Network effects occur when a product becomes more valuable as it is used more widely. Telephones are a classic example, where a single phone is useless but every phone added to the network can call every other phone and the number of possible conversations increases exponentially as phones are added. The larger the network grows, the more valuable it becomes.
The thing about network effects is that they create a winner takes all proposition. A larger network is exponentially more valuable than a smaller network, and even a modest difference in size can lead to a great difference in value. Customers buy the computer that has the most software, yet developers write software for the platform that has the most potential customers. An early lead can quickly become insurmountable, and an early monopoly can only be overcome in one of two ways.
The first way is to tap into and share the existing network. This is what Digital Research tried to do in the 1980s with DR-DOS: they created their own version of DOS that could run the same programs as Microsoft's DOS. Microsoft responded by tying its most popular applications, including Windows, to its own version of DOS so they would detect and refuse to run under Digital's version. The current owner of DR-DOS (Caldera) is suing Microsoft for anticompetitive practices because even Windows 95 can be made to run under DR-DOS instead of MS-DOS once Microsoft's "which version is this" tests are disabled.
Unfortunately, challenging a ruthless monopoly with unlimited funds and outright enthusiasm for playing dirty is not an easy way to go. (Although now that Microsoft has officially been declared a monopoly by the federal government, Caldera's own antitrust lawsuit becomes almost trivial.)
Preventing any of their competition from sharing their market is why Microsoft has gone to such trouble to tie its applications so closely to Windows. It got rid of DR-DOS when the field of battle changed from DOS to Windows. It fought dirty and it got lucky, but it still had a fight on its hands. Now it tries to prevent anyone from cloning Windows with applications that use as much unpublished, inside knowledge about the workings of Windows as possible. Microsoft's drive to leverage its way into new markets is in part about providing revenue growth, but the other half of its drive is to protect its core monopoly by destroying any software products that can easily be ported to other platforms and replacing them with products that will not run on anything but Microsoft's specific implementation of Windows.
Microsoft is very good at leveraging an existing monopoly to muscle its way into a new market. Its monopoly revenue stream provides virtually unlimited funds with which it can pay for all the development it wants. It can then wait forever to recoup those costs, selling its new products at less than the competition could possibly afford to (or that Microsoft could afford to if it didn't have other sources of income) until that competition has gone into debt and out of business. (In antitrust terms, this is called predatory pricing.)
Microsoft also has the money to buy other companies in a business it wants to be in, which can give it new products to sell when it can't successfully develop them in-house. Purchasing competitors and potential competitors is also an easy way to eliminate competition, and several of the companies and products Microsoft has purchased never see the light of day again. (For a list of Microsoft's purchases, read The Whole Microsoft Catalog.)
Finally, Microsoft has the powerful psychological weapon called FUD, Fear Uncertainty and Doubt. Invented by IBM, this marketing technique can make success a self-fulfilling prophecy by convincing customers that a competitor's product will become obsolete in the future, so customers should abandon it today to avoid finding themselves dependent on a dead-end technology later.
Microsoft often uses this approach to compensate for the fact that it's often late to market. Windows 95 and 98 each shipped two to three years after their originally announced ship date, and Windows 2000 is still in development almost three years after customers were first told to expect it. Its other products have a similarly dismal track record of shipping on time, but it doesn't matter because even though the competitor is shipping something that works just fine today, as soon as the Microsoft version comes out it will obviously take over, right? As long as nothing disrupts Microsoft's track record, it's really hard to disrupt Microsoft's track record.
The problem with all this is that it prevents Microsoft from diversifying away from its original monopoly. They haven't even gotten rid of DOS yet; Windows 98 still runs under it, as does their proposed "Millennium" product (the consumer version of Windows 2000). Since every new territory it conquers is bent back to protect the whole, everything they do is firmly tied to their core monopoly. Despite the multi-billion dollar size of the company, their future still rests on the success or failure of new versions of Windows.
Microsoft has almost completely immunized itself against the first way of overcoming a network effect: monopoly. It has prevented anyone else from sharing its market, and vigorously guards against the slightest intrusion on that front. But in doing so, it has made itself extremely vulnerable, almost brittle, to the second way of taking on such a monopoly. More about that tomorrow.