The network effect is the impetus behind today's software platforms, but a balance must be struck between homogeneous vulnerability and fractured inefficiency. Comparing J2EE to .NET shows clear advantages for J2EE through vendor diversity, portability, standardization community, educational opportunity, language commonality, and security. .NET's attempt to replicate J2EE is shallow, providing technological similarity in a disconnected and proprietary package.
Broadly speaking, the network effect is the growth experienced by networks due to the feedback loop induced by the increasing value of joining a growing network. Consider fax technology. It has been successful because the network of fax machines, connected by the telephone system, communicates reliably - thanks to a common standard. The adoption of fax machines showed runaway growth because, in a sense, the value of a fax machine increased as the size of the entire fax network increased. 
Such networks may start slowly, but when they do succeed the effect is dramatic. Clearly, interoperability is crucial to their success. Why not guarantee interoperability by insisting on a single manufacturer, a fax machine monopoly?
The reasons against this have been repeatedly established: an absence of competition leads to lower quality, higher prices, a lack of innovation, and a vulnerable system. The free market demands a diversity of suppliers, unified through common standards. This is not only an economic principle; looking at software platforms, we see the network effect on many levels. Let's examine how J2EE and .NET compare.
Everything derives from the human network of education, and both J2EE and .NET technologies are excellent starting points for an education in computer science. The difference is that only J2EE is suitable for a formal curriculum. Unlike the .NET unmaintained prototype in "Shared Source," J2EE code is freely available in its entirety for educational and research purposes. Academic integrity is preserved only when full examination and discourse are encouraged.
Isn't .NET better for teaching? With its Common Language Infrastructure, it can be used to teach any language. However, this is a dangerous illusion. .NET reduces the interesting differences of programming languages to a syntactic tower of Babel. Exposure to a variety of languages is an essential part of every education, but they must be seen in their true form to be of value. In contrast, Java makes no such claim of universality; it simply unites developers with a common language, allowing them to effectively share source code and ideas. The network of developers is connected through Java, not through compiled bytecode.
Less abstract is the network of middleware, virtual machine, and operating system suppliers. Once again, the free market demands that this be a diverse collection. J2EE, through its focus on portability, provides an integration point that gives developers and deployers choices at every stage. In opposition, .NET holds up a single supplier, eager to collect high taxes and exercise control.
