BUSINESSWEEK ONLINE:   Business Week ebiz


Business Week e.biz

PERSPECTIVE By Mike France February 14, 2000


What Would It Cost to Split Microsoft?
Whether competing Baby Bills would fragment Windows and cost software developers billions is becoming one of the case's hotly contested questions

Should Microsoft Corp. be split up into a series of identical, vertically integrated Baby Bills? That's one of the main breakup scenarios for the software company, and now that government remedies are seemingly leaning toward this solution, the question is getting more serious consideration than ever before. To a large extent, the answer depends on one critical issue: whether this type of vertical division of Microsoft would fragment the Windows operating system.

Surprisingly little serious attention has been focused on this issue. In April, Stan J. Liebowitz, an economics professor at the University of Texas at Dallas and a longtime critic of the Justice Dept. antitrust suit against Microsoft, wrote a study in which he estimated that a vertical breakup would increase costs for independent software developers by about $10 billion per year. Last month, a trio of economists, financed by Oracle Corp., attacked Liebowitz' thesis, arguing that the costs of a vertical breakup would be negligible -- and that there was little danger of operating-system standards being splintered. The group includes Georgetown University's Steven C. Salop, as well as R. Craig Romaine and Robert J. Levinson of Charles River Associates.

Who has the better argument? In my opinion, the Oracle group. But their study is far from comprehensive -- and this issue certainly deserves much closer attention if vertical breakup later comes under serious consideration.

"A SERIOUS DRAG." Liebowitz' argument is based on one key assumption: Because the Baby Bills would sell competing operating systems, they would have no choice but to differentiate their offerings. As a result, he argues, the uniform Windows operating system would inevitably be fragmented. Once that happened, software-application developers would have to hire new programmers, testing personnel, and product-support staff. Of 11 companies that responded to Liebowitz' request for estimates on the incremental costs of developing programs for additional operating systems, the average estimate was that the research-and-development costs would increase by 78% with each new Windows platform, and that support costs would rise 47%.

Several software developers buttress Liebowitz' conclusions with firsthand testimony about the difficulty of supporting multiple operating systems. Says Charles Chrystie, CEO of Chili!Soft, a Bellevue (Wash.) maker of Web applications: "Maintaining software for different operating systems is very costly and is a serious drag on our business. To begin with, it takes about six months for a team of 12 people to port a new operating system. The additional cost of each OS is about 80% of the original work.... [I]nstead of improving our software for our consumers, we'll be busy moving to a new OS." Based on this data, Liebowitz "conservatively" concludes that developers' costs will increase by 6.46% of revenues, or about $11.9 billion industrywide in 2002. If he's right, that could be a powerful argument against a vertical breakup of Microsoft.

But the Oracle group roughs up his analysis quite a bit. For starters, Salop and his colleagues argue that the Baby Bills would have a strong incentive to maintain compatible standards. Because of the enormous installed base of Windows applications, consumers would be unlikely to switch to an incompatible new system, they argue: "If one of [the Baby Bills] drastically deviated from the Windows standard, it would risk financial ruin because consumers might not be willing to suffer the costs of transition. In contrast, a [Baby Bill] that adheres to prior standards would stand to gain additional sales."

 


The Oracle report says the costs of getting software to work with new Baby Bill operating systems has been much overstated
 

They also contend that Liebowitz dramatically overstates the costs of developing software that would port from one operating system to another. Many of his estimates, they note, are based on data supplied by companies that make software for both Microsoft and Apple operating systems, which "differ greatly in their design and...run on radically different hardware." But because all of the Baby Bills would be starting from the same code base and hardware platform, any necessary code rewriting would be cheaper. They add that the companies also would have a strong incentive to join a standards-setting body. According to the Jan. 20 Oracle study (called "The Flawed Fragmentation Critique of Structural Remedies in the Microsoft Case"), "Professor Liebowitz...assumes that the new operating system competitors will opt out of the Windows standard, forcing consumers to jettison their existing investments in applications software and coercing applications developers to rewrite their products. In reality, the costs of switching and the benefits of network effects create tremendous incentives for consumers to prefer a compatible operating system and for the new operating system companies to retain compatibility with the installed base and each other."

Expect to hear more of this line of argument if Microsoft and the government are unable to reach a settlement in upcoming months. Federal judges are notoriously conservative. Unless the pro-breakup crowd does a good sales job, it will face an uphill battle.

France covers Legal Affairs for Business Week in New York.
Have a question or a comment? Let him know at mike_france@ebiz.businessweek.com.


_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Geoff Smith
Mike France covers Legal Affairs for Business Week


WEB POINTERS
Click here to visit some of the sites mentioned in this column:
Stan Liebowitz


Copyright 2000, by The McGraw-Hill Companies Inc. All rights reserved.
Terms of Use   Privacy Policy