Open Source Software:
Free Provision of Complex Public Goods
by James Bessen (Boston University School of Law)
FULL TEXT
Introduction
Empirical research on Free/Open Source software (FOSS) shows that a large part of the development effort is directly supported by firms, not just individual hobbyists. One survey finds that about half of the development is done by employees at work, with the knowledge of their supervisors (see Lakhani and Wolf). And although some firms may participate in order to undercut direct rivals—for example, a direct competitor of Microsoft may support Linux—most involvement does not seem to involve such rivalry. Indeed, direct rivals may even share code with each other (see Henkel and Tins).
Much of the initial research on FOSS has focused on the motivations of individuals. This paper asks, instead, what motivates firms to participate? The answer, developed in a formal economic model, is that participation in FOSS development often gives firms the best way to get the software that most closely meets their needs. As Richard Stallman has argued, the freedom to modify and improve software delivers a key benefit of Free Software.
The Limits of Proprietary Software Provision
Of course, an economic intuition holds that proprietary provision should be more effective than voluntary contributions. In other words, volunteers won’t contribute enough effort to the production of a “public good”—a good that is shared without one user’s consumption reducing another’s—because they don’t directly benefit from the enjoyment others receive from the good.
But this paper argues that this notion is misleading, because so much software is not a simple, homogenous good used by all consumers in the same way. Much software is complex, and its users have highly diverse needs, which often require specialized features. Despite the common view that Microsoft is a typical software company, most software investment by firms is actually made in custom development, not in purchasing pre-packaged programs.
Firms are willing to make this investment despite the difficulty of writing contracts for custom development. The complexity of software makes it difficult to specify the desired product—indeed, the software itself is the complete description of how the product will function under all circumstances and this is what the developer is paid to create. A standard economic result is that with “indescribability,” an “incomplete” contract will not elicit the socially optimal level of effort from the developer, leaving the customer with a poor quality or otherwise inadequate product.
To the extent that pre-packaged software can group features together and aggregate demand over many prospective customers, it may generate a more optimal level of effort and deliver goods that match many, but not all, customers’ needs. To maximize revenues, pre-packaged software vendors bundle as many features as possible into their products, but they are limited because complex interactions between these features drive up the cost of debugging and maintenance, which account for most of the cost of software development. Firms limit the feature set to contain these costs, which limits the applicability of the software to many users. In a larger market, firms can profitably include more features, and so the market share of pre-packaged software has grown together with the number of computer users. But the limitation on the feature set explains why so many customer needs are not addressed by pre-packaged products.
A pre-packaged software firm may address some of these needs by creating an Application Program Interface (API). The customer purchases a developer’s toolkit to access functions coded in the pre-packaged software through the API. This permits customers to develop their own applications, but to piggyback on the code of the pre-packaged software firm, saving some development cost. However, this functionality comes at a cost—the price of the developer’s toolkit—and some customers will be priced out of the market.
FOSS
Free/Open Source can improve further on this model. FOSS can be viewed as a free API. Moreover, because developers add newly coded features back into the shared code, the underlying functionality of a FOSS project may grow more complex—and meet more needs—than a competing pre-packaged software product. Overall, FOSS is a more effective way for firms to obtain specialized software.
In the paper, the model shows that FOSS and pre-packaged software can both exist in the same market because to a large extent they address different groups of customers. Pre-packaged software addresses simpler needs desired by large numbers of people; FOSS addresses more specialized, often “geekier” needs. In effect, FOSS does not replace proprietary provision in general (though it does for some customers); for the most part, however, it complements proprietary provision.
The paper concludes that it is a mistake to assume that FOSS is somehow less robust because it is based on voluntary contributions rather than driven by the profit incentive. The firms that participate in FOSS are driven by the profit incentive—FOSS is just an alternative to simple “make-or-buy” decisions.
