The goal of this book is to provide a snapshot of current and future research being conducted into HCI on the Internet. Twenty- three experts in HCI-related fields contributed the 16 chapters of this book. Well-known names like Ed Chi (Xerox PARC), Mary Czerwinski (Microsoft Research), and Andrew Sears (UMBC) all wrote about their research into how we interact with the web.

The book is thought-provoking in the breadth and depth of knowledge presented. The book moves from "digital strategy" and planning, through analysis, to design and evaluation. Highlights from the first two parts include Shneiderman's "universal usability," where web sites are designed for a world-wide audience (presented ably by Andrew Sears), cost-justifying web usability (Mayhew and Bias), usability metrics (Opaluch), and "phone usability" where usability surveys are conducted at low cost over the phone (Ratner). Talk about discount usability engineering!

The third part, comprising more than half of the book, explores the cognitive aspects of web usability. Microsoft's Czerwinski and Larson offer a glimpse into HCI research at Microsoft. They propose that a stronger cognitive foundation must be achieved before we can truly optimize the design and usability of web sites. In their "Cognition and the Web" chapter they highlight areas where works needs to be done, and show some interesting examples from their research.

Czerwinski and Larson offer a new metric to better fit cognitive science to web design. They propose that Relative Subjective Duration (RSD) is a better measure for gauging user satisfaction. RSD measures the difficulty users are having with a task by time estimates. Users who are interrupted tend to overestimate task duration while completed task times tend to be underestimated. They also offer a glimpse into Microsoft's research with sections on nonspeech audio to enhance the web interface, telephony, and speech recognition.

The "Scent of the Web" chapter is especially relevant to web designers as Ed Chi (Xerox PARC) offers an automated way to analyze the usability of sites before users see any pages. Not unlike an animal's foraging behavior users use "information scent" to optimize their efforts to find what they want. Information scent is the perceived value and cost of assessing a piece of information.

Chi has developed a tool to called Bloodhound to automatically sniff out a site's usability based on a predictive model suggested by information scent and information foraging theory. As we scan and skim web pages we use information cues like link text and surrounding text, graphics, position, and frequency of occurrence. Chi has mathematically codified these cues into a "scent matrix" that predicts the probability that a user with a particular information need will move from page A to page B. Using this matrix Chi's tool can predict paths where users will go with particular information needs, and how effective particular sites are for particular sets of keywords.

Their Bloodhound tool can accurately predict where users will surf within sites, how usable sites are for particular information needs, and even summarize sites by keywords. Using simulated users (each with a set of a keywords representing their information need) their Bloodhound service can automatically return metrics that rate web site usability.

Chi says that a scientific understanding of human behavior in web surfing is essential to web usability. Innovative tools like Bloodhound can lower usability testing costs and improve our understanding of human behavior on the web.

The book essentially says that heuristic rules of thumb and best practices can only take you so far in usability engineering. For real progress we need a stronger cognitive foundation on which to improve the usability and design of web sites. Guidelines may work in isolation, but on the web there are so many variables that applying these guidelines to all sites for all users doesn't always work.

We require multidimensional analysis to assess the best ways to design actual sites. We need to study more complex, real-world problem domains that better match what users actually experience. This book offers a fascinating glimpse into current and future HCI research on the web.