At every point in the development of fixed line Internet, the underlying infrastructure has been non-prescriptive. The designers of the infrastructre made no attempt to figure out the best uses for it: what they did was make it as flexible as possible so that individual users and software designers could figure this out for themselves. The designers of the underlying TCP/IP protocols in the 1970s and 1980s had absolutely no way of imagining the WorldWideWeb as it would come to exist in the 1990s.
The successful strategy was to design an underlying network that worked, and then allow consumers and the market to decide what should be done with it. The TCP/IP protocols were thus designed with two main features in mind. First, they were made as hardware independent as possible. Once the data was split up into packets, these packets could be sent over any other network running TCP/IP, and TCP/IP was designed to work on virtually any underlying hardware. Secondly, they were made as flexible as possible, so that they could be used for virtually any use that involved two computers sending information to each other. If I wished (or wish) to design a new aplication and application protocol to support online chat over the Internet, or to design a new method of online file sharing, I do not have to get permission from anyone to do this. I just write the software. This is exactly how e-mail came to be invented in 1973, and how the WorldWideWeb came to be invented in 1991. Over the years, this fact has been incredibly powerful. The wider the net becomes, the more powerful this flexibility becomes, as people can share their new software and new ideas almost instantly.
-- [Michael Jennings] et al. WAP or i-mode, or Slowly Merging--Lessons from the fixed-line Internet and from Japan. Credit Suisse - First Boston, 14 November 2000.