First: why is a monopoly "bad" for society? Monopolies have historically stifled innovation within their industries, causing consumers to pay higher prices for inferior products. The counter-innovative tendency occurs both internally and externally to the company. The holders of a monopoly have little reason to innovate because their profits will not be noticeably improved by new methodologies and products. Those external to the monopoly have little incentive to invent new methodologies or products because they would not have an opportunity to exploit and gain from their new ideas in the environment of a monopoly.
What solutions to the Microsoft monopoly have been proposed? The simplest solution is, of course, to do nothing. I find this unacceptable due to the high cost to the consumer of inferior, shoddy products.
The next most obvious approach is heavy government regulation. Because the reaction time of government regulators is far slower than that of the computer industry, this method would likely be an abysmal failure. Either the government regulation will not be strong enough and Microsoft will easily be able to work around the regulations, or it will be too heavy, quashing any desire for invention which does exist in the company.
Another proposed solution is dividing the monopoly up based on applications - for example: MSOS, MSOFFICE, and MSGAMES. But MSOS and MSOFFICE will each still have effective monopolies within operating systems and office applications. The division by workgroup solution may offer some short term improvement, but it is important to remember how Microsoft achieved its monopoly: by leveraging its widely used operating system. MSOS will be able to use exactly the same approach yet again, swallowing MSOFFICE bit by bit, just as Corel was locked out of the office application market and Internet Explorer was assimilated into the operating system to defeat Netscape.
Luckily, there is something very unique about the Microsoft monopoly which can be exploited to produce a remedy. In monopolies such as steel, oil, train, or telephone, the company had control of physical facilities which were difficult to cleave into separate viable pieces. When these previous monopolies were divided, partitions were typically made by region or manufacturing process. Microsoft's monopoly, on the other hand, is almost entirely based on _bit patterns_, which are trivial to reproduce. The reason these bit patterns haven't yet been copied by others in significant quantity is software copyright, patent, and trade secret. In fact, the Microsoft monopoly is a government encouraged monopoly - given the way copyright laws are applied in the industry and the leverage of having the most common operating system, a large software monopoly is almost inevitable.
One obvious solution to the Microsoft monopoly is then to remove or reduce software copyright laws. A meaningful reduction of the time scale of software copyright would have to push the length of a software copyright down to something akin to the typical time scale in the computer industry - 1 1/2 or 2 years. For various reasons including the strength of corporate influence in government and American dominance of the international software trade, I seriously doubt that the government will be willing to implement this solution at this time.
Nonetheless, there is a way to exploit the virtuality of Microsoft's monopoly to produce a meaningful, lasting solution. Essentialy, we create 4 little Microsofts: MSUP, MSDOWN, MSLEFT, and MSRIGHT. Each MS* would have its own copy of all source code, software, trademarks, and other intellectual property, along with the right to license various uses of the property. There are a few problems with this method, but none of them are insurmountable. Microsoft does own some physical objects - buildings and employees which could be divided in more conventional ways. For stocks, it is important that the current stock holders do not own the same proportion in each MS* to avoid monopolistic tendencies. I would suggest a maximally dichotomous stock split - each stock owner recieves stock in one of the MS* companies.
Why is this a good remedy? Each MS* would have a fair division of resources and no significant incentives for cooperation, implying a strong potential for rich competition with compelling innovation incentives in the forseeable future. It should take a long time (perhaps 15 years?) to break the symmetry of the initial division of the 4 MS*'s, allowing one company to return to a monopoly position.