SOFTWARE REFEREES

Arnold Kling, "Arguing in My Spare Time", No. 11

On-demand Printing

In the Internet arena, will supremacy in book selling and distribution belong to Barnes and Noble, Amazon.com, . . .or Kinko's?

This question occurred to me when I heard Robert Main of Electric Press in Reston, Virginia describe the "on-demand printing" system that his company has developed. The idea is that any digitized content can be sent at any time to any Kinko's copy center to be queued for printing. With high-quality copiers, this means that even slick magazines can be printed at a copy center.

For publications as they exist today, this might have two advantages:

1. Think globally, print locally. Because it is cheaper to ship bits than atoms over long distances, the cost of distributing high-quality publications could be less with an on-demand printing architecture.

2. Just-in-time inventory. Today, in order to sell a book you have to maintain it in stock. This creates a huge inventory expense. With on-demand printing, there is no need for physical inventory.

Reduced shipping and inventory costs may or may not cause on-demand printing to supplant traditional distribution for books and magazines. The impact may be relatively larger for low-volume publications, for which inventory and shipping costs might represent a higher proportion of total costs.

However, on-demand printing might change the nature of publishing altogether, particularly for nonfiction. On-demand printing reduces some of the fixed costs associated with printing and distribution. The cost of printing the first copy is not that much higher than the cost of printing the 1000th copy.

In the academic world, a professor could pick a set of chapters or articles from various sources and have them printed as a reading list, a journal, or a textbook. Textbooks could be updated quickly by printing out substitute chapters on an as-needed basis.

Using a metering approach along the lines of Brad Cox's Superdistribution, authors could be paid each time an article or a chapter is converted from bits into atoms. The same article might appear in several different collections.

On-demand printing would not bring about total disintermediation in the sense that there would be nothing between authors and consumers. The importance of editors, reviewers, referees, and other intermediaries would be likely to increase in an environment in which the fixed costs of printing are low. There may be many more possible collections of articles, and I will not want to read every collection. I will be willing to pay someone else to filter out articles and recommend them to me.

Refereed Software

What I am suggesting about on-demand printing is that it could bring about a

subtle shift in the design of books and journals. The design would become more decentralized, moving closer to the end-user.

If this were carried to the extreme, then as a reader I might design my own magazine by selecting articles from authors that I like. More likely, I will delegate this task to other people whose judgement I trust. However, unlike today, where to be a viable editor one must obtain a large circulation, with on-demand printing an editor with a much smaller subscriber base could cover the expense of reviewing articles. Magazines might evolve from a relatively few mass-market publications to many refereed journals fitting various niches.

Is there an analogy with software? Could the importance of mass marketing decrease, while the importance of software referees increases? Who would become the referees?

With the Internet, software distribution already has moved in the "on-demand" direction. Instead of going to the store to obtain a hard copy, you can download software from the developer's web site.

One interpretation of Netscape's decision to release the source code for its browser is that Netscape has decided to become a referee rather than a developer for its browser. Netscape will set standards for how components are developed, but it will not develop components.

Referees set standards for software. They also provide recommendations to consumers. In that sense, people who write software reviews are referees. The committees that resolve software standards (for Internet Protocols, for example) are referees.

A controversial situation is when a company is both a developer and a referee. Sun apparently has taken on this role with Java.

Microsoft also acts as a referee. For example, when it bought the Front Page development tool from Vermeer, it effectively anointed Front Page as the leading software of its type. Late last year, Rich Seidner, a software developer in the U.K., wrote an article for CNET called "Crushed by Microsoft" in which he described how another firm's video driver was chosen over his by Microsoft. Again, my interpretation would be that Microsoft acted as a referee.

If the concept of software clubs, described in the previous essay, were to get off the ground, it would require high-caliber referees. Software clubs would have to pay referees to evaluate software. These referees would have considerable influence on the design of software.

Today, most software referees cater to a mass market. Software reviews primarily enumerate features, and one's impression is that the product with the most features is declared the winner. This reinforces the development of bloated, lowest-common-denominator software.

Smaller niche markets might become more prominent. If "bloatware" is to be replaced by more specialized products, then I believe that this will come about because the economics of software distribution change to give more power to specialized referees. It will have little or nothing to do with Network Computers or Java.

The question is whether the market can support a diverse set of referees. If so, then there can be some evolution of software away from requiring mass market acceptance.