ftp.umcs.maine.edu/2015-02-07.ftp.umcs.maine.edu.tar/ftp.umcs.maine.edu/pub/WISR/wisr4/proceedings/detex/lavoie.detex

Text File | 1992-04-05 | 17KB | 367 lines

 [12pt] article








  Increased Productivity Through Reuse:
An Economist's Perspective

  Don Lavoie, Howard Baetjer, and William Tulloh
Center for the Study of Market Processes
George Mason University
Fairfax, VA 22030
[e-mail]





From the economic viewpoint, increased software reuse seems
to depend on these factors:



continuing development and use of the modularization
techniques of encapsulation and information hiding in order
to save on maintenance costs

division of programmer labor into increasingly
specialized areas, accompanied by area-specific tools,
frameworks, and reusable components

investment in a healthy reuse environment, including
market-based management structures that address the
communication, discovery, and incentives problems, and the
increased use of component libraries and/or markets

property rights regimes that appropriately balance
producer and user interests.



  Introduction
Widespread reuse of software components has long been
advocated as a key to increased productivity and
better quality software.  In addition, markets in reusable
components have been seen by some as an important element in
realizing these improvements  .
While much progress
has been made in addressing the relevant technical factors,
it has become increasingly apparent that organizational,
cultural, institutional, and economic factors must also be
addressed in order to reap the gains from wide-scale reuse.

Among the important technical advances have been developments in
abstraction mechanisms, modularization
techniques and object oriented technologies.
These technologies promise to play an important role in helping to
realize the benefits of reuse by enabling alternative ways
of packaging software through modular construction.
Software developers will be able to take advantage of
economies generated by the increased division of labor and
capital,  while maintaining the flexibility needed to adapt
to changing requirements and demands.  Some of the benefits
taht are seen as deriving from reuse include lower
development costs, reduced time of development, greater
predictability of budget and schedule, and lower maintenance costs.

The Agorics Project, a team of researchers at the Center for the Study of
Market Processes at George Mason
University, is now engaging in a comprehensive study of the
software components industry.  Our study will examine the
interaction of these technological advances with the
institutional and economic factors that will affect the
development of a vigorous software components industry.  In
particular we will investigate the possibility of markets in
reusable components.

  Modular Construction and Object-Oriented Technologies

Many different visions of markets of software
components have been proposed, based on a wide variety of
abstraction mechanisms and resulting definitions of
components.  Our own investigation is taking a rather
inclusive view of the possible scale and nature of these
components, but we see object-oriented methods as an
indispensable part of the move to reuse.  The more modular
methods of software development that result from the
adoption of object-oriented technologies promise to provide
a dramatic increase in the reuse of components and
frameworks.

A key aspect of object-oriented programming in
encouraging reuse is the support of encapsulation.
Encapsulation of domain abstractions allows for greater
reuse of others' experience.  Information hiding through
encapsulation also eases the task of integrating different
components.  Mark S. Miller and K. Eric Drexler have
liked encapsulation in object-oriented programming to the
discovery by programmers of property rights over data  .
They write, ``encapsulation in software serves the same
crucial function as property rights in human affairs: it
establishes protected spheres in which entities can plan the
use of their resources free of interference from
unpredictable external influences.''  This modularization of
data and action helps provide greater flexibility in the
reuse of software components.

One great virtue of property rights in markets is the
ability to rearrange them in response to changing
circumstances and in order to meet new demands.  Similarly,
object-oriented methods promise better maintainability and
evolvability through encapsulation and information hiding.  The modular
nature of object-oriented development encourages
reuse by facilitating alternative packaging of functionality
across different applications, allowing
 a wider variety of
 user requirements to be met.  In addition, modular
 construction allows greater scope for reuse in sequential versions of an
 application by facilitating quicker response
 to changing user requirements through time.

  The Structure of Production of Software

Reuse in software involves a shift to a more
capital-intensive software development process  .  The
accumulation
of capital is the accumulation of knowledge and experience.
Capital goods such as components, frameworks and designs
embody domain knowledge, while tools, environments and
methodologies embody knowledge that aids in the solution of
recurrent software development problems.  Effective reuse
must build on prior experience and adapt this knowledge to
constantly evolving needs.  To achieve large gains in
productivity and quality from reuse will require reuse
across a variety of developers and organizations.  New
software communities   must emerge that enables large-scale
cooperative reuse.

