So, how's your web site doing? Are people reading what you blog? Downloading your software? Simple statistics are fairly easy to gather, but if the web is how you make your living, simple stats may not be enough: you might need to find out what percentage of your customers are being referred from where, or how often they wander down blind alleys in search of the Mac beta of your latest game.
If this is your current migraine, Web Site Measurement Hacks may be your new best friend. It mixes practical "do it this way" advice with cogent discussion of why you'd want to do some things some way, and other things not at all. It did jump around a little more than some of O'Reilly's other "Hacks" books, but that's partly the nature of the subject. Recommended.
Modeling the Internet and the Web sounds like it's about the same topic, but it isn't. Instead, it surveys various mathematical and statistical models of page connectivity and user behavior. There are chapters on text analysis and advanced web crawling, and lots of comparisons of theoretical predictions with actual data to inspire confidence in the modeling techniques used. If this is your thing, this is your book.
Brian Hook's Write Portable Code is also recommended. Anyone who has ported a non-trivial C or C++ application from one OS to another will have experienced some of the pain he describes. What makes this book worth owning is the breadth of its coverage, and the way the author ties various solutions together. Everything from line endings and byte ordering to stack layouts and exception handling is present and properly indexed. What's more, the writing and code samples are cleanly laid out and very readable.
Finally, there is Bruce Tate's Beyond Java. It's central messages are that (a) Java's accelerating bloat is getting in the way of leading-edge developers, (b) dynamic languages like Ruby are a lot more fun, and (c) you can never have too many kayaking anecdotes. I agree in part with the first point, though of course one geek's bloat is another's gotta-have-it. I'd agree with the second point, too, although more data and less ramble would have made it more convincing. As for the third, well, I think the analogies are forced, and make the book feel more Tony Robbins than Brian Kernighan.
Pierre Baldi, Paolo Frasconi, and Padhraic Smyth: Modeling the Internet and the Web: Probbailistic Methods and Algorithms. Wiley, 2003, 0470849061, 285 pages.
Brian Hook: Write Portable Code. No Starch Press, 2005, 1593270569, 248 pages.
Eric T. Peterson: Web Site Measurement Hacks. O'Reilly, 2005, 0596009887, 405 pages.
Bruce Tate: Beyond Java. O'Reilly, 2005, 0596100949, 185 pages.
comments powered by Disqus