What are the pseudosciences of today? What will folks look at a century hence and ask "How could they have taken that seriously?" Three perhaps-related candidates that have crossed and recrossed my path:

• Social Network Analysis — which attempts to discover relationships based on fragmentary clues of interpersonal connections via graph theory and statistics
• Criminal Profiling — which purports to identify perpetrators of various heinous acts based on apparent patterns among where/when/how they have struck their victims
• Text Mining — which claims to take large natural-language corpori and automagically organize them based on co-occurrences of words and character patterns

Common features of these disciplines: big headlines, simple explanations of complex phenonema, and a near-invulnerability to quantitative testing. Where are the real double-blind experiments, the validated test cases? Instead, all one sees are just-so stories.

I hope that I'm wrong and there really is a pony in the barn ... but meanwhile, I'm awaiting solid evidence. Which reminds me of a med-school aphorism:

(see also ScienceVersusStampCollecting (20 Jun 2000), LogicAndInformation (1 Aug 2001), ... )

TopicScience - 2003-10-19

(correlates: YouCanHaveItAll, FinalLesson, InvisibleWeb, ...)