Three-Toed Sloth

Books to Read While the Algae Grow in Your Fur, September 2010

Fred Vargas, The Chalk Circle Man and This Night's Foul Work

Vastly entertaining, though the level of coincidences required for the mysteries to be solved — for the plot to go at all — is well above the regulation dose of a single million-to-one chance per month, and indeed only just below Adamsberg finding a confession in a bottle in a shark he just happens to catch and cut open.

Bérénice Geoffroy-Schneiter, Gandhara: The Memory of Afghanistan

Pretty photos of Gandharan art, with an introductory essay that emphasizes the French archaeological mission in Afghanistan. Dates from the end of 2000, with a last-minute addition about the destruction of the giant Buddhas. I doubt the essay would make much sense to anyone who didn't already know about the subject.

Arthur M. Hind, A History of Engraving and Etching: From the 15th Century to the Year 1914

Much old-fashioned art-historical information. Needs more pictures. (Did not answer my questions, but I hardly expected it to.)

John Layman and Rob Guillory, Chew: International Flavor

Mind-candy, you should forgive the expression. (Previously, subsequently.)

Lauren Willig, The Deception of the Emerald Ring, The Seduction of the Crimson Rose and The Temptation of the Night Jasmine

Mind-candy; nearly weightless (despite the evident historical research!), but highly enjoyable while they last.

Karin Slaughter, Broken

Compelling, even when some view-point character's consciousness are dark pits I wish I could crawl out of; Hell, they wish they could crawl out of them. (Deliberately vague to avoid spoilers.) — Previously; sequel.

Steffen L. Lauritzen, Extremal Families and Systems of Sufficient Statistics

A fascinating look at what can be done by starting with postulating certain sufficient statistics, and distributions of observables conditional on them, and then building the model class from these. In particular, a very special role is played by "extremal" distributions, which can be interpreted as ones which cannot be obtained as convex mixtures of other distributions in the same family, or, what turns out to be equivalent, the models where all the parameters are identified in the limit. Particularly nice results hold for models where the sufficient statistics take values in a semi-group, including powerful extension of the usual results about exponential families. All in an all, it's an excellent book with some rather profound statistical theory, but it's horrible to read math written on a typewriter. Someone really needs to re-set it in LaTeX and maybe put it on the arxiv.

Lauritzen's "Extreme Point Models in Statistics", Scandinavian Journal of Statistics 11 (1984): 65--91 (with discussions and reply) is selected highlights of the book, without proofs, details and extensions, but with decent typography. It's available via JSTOR.

Books to Read While the Algae Grow in Your Fur; The Pleasures of Detection; Writing for Antiquity; Enigmas of Chance; Afghanistan and Central Asia; Scientifiction and Fantastica