Why do
people spend time and effort on innovative projects without direct
compensation? In a recent paper with Mark Schankerman (London School of
Economics), I explore this question in the context of open source software
development. In open source software
(OSS), programmers are typically unpaid, and the code is available for public
use and development under project-specific conditions which often make it
essentially free.
But how
is open source innovation sustained when its workforce doesn’t usually get
paid? This question has broader applications, since we see similar situations
in other areas in which “open commons” production has been proposed, including
biotechnology and nanotechnology. Scholars have proposed four broad
explanations: ideology, reputation, enjoyment or other personal value, and
future gain.
In our
paper, we observe code contributions in order to quantify how each of these
motivations drives open source innovation. We exploit the fact that there are different types of
projects. Though open, unrestricted access was the original driving force
behind the “free software” movement, many OSS projects now include licenses
that limit terms of use. For the sake of clarity, we call these
"closed" projects.
We focus
on four distinct groups of developers, whose profile we infer based on the
types of open source license governing the projects with which they are
affiliated: open, closed, mixed, and anonymous. Open developers are developers that are associated only with
projects that keep the code essentially free. On the other extreme lie the closed developers who belong only to
projects that allow the commercial exploitation of the code. Mixed developers
are developers that are not strongly affiliated with either open or close
projects. Lastly, anonymous developers are developers that do not reveal their
identity when making code contributions. The key to our approach is that we
examine the pattern of contributions –what economists call the “revealed
preference” of developers -- to infer their underlying incentives. That is, we
try to match patterns of code contributions to each developer type, according
to theoretical predictions underlying different motivations. Our results show
that software developers have a variety of motivations to contribute to OSS
projects:
1)
Strong identification with the “ideology” underlying the open source movement.
Open contributors almost exclusively contribute to projects with open licenses,
indicating dedication to the ideology of the open source movement. This very
important findings suggests that intrinsic, as opposed to extrinsic, motivation
governs effort decisions. In other words, how much a developer identifies with
the mission of the project should determine his or her level of involvement in
the venture.
2)
Increased reputation. Contributors from closed projects are more likely to
contribute to larger and corporation-sponsored projects. This findings suggest
that gaining reputation is also important for OS participation, extending
support to extrinsic reasons to devote efforts to specific projects. But
importantly, extrinsic type in the form of increased reputation is likely to
affect more specific type of developers (closed), and may actually have no
effect on others (anonymous). Importantly, we find a small effect of project
size for anonymous developers (who cannot gain reputation, as they do not
reveal their identity), and a negative effect of corporate sponsorship.
3)
Enjoyment. For “hobbyists,” enjoyment may outweigh the effort they put in. Open
contributors are much more likely to contribute to projects aimed at end users
(e.g., computer games), while closed contributors target developer-oriented
projects (e.g., programming tools). This suggests that open source development
is more viable as a substitute for proprietary software innovation on the
end-product side.
4)
Future gains through reciprocity. The last theory suggests that developers may
expect future gains from projects to which they have contributed. We find,
however, that reciprocity plays a limited role. Developers are more likely to
contribute to projects from which they have previously received contributions,
and it is more common among closed developers than open developers. This
suggests that reciprocity is associated more with building reputations than
with intrinsic motivation.
So what
does this all mean? The implication is that whether or not a project has an
open source license, and the openness of the license, affects contributions to
the project. For example, a more open license increases contributions by one
kind of developer but reduces those of other developer types. In the context of
setting license criteria, open source managers face an important trade-off: on
the one hand, choosing a more closed license would mean more potential profits
due to fewer commercialization restrictions. But on the other hand, the same
closed license would substantially reduce contributions that come from
developers that strongly identify with the open source ideology.
These
findings -- that motivations are varied and induce sorting behavior -- should
also be relevant beyond the OSS context. A combination of indirect and direct
rewards can be more useful than just direct rewards. When indirect motivation
is strong (e.g., in the academic and NGO sectors), low-powered incentives may
be more effective, and direct incentives may even cancel out indirect
motivation.