Group annotations on this page
The best way to launch web products is to first release the smallest parts that will be useful and which can stand up as a "product." Then, follow this up by watching user behavior closely and letting your users steer the product toward the real demand while adding more features. Leave your product in "beta" for a year or more if you want.
The biggest problem with the old "big release" model is that it required design and development teams to go quite far down the road of development before seeing any real-world user action, which meant that the builders had to make many more predictions about how users would behave before seeing real user behavior.
Mashups add value when two or more web apps have more meaning mixed together than the component parts did separately. If the mashup doesn't add considerable value, it's not worth doing. Unless you're doing it for fun.