A correspondent pointed me at a paper by Carliss Baldwin and Eric von Hippel, Modeling a Paradigm Shift: From Producer Innovation to User and Open Collaborative Innovation, which builds on my papers in some interesting ways. Here’s one of the money quotes:
Building on arguments of Ghosh (1998), Raymond (1999), and von Hippel and von
Krogh (2003), Baldwin and Clark (2006 b) showed formally that, if communication costs are low
relative to design costs, then any degree of modularity suffices to cause rational innovators that
do not compete with respect to the design being developed to prefer collaborative innovation
over independent innovation. This result hinges on the fact that the innovative design itself is a
non-rival good: each participant in a collaborative effort gets the value of the whole design, but
incurs only a fraction of the design cost.
If this sounds familiar, it should. It’s descended directly (as the authors acknowledge) from my argument in The Magic Cauldron that open-source projects are an inverse commons which encourage collaboration because that minimizes downstream costs to participating selfish actors. But Baldwin and von Hippel cast a much wider net than just software by developing a picture of user-driven and collaborative innovation in other industries — sporting goods, of all things, make several walk-ons in the paper.
Their demonstration is interesting in part because it bolsters my economic case against the GPL. The authors are saying, in effect, that we do live in what I called a “Type B” universe, where efficiency incentives favor the open-source design of software. In fact, they think they have a formal proof of this result, but I haven’t seen their 2006b paper so I can’t evaluate it. And they argue that the result applies to other forms of engineering design as well. I am cautious about this: they may be right, but my own intuition is that the effectiveness of open-source methods is limited when the limiting factor of production is something other than human attention.
There is one assumption they make that I think is stronger than necessary:
By focusing on anticipated benefits and costs we assume that potential innovators are rational actors who can forecast the likely effects of their design effort and choose whether or not to expend the effort
But, in fact, forecasting of this kind is extremely difficult and I don’t think any model should assume open-source contributors actually do it routinely. That would be equivalent to solving the patch-valuation problem I pointed out in The Magic Cauldron; I think there are good Hayekian reasons to doubt this is even possible.
I think Baldwin and Hippel are closer to the truth of the matter when they talk about institutions being self-reinforcing collaborative games, and I want to suggest a way in which that can explain rational-actor behavior without assuming a degree of knowledge and