariant (rather leaving particular realizations invariant). They are good on connections between self-similarity and long-range dependence, not quite so good on how self-similarity relates to heavy tails. (To be fair, the latter is a more complicated topic, because, e.g., there are Gaussian self-similar processes.) The implied reader has some knowledge of measure-theoretic probability, the Wiener process, Itô integrals, and Lévy and stable distributions --- it's pitched at applied probabilists rather than, say, physicists. (Someone who took my course would know all of these prerequisites except for the stuff on stable distributions, so I need to add them.) Many proofs are abbreviated or referred to the literature in their entirety; those they do give are generally nice.

Why Oh Why Can't We Have a Better Academic Publishing System Dept.: This book consists, as I said, of about a hundred pages of text, including computer-generated figures, plus references and an index. In other words, it's only slightly more elaborate than a long review paper at arxiv.org. It is also sturdily bound, which is useful. However, it lists for over \$40, and a generous price on printing from the archive (paper, ink, bandwidth, staples, time) is 4 cents a page. I do not see an extra 30 cents per page of value being supplied by the publisher, or even the publisher and the binder together. §

Patricia A. McKillip, Solstice Wood

A sequel, of sorts, to Winter Rose, but also a rare venture into early twenty-first century America. Less ornate prose than usual, but still more Elfland than Poughkeepsie (to use an old line of Le Guin's).

Books to Read While the Algae Grow in Your Fur