[Patents] James Bessen: Strong Property Rights Can Limit Provision Software

[seth.johnson@RealMeasures.dyndns.org (Seth Johnson)]

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=278148

Open Source Software: Free Provision Of Complex Public Goods

JAMES E. BESSEN
Research on Innovation

Abstract: 
Open source software, developed by volunteers, appears counter to
conventional wisdom about private provision of public goods. Standard
theory holds that without property rights (or excludability),
free-riding inhibits private investment in non-rival goods such as
information and software. But complex open source products challenge
commercially-developed software in quality and market share. I argue
that the complexity of software changes the results. For complex goods
under asymmetric information, open source developers self-select,
offsetting free-riding losses. But commercial firms lack information
necessary for effective provision. I compare different forms of
provision and different property rights. Strong property rights can
limit provision of complex public goods.


Excerpt from Introduction:


On first examination, open source software seems paradoxical. Open
source software is a  public good provided by volunteers---the ``source
code'' used to generate the programs is freely  available, hence ``open
source.'' Networks of thousands of volunteer programmers have developed 
widely used products such as the GNU/Linux operating system and the
Apache web server.  Moreover, these are highly complex products and they
are, arguably, of better quality than  competing commercial products,
suggesting that open source provision may be highly efficient. 

This appears to counter the common economic intuition, supported by a
large literature,  that private agents, without property rights, will
not invest sufficient effort in the development of  public goods. As
Arrow [1962] argued about information goods generally, under­investment
arises  because each agent faces a free­rider problem or public
externality. Although each programmer  realizes some benefit from her
own development effort, she gains nothing (ignoring altruism) from  the
benefit that all the other consumers realize from her software. Since
her effort is guided only by  her private benefit, she invests less than
the socially optimal amount. 

As Coase [1960] famously argued, property rights allow firms to
internalize such  externalities as long as transaction costs are zero. A
firm that develops a proprietary software  product obtains profits from
all of its customers and so it may invest more. In a large market with 
many customers (as opposed to, say, a bilateral bargaining situation),
the market centralizes  information on the demand for simple
commodities, reducing transaction costs and permitting a  greater level
of private provision. 

This paper argues, however, that software and many other information
goods are not  usually simple commodities. Instead, software is a
``complex product,'' meaning a product with  many features that work
together. Because consumers have different preferences for each feature
in  a complex product, they use different combinations of features. This
creates a problem relative to  simple models because, in effect, the
customer consumes only a specific instance (the ``use­  product'') of
the general product the firm sells. The firm sells a product with M
optional features  that may or may not be used with each other. This
represents M ^ 2 different use­products, of which  only one is of
interest to any given consumer. 

This distinction creates a real economic difference when the firm faces
a cost, even a slight  cost, for each use­product. And indeed, the
quality of software depends on the extent to which  different
use­product combinations are tested and debugged. Because the features
in a complex  software program interact with each other, each
use­product must be individually tested to ensure  that it works. Yet
firms cannot feasibly test all possible use­products because the number
of  possible combinations is astronomical. 1