Traditionally economists have seen increased
productivity as resulting from the increased complexity of
the structure of production -- the extension of the
division of labor and greater use of capital.  Software development,
however, is still based largely on a model of craft
production.

Craft production has certain advantages; most important
among them is its flexibility in adapting to rapid change
and in customizing the product to meet specific needs.  Human capital,
the skills and experience of programmers, represents the main
source of reuse, as well as the main
source of creativity and flexibility.  But a reliance on specialized
human capital can be risky: wide diversity in skills, high
programmer turnover, and the expense of training and
education, all limit the effectiveness of this
form of reuse.  More importantly, however, this form of
production does not scale up well in its ability to cope
with more and more complex software projects.

A common solution to this problem in other industries
has involved a shift from a complex production
process to a more complex structure of production.  Rather
than build in a single complex stage with multipurpose human
capital, the idea is to build a complex series of simpler
stages with more specialized human capital.  The increasing
division of labor is aided by the division of tasks into
easily manageable units, and by the standardization among
resulting static advantages of the division of labor: 1.
increased mechanization -- the substitution of capital goods
for human capital 2. reduction in the amount of human capital (skill)
required for each task due to the increased
routinization of tasks, and 3. economies of scale in
manufacturing.

But software is not amenable to traditional scale
economies of mass production.  Software development requires
greater flexibility than is offered by the efficiency gains
in standardized manufacturing processes.  In software,
development, not manufacturing, is of the essence.
Moreover, attempts to benefit from an increased division of
labor are often overwhelmed by coordination and
communication problems  .

What is needed is the ability to maintain the
flexibility and creativity of the craft approach, while
reaping the gains in productivity due to a greater division
of labor and a greater use of capital, thus enabling more
complex software projects to be undertaken.  Such an
approach which some have termed ``lean production'' as
contrasted with craft and mass production, must focus on the
dynamic advantages of division of labor and capital  .
These advantages come from the more effective use of
knowledge and enhanced learning that accompanies such
coorperative efforts.  The advantages can be found in: 1.  the
increased rate of discovery of product and process
improvements 2. the move from human capital to knowledge
embodied in capital goods seen not as a dumbing down of
production, but as a freeing up of human capital for
application on the margin of improvement, and 3. economies
of scope, not of scale, in development from the sharing of
resources and experience across multiple projects  .

  Investment in Reuse

To realize these benefits requires new organizational
and institutional responses to the increased division of
labor and capital.  Changes in the technological structure
of production must be accompanied by changes in the
institutional structure of production.  A variety of
property right regimes, contractual relationships, and
management structures must be explored to solve problems
that arise with the communication and incentive that
coordinate different stages of the production.  The forms
that will emerge depend on tradeoffs between various
institutional and technological solutions to these
coordination and communication problems.

One important change that arises is the need to ensure
an adequate return on the investment required in developing
and maintaining the capital asstes necessary for wide-scale
reuse.  This problem emerges whether reusable components are
produced within or across firms.  In fact,  the problems
within firms can often be usefully analyzed as analogous to
problems in the market.

Investment in reusable software requires that software be managed as a
valuable asset, and that the necessary organizational support be given to
ensure a chance at an adequate return on the investment. This requires a long
term commitment by management, including investment in the complimentary
assets (infrastructure), and changes in evaluation criteria and reward
structures. One of the key problems facing the firm is how to provide
appropriate incentives for both the creation and use of reusable components,
and assuring adequate communication of what components are available.

Much work on reuse has dealt with the most appropriate repository for reusable
software components. Advances in search techniques and classification schemes
provide an important area of research in improving access to components.
Technological advances can help reduce the transaction costs of search: such
techniques as browsers, hypertext navigation, classification methods, software
agents, and good old librarians can all play an important role.

In addition to such improvements in library technology, increasing attention
is being paid to the organizational incentives and communication flows needed
to maintain a flow of reusable components into the library and a flow of
reusable components to projects when and where they are needed. Repositories
by their nature are not well suited to providing useful incentives. The
problem arises of how to allocate the costs of building and maintaining the
library among the various projects that use components from the libraries.
Software, of course, is not subject to wear and tear, but it is subject to
economic obsolescence. Continual investment is required in the maintenance of
value to user. Rewards for useful contributions to the library and rewards for
making use of the library need to be a fundamental aspect of project
development.

Libraries as a shared resource serving many masters can be viewed as what
economists call a commons. The problem is the opposite of the usual story of
the commons; the problem is not the danger of overgrazing a fixed supply of
pasture, but of ``undergrazing'' reusable components due to the lack of
incentives and information regarding what makes for good reusable components,
and when it is appropriate to use them. Care must be taken in setting up the
appropriate rewards for contributions, and in reducing the costs of accessing
and learning about the contents of the libraries on the part of users.

Electronic markets also offer a potential solution to the problem of
affordable access to a wide variety of reusable components. Libraries are
essential passive; they wait for users to come search them. Electronic
markets, in contrast, reflect the active matching of buyers and sellers.
Component vendors have an economic incentive to produce useful components
and to reduce the transaction costs to users of gaining knowledge of and
access to components. We can imagine the emergence of component brokers who
try to match available components to developers' needs.


  Property Rights and Markets in Reusable Software

Many barriers must be addressed before vigorous markets in reusable components
will emerge. Problems of importance include the limited extent of the market
due to lack of standards, language incompatibilities, and domain specificity,
and the need to establish certification, testing and reputation mechanisms.
The biggest challenge facing the emergence of markets, however, is
establishing appropriate property rights regimes which can balance producer
and user interests. Entrepreneurs must discover combinations of attributes and
functionality that users value (and means of delivery which keep low the
transaction costs to the user) and price them so as to earn profits. The
challenge to establishing markets in software components lies in finding the
institutional and technological solutions which allow producers to be
adequately rewarded, and ensure users receive quality software in a timely and
affordable manner.

The ease with which software can be copied signals the difficulty of defining
property rights in software. Component pricing policy, and the very nature of
what is bought and sold, may well be shaped by tradeoffs between legal and
technological means of protecting intellectual property. Current developments
in patent and copyright law may determine what types of reusable components
may be bought and sold. In addition to intellectual property law, liability
issues in regard to reusable components are also an important issue that must
be addressed, but as recent Nobel Laureate Ronald Coase has argued, this too
is at root of a question of ownership  .

We need not depend entirely on legal means; as experience from other
industries has shown us, there exists a wide array of possible means of
appropriating sufficient return on the investment in intellectual property. In
addition to legal solutions, there are a variety of technological and
marketing options that have worked in other industries. The solutions most
appropriate for different kinds of granularities of software components should
emerge as markets evolve.


   mcil68

McIlroy, M. D. (1968) ``Mass Produced Software Components'' in P. Nauer and
B. Randall (eds.)    Software Engineering , NATO Science Committee, October.

Brooks, Frederick (1987) ``No Silver Bullet: Essence and Accidents of Software
Engineering''    Computer , April.

Cox, Brad (1990) ``Planning the Software Industrial Revolution''
IEEE Software ,
November.

Miller, Mark S. and K. Eric Drexler (1988) ``Markets and Computation: Agoric
Open Systems'' in B. Huberman (ed.)
The Ecology of Computation , North-Holland.

Wegner, Peter (1984) ``Capital-Incentive Software Technology''
IEEE Software ,
1:3, July. See also Ludwig M. Lachmann (1956) Capital and Its Strucutre
Kansas City: Sheed Andrews and McMeel.

Gibbs, Simon, Dennis Tsichritzis, Eduardo Casais, Oscar Nierstrasz, and Xavier
Pintado (1990) ``Class Management for Software Communities''
Communications of
the ACM , September.

Brooks, Frederick P., Jr. (1975)
The Mythical Man-Month: Essays on Software
Engineering , Addison Wesley Publishing Co.

Womack, James P., Jones, David T., Ross, Daniel, and Carpenter, D.S. (1990)
   The
Machine That Changed the World  New York: Maxell MacMillan International.

Cusamano, Michael A. (1991)
Japan's Software Factories: A Challenge to U.S.
Management , Oxford university Press, pp. 427-28.

Coase, Ronald (1960) ``The Problem of Social Cost''
Journal of Law and
Economics , October.



  About the Authors

Don Lavoie is Associate Professor of Economics at George Mason University.
He is a leading critic of centrally planned economics and co-director (with
Mark S. Miller) of the Agorics Project, a research effort sponsored by the
Center for the Study of Market Processes and aimed at investigating issues
that overlap between computer science and economics. Howard Baetjer and
William Tulloh are graduate students in the Economics Department at George
Mason University and members of the Agorics Project.