Interesting article in Wired News: “Covert Crawler Descends on Web”.
The smart folks at SPI Dynamics have identified a crawler that wanders the web, looking like a human: it works from a “real” browser (one that caches and only downloads new material), works at “human” speeds and varys it's behavior by content or site. This sort of activity is less likely to discover in web server logs and with a large enough zombie machine base could participate in a variety of illegiitimate activites.