Summing Up

"Arguing in My Spare Time," No. 28

by Arnold Kling

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


November 18, 1998

I have been writing these essays for a year, and I believe that I have used up over 85 percent of my ideas on the topic that is at the intersection of economics and software. My intention is to take a few months off from essay-writing. I may send short messages to the list from time to time about interesting books, articles, or what have you.

This essay is a review of the main themes of the aimst essays, with a few comments thrown in. The essays are referred to by number, with the number of the essay in brackets [ ]. The themes will be reviewed more or less in the order in which they were introduced.


  1. The trend in software is for more components to be generic and fewer components to be proprietary [1], [4]. This trend’s primary impact is to create a shift in software development between the Dilbert sector (large organizations in government, banking, telecommunications, etc.) and professional software firms (Microsoft, Oracle, and many smaller firms). New applications in the Dilbert sector will have relatively more components derived from generic software and relatively fewer components derived from proprietary development. For example, the Web interface will be used more, and development of proprietary interfaces will diminish.
  2. Companies are increasing profits not so much by raising productivity but by holding down salaries of workers in general, and technical workers in particular. [2], [13], [14], [21]. The "labor shortage" should be interpreted as a temporarily below-equilibrium wage rate.
  3. Bureaucracy has a useful role in controlling risks undertaken within large firms. [3], [15], [16], [26], [27]. However, the traditional source of risks in large firms—the need for large investments in physical capital—is declining in importance. I particularly question the role of bureaucracy in systems development. Going forward, we may see more resources shifted to entrepeneurial environments, where the incentives for managing risk are better aligned.
  4. Network effects are over-rated, and they are not the only explanation for Microsoft’s success. [4], [5], [19], [23]. The Internet stocks that the market is swallowing today all rest on the theory of network effects. In the absence of network effects, these companies have no assets that could conceivably be worth millions of dollars, much less billions. I would love to own a long-dated put option on an index of these tulip bulbs. As to Microsoft’s success in spite of flawed software, all software is flawed relative to some theoretical ideal. Microsoft’s software tends to be less flawed than that of its competitors. Microsoft’s architecture arguably is too closed and proprietary for the state of the art in 1998. However, this fact will be of interest only to a small elite among software developers. It will not affect the typical corporate systems department still struggling to free itself from its mission-critical COBOL applications, or the many people who still have not gotten comfortable with computers.
  5. Software is difficult to price, because the marginal cost arguably is zero. [9], [10], [11], [20]. I suggested that an intermediate solution between freeware and cost-per-license is a software club, where people pay dues to a club and are given unlimited usage of software developed or licensed by the club. I do not believe that "open source" software will become a significant factor in the end-user market without some kind of economic model. The tedious work needed to polish software for average end-users is not something that developers are going to want to do for free. It is possible that software clubs will have some of their software developed using the "open source" model, as long as they can offer complementary products and services in a way that justifies the investment in "open source" development.
  6. The critical resource to allocate in today’s economy is no longer capital, but talent. [12]. [18], [22], [24]. The institutions best suited to allocating talent probably have not yet evolved. Neither bureaucracies nor the stock market are likely to continue to be as central as they are today.
  7. Java is overrated as a force for change [6], [7]. I continue to believe this. I do not believe that the problem that Java purports to solve (enabling applications to be written once and run everywhere) is on the critical path to moving business forward. What is on the critical path is enabling data sharing across enterprises. To the extent that XML becomes the solution to that problem, my prediction is that by the year 2005 XML will be regarded as a more important development in computing than Java.
  8. On the other hand, I did change my mind on network computing between [8] and [23]. I now believe that the evolution away from desktop devices to mobile tools is going to be a major driver of innovation going forward. However, one point that I made in [8] I continue to believe. That is that a thin operating system makes a lot of sense for a server. Far be it from me to second-guess Microsoft, but I fail to grasp the point of a server operating system with tens of millions of lines of code. Microsoft’s web server and database server products would be very attractive to me if they would run under Linux or some other reliable operating system. My sense is that the web server, in particular, is the most developer-friendly product of its type on the market. The many excellent web pages that one sees with the .asp extension reinforce this impression. But we do not use it because we are not willing to suffer with Windows NT. From my perspective, NT looks like a classic white elephant software project.