Hanson describes mashups in the context of the so-called Web 3.0. Three types of mashup codings are given - presentation, data oriented and process oriented. Of these, the presentation approach is the simplest to understand and code, but also the most limited. It involves the mashup happening directly in the browser, when it loads a web page of mashup instructions. A big drawback is the browser sandbox. So if you load the page from Alpha dot com, then it can only load data from that domain.
The other approaches constitute the bulk of the book. Much harder. By the way, the text also gives a usage for JMX [Java Management Extensions]. About 8 years ago, JMX was hot, as a great new thing to control remote java code and access remote data feeds. Hanson in fact wrote a book on JMX. Unfortunately, JMX fell into some abeyance as too limited for a difficult problem, and was overshadowed in part by SOA and Web Services. Now the current book shows how JMX can be applied in constructing a mashup. In essence, a mashup can be considered part of what JMX was originally intended to do, though the word mashup in its current usage did not then exist.
The book also has a list of reputable organisations that provide data feeds for you to experiment with. (Though you often need to register and perhaps pay them first.) These include Google, US post office, NOAA and AOL.