Arnold Kling     Essays | Short Book Reviews | Favorite Links | Internet Bubble Monitor | Home


"Arguing in my spare time," No. 25

by Arnold Kling

Oct. 17, 1998

May not be redistributed commercially without the author's permission.

Microsoft is the poster child for the economic concepts of "network effects" and "lock-in effects." These are effects that allegedly can cause a bad social outcome.

A simple way to describe a network effect is to say that you buy an inferior product in order to be compatible with what other people are buying. (I owe that formulation in part to Bob McDonald).

A simple way to describe a lock-in effect is that you stick with an inferior product in order to avoid the cost of switching to a new product.

In my opinion, a clear-cut example of network effects and lock-in effects would be America Online (AOL). The only reason that I have an AOL account at home is for teenager compatibility. This is a network effect. My teenage daughters want to "chat" on line. The friends with whom they want to chat are on AOL. Although I suspect that these chats are at best insipid, I gladly will pay $20 a month to anyone who reduces the time that my daughters spend watching TV. In fact, if cable TV offered a plan that offers fewer channels than ordinary television, I would consider getting a subscription.

The only reason that many people have for keeping AOL accounts is that once they have given out their AOL email address to a lot of people, it is costly to change. This is an example of a lock-in effect.

Some people attribute AOL’s success to the fact that it is "easy." Be careful. Hal Varian has pointed out that it is important to distinguish "easy to learn" from "easy to use." He points out that the incentives can lead a software developer to over-invest in ease of learning relative to ease of use. AOL illustrates this distinction. Compared with the suite of tools from the typical Internet Service Provider, AOL is easy to learn. However, it is much harder to use.

In my opinion, much of AOL’s success is due to network effects, lock-in effects, and the advantage of being easy-to-learn rather than easy-to-use. It is easy to imagine a social engineer improving the equilibrium by breaking up AOL and forcing its users to get accounts with generic ISP’s. I am not advocating that government should do this. In fact, I would be opposed, for various philosophical reasons. I think that the welfare gain, if any, of such a move would be minimal. Moreover, I believe that it is a good idea for the government to allow private markets to operate even when it sees opportunities for improvement over the private outcome.

Which brings us to Microsoft. Here are the reasons that I believe that Microsoft has been successful:

  1. Microsoft developed the mass-market software business.
  2. Inventing a new mass-market industry confers great advantage. Look at Henry Ford. Other companies came close to being the software equivalent of Ford. Visicalc, for example. But Microsoft really perfected the process of developing software that fit the hardware and the consumer needs of the time.

    In the mainframe world, companies sold hardware and service, with software bundled in for free. Microsoft was able to avoid the costs of the service business and sell software as a stand-alone product. It may seem like an obvious strategy in retrospect, but at the time they were moving into uncharted waters. They were early, and they learned a lot about their business as they invented it.

  3. Until recently, Microsoft was more open and less proprietary than its competitors.
  4. As I argued in some early essays in this series, one can think of the Dilbert sector (large, bureaucratic companies and organizations) as competitors to Microsoft. Much of the growth of PC-based tools has been due to their superior flexibility when compared with proprietary Dilbert-sector systems.

    As an operating system, DOS and Windows were more open than Apple. Many observers believe that this relative openness was a major factor in the success of DOS/Windows over Apple. Only with the advent of Linux has a truly open operating system arrived to challenge Microsoft.

  5. Microsoft treats its key workers well.
  6. Just as Henry Ford was innovative in paying workers a "living wage," Bill Gates was innovative in offering computer programmers stock options. The tradition in the Dilbert sector was to treat programmers like the secretarial pool or janitorial staff. Naturally, Microsoft has been able to attract talented developers.

  7. Microsoft looks for revenue from high volume, not from high prices.
  8. Nicholas Economides has argued that Apple took the view that because it was the best, its prices could be higher. Thus, they deliberately eschewed market share dominance. If this is true for Apple, then it is true in spades for Oracle. Microsoft is saying that a relational database is a commodity, and Oracle is saying that people will pay more (by orders of magnitude) for a "better" brand. History is not on Oracle’s side.

  9. Microsoft frequently has best-of-breed applications.

For example, many people view Excel as the best spreadsheet program. Many people view Frontpage as the best Web page development tool. From an application developer’s perspective, Internet Information Server is the best Web server.

In my opinion, each of these 5 factors is more important than network effects, lock-in effects, or ease-of-learning effects in explaining the success of Microsoft.

Unless one is careful, arguing over what explains the success of Microsoft can be a dangerous exercise. It tends to be unscientific. One way to torture economists is to force us to read a best seller filled with anecdotes about winning businesses. We end up fantasizing about writing our own book, called "In Search of a Controlled Experiment."

With that difficulty in mind, let us proceed to consider the policy question of whether to break up Microsoft. This is a more interesting question than what is likely to be asked in the antitrust trial.

I myself cannot get worked up over the browser war. As a Web developer, I want browser features to be decided by standards bodies, not by unilateral browser company decisions. These days, Microsoft plays by those rules. Microsoft deserves to beat Netscape. Almost three years ago, I implemented a JavaScript program according to Netscape’s specifications and it bombed on Netscape but worked fine on Microsoft Internet Explorer. That pretty much has been the pattern ever since. Explorer usually does a better job of implementing standards than Netscape, even when Netscape wrote the standard.

Nor can I get worked up over "hardball" tactics. When we make deals with other Web sites, all sorts of hardball tactics are tried by both sides to try to gain exclusivity or other advantages. I am not saying that Microsoft is entitled to do anything illegal. Moreover, their quasi-monopoly position in the operation system may give them tighter legal obligations than those of us who are in more competitive situations. However, at the end of the day I do not think that a crackdown on "hardball" is going to make much difference.

The interesting question is, what would happen if Microsoft were broken up into two separate companies? Suppose that were split into an operating system company and an application-development company.

The anti-Microsoft thesis is that if this were the case, the application-development company would lose a major unfair competitive advantage. The hypothesis is that people choose inferior Microsoft applications because they believe that it is necessary to do this in order to achieve compatibility with the Microsoft operating system, which is so predominant.

I would like to see someone make this case logically and in detail. It does not sound right to me, particularly in the age of the Internet, where, as someone has pointed out, the joke that "No one knows you’re a dog" also means that no one knows you’re on Apple or Unix.

Instead of an advantage to owning the operating system, I see almost the opposite. If Microsoft’s database and server products were freed from the ball and chain of the NT operating system, then the demise of the competing products from Oracle and Netscape would occur even more quickly.

Microsoft’s NT developers in turn might be relieved by the break-up, because this would free them from the ball and chain of legacy applications running on DOS and Windows 95. To get users to adopt NT, they will have to overcome the lock-in effects of these earlier operating systems. Moreover, the development of NT has had to be compromised by the perceived need to replicate some of the functionality, look, and feel of Windows 95.

In conclusion, the case for breaking up Microsoft may be difficult to make. It might not change things very much. Or it could strengthen Microsoft in both markets. It’s not as obvious as the case for breaking up AOL.