Taking on Microsoft

It's time to rewrite the rules of the game. Here are some ideas that don't depend on the Justice Department.

By Karen Linn

I'm no fan of regulation. I think it impairs progress, letting the players figure out how to fight today's battles while the rest of the world marches on to the next competitive grounds. If the software market as a whole, and the Unix vendors in particular, want to take on Microsoft, it's time to re-invent the rules of the game. Here are some subtle but crucial shifts in the industry that should force us in that direction.

Operating systems are becoming more cost- than value-add. When was the last time you thought, "That application runs better because of POSIX compatibility"? If a vendor suddenly introduced an operating-system technology that fundamentally changed the way we code applications, or reduced the cost of managing a distributed network, the cost-value pendulum would swing back again. The gauntlet has been laid down: It's time for the Unix campers to deliver the next wave. There are several signs of progress: OpenStep (Sun's distribution of the Spring research OS) and most interestingly, the Linux movement. It's time to move the operating system wars to the next level -- application interoperability.
Marketing and software distribution benefit from the flat namespace of the Internet and the Web. Microsoft.com and Little-guy.com are just an ftp -- the same ftp -- away. If you have a good product, it's easy to get it into the hands of millions. Fighting over shelf space at retail outlets is reliving a 1980s model of software distribution.
Modularity is reducing the cost of entry for those new products. Objects and OMG standards, OpenStep (again), STREAMS interfaces, and even the interest in microkernel architectures -- all are forcing us to build software that is self-contained and well behaved when placed in the same box with other code. Modularity is possible with well-defined interfaces between components. As long as there are strong, sufficient, broadly defined standards, there will be competition for their implementation. If you only have to do a small part of the whole solution to deliver value, you can get in more cheaply. Why build a Microsoft Word competitor when your brilliant idea is really for a table typesetter? Use OLE and just do tables. If you fear Microsoft entering the transaction market, think about where you could provide the same transaction vehicle more reliably or less expensively. Increasing margin pressures should force competition on price; pressures on timeliness and correctness force competition on features and performance.
Modularity wins when the components are well integrated. Modularity buys you a measure of risk diversification -- if you don't like Vendor A, you can buy from Vendor B, and the plugs will be about the same shape -- until you find out that your risk is diversified over zero products. (Know anyone who standardized on DME?) IBM provided complete interoperability in the mainframe data center. Microsoft provides complete interoperability on the desktop. The Unix vendors need to move from diversification to syndication -- spread the risk over several possible players, but also take the lead in bringing them together and guaranteeing that they will work together.

Regulation scares me because it sets a bad precedent. Does the government have a place in regulating the best Internet providers or the best cross-vendor interoperability tools? Unix came into the mainstream market through technology leadership: It lowered costs and provided flexibility. The Unix industry needs more of that leadership to compete with Microsoft.

About the author
Karen Linn is the pseudonym of a Unix professional working for a computer vendor.

If you have problems with this magazine, contact webmaster@sunworld.com
URL: http://www.sunworld.com/asm-05-1995/asm-05-commentary.html.
Last updated: 1 May 1995.