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



Arnold Kling, "Arguing in My Spare Time", No. 10

March 29, 1998


At the end of the previous essay, I introduced the concept of a software club. The idea is that individuals and businesses could join clubs. The fees for joining these clubs would be sign-up charges and membership dues. The clubs would purchase the services of software developers, and members would be able to use all of the software provided by the club at no marginal cost.

This essay will explain the "club" model in more detail. What problems does it solve? How would the business model operate?

What's Broken?

A popular adage goes, "If it ain't broke, don't fix it." For the concept of the software club to be worthwhile, it should fix something that is broken.

Here are some aspects of the software market today that one can argue are broken:

1. The equililibrium market structure is monopolistic. With software revenues based on licenses, and the marginal cost of supplying a license zero, the tendency is for one company to achieve dominant market share. For example, Microsoft's share of PC operating systems is almost 100 percent, and Oracle's share of high-end relational databases is over 60 percent.

2. Innovation may be hampered by fear of imitation. Again, the marginal cost of providing a licensed copy

of software is close to zero. Therefore, a company that copies someone else's software is in the most

advantageous position. It does not have the R&D cost of developing the software, and it can choose to

sell at low marginal cost in order to gain market share. Therefore, innovative software companies have

much to fear from immitators, particularly if the immitator has a better distribution channel.

As Mario Morino and others in the software industry have pointed out, pure R&D is not the biggest

expense in software development. I would stress that creative inspiration by itself does not produce quality

software. It is often the steps that involve the most pain--documentation, testing, and enhancements--that

provide the most gain. But even with these functions performed successfully, the challenge of positioning

and distributing the product in the market is very formidable.


3. There are incentives to skimp on quality. As long as you can get away with selling licenses for a flawed product, there is not much incentive to improve it. The users/victims have little or no leverage over the developers.

4. There are important "externalities" that are not addressed. In economic jargon, an "externality" is an activity that has economic impact that is not priced. Pollution is the classic example.

In the case of software, externalities are a result of complementarity. Complementarity means that two products reinforce one another. For example, if there were no gas stations, people would not find automobiles very useful. However, if people did not have automobiles, there would be no incentive to build gas stations. In that sense, gas stations and automobiles are complementary.

Computer software often involves complementarities. For example, spreadsheets and printer drivers are complementary. If no one writes printer drivers for an operating system, then no one will write spreadsheets for that operating system. If no one writes spreadsheets, then there will not be any demand for printer drivers.

I believe that the main reason that Linux, the elegant freeware operating system, cannot compete with Windows is that there is no institution that can internalize these externalities and turn Linux into a practical mass-market product.

5. Very high risk in software development

With its "winner-take-all" characteristics, software development involves very high risk. It is an environment in which there will be a few big "hits" and a lot of well-meaning failures. The ability to develop good software can be overshadowed by the need for capital and marketing skills as well as by sheer luck.

6. Expensive distribution channels

Many capable software developers cannot reach the market with their products, because it is very expensive to reach distribution channels. Even though the Internet has alleviated the need to obtain physical "shelf space," there remains a tremendous challenge of getting word out to consumers about a product.

The Software Club Solution

If the software market consisted of software clubs, rather than by-the-license marketing of software, this would alleviate the problems described in the preceding section. A software club would be competing with other clubs to try to achieve a maximum membership rate at minimum cost. This would create incentives to address the problems that exist in the current software market.

1. A large club would have more power than an individual buyer. This would enable the club to address the issue of monopoly power more effectively. While no individual buyer can make it possible for an alternative database product or operating system to be viable, a large software club would be willing to pay millions of dollars to develop such a competitor in order to avoid paying billions of dollars to a monopolist.

2. With several software clubs competing, the benefits of being an imitator might be reduced. People might prefer to join a software club with a large R&D budget, in order to get early access to leading-edge software. However, not all of the advantages of being an imitator would be eliminated by the software club concept.

3. The incentive to provide high quality software could be increased with a software club. Revenues depend on the satisfaction of the members, not on the ability to ship many copies of a product. Testing and documentation might be accorded their proper importance.

This would probably be the biggest advantage of a software club over the alternative of "free software." Many people have questioned whether "free" software is a good idea, because of the question of where to go for support. The support problem may be soluble, but it is part of a larger issue of follow-through. When a user adopts free software, they are doing so knowing that nobody has any incentive to improve, enhance, upgrade, and document the product. Maybe these activies will occur, but chances are they will not.

I would not suggest that the software club approach would guarantee high software quality for everyone. Some clubs might offer low membership fees and offer a broad array of mediocre products, and the market may support this. However, this would leave room for other clubs to offer high quality products at higher membership fees.

4. A software club could internalize some of the "externalities" in software. The club would have a large enough budget to fund the development of complementary products. This might be the most powerful advantage of software clubs over the current industry structure. Right now, only a company with the near-monopoly power of Microsoft is able to deal effectively with these externalities. It is, in effect, the prototype software club.

5. A software club would absorb much of the risk of software development. It would pay for development, regardless of whether the product that results is a "hit."

6. Software clubs would have built-in distribution channels, so that a software development company that works with the club would not need to execute a consumer marketing strategy. The process of marketing software becomes one of convincing the software club managers that its members would use the product.

The Business Model

Clubs would be owned by shareholders and managed for profit. A software club might operate in some ways like a television network. The club would choose a "lineup" of software to offer to its customer base. It might purchase the distribution rights to software from independent studios or develop the software with salaried staff.

Another analogy might be a reseller of long distance telephone services. Like a long distance reseller, a software club needs a marketing and billing operation, but it may not own the technical infrastructure.

Suppose that there are 30 million potential members of software clubs, and each member has $1,000 per year to spend on membership dues. Remember, many memberships may be provided by corporations to employees. A typical club might charge $200 - $400 per year for membership, and a typical individual might belong to three to five clubs.

Continuing with this arithmetic, the total revenue of software clubs would be $30 billion per year. Assume that half of this goes to overhead, including marketing, and the remainder goes to pay developers (including people to perform testing and documentation, as well as programmers). Then $15 billion would go to developers, or enough to pay 150,000 developers a year an average of $100,000 each. This would be enough for 5 Microsofts.

Suppose that there were eight major software clubs. They might have very different sizes and structures. One club might be structured like Microsoft, with a large in-house staff and a policy of purchasing smaller "studios" outright whenever it wants to acquire software developed elsewhere. Another club might have no development staff at all, and instead could put all its projects out to bid to independent studios.

Software clubs might not have exclusive rights to distribute software. A talented software studio might have products that are in demand by all of the major software clubs, so that it would get paid by each club for distribution rights.

To differentiate themselves, software clubs might aim to focus on certain vertical markets. Some clubs might be focused on providing basic office functionality to large corporations (document processing, email, etc.). Other clubs might focus on the SOHO (small office/home office) market. Another market might be for companies engaged in online retailing. Another market might be software for children, including education and games.

Economic Analysis

The software club is a different model for software pricing. Pay-per-license would be replaced by membership dues which convey rights to unlimited usage of software that is provided by the club.

My conjecture is that the software club leads to a different industry structure. Instead of pure "winner-take-all" markets for pay-per-license products, the structure would be one of oligopoly: in each major vertical market, two or three clubs would compete.

Even if there are not many software clubs, the market might prove to be highly competitive. If many software "studios" remain independent, then it may be fairly easy to enter the software club market. By making agreements with the enough software studios, a new entrant could compete as a software club without having to incur huge costs of hiring, research, and development.

Consumers want integrated, robust solutions to problems. Each software club would have the incentive to take a customer-centric perspective, rather than a product-centric perspective. Unlike a single small software a company, a club would have the resources and incentives to develop complementary products that provide a comprehensive solution to a problem. Right now, only Microsoft has enough scope to internalize this as an incentive. Other companies are focused on developing a single "hit" product.

Large software clubs would not have the incentive to meet the needs of unusual customers. Their needs might be met by small software clubs, or by software that continues to be sold on the pay-per-license basis.

In conclusion, I believe that software clubs would improve efficiency. Today's approach is to charge for something--a software license--that costs nothing at the margin to produce. A software club would pay software developers for their time. It would allocate their time in order to maximize member satisfaction, thereby increasing its ability to extract revenue from its members. This would come closer to pricing the correct resource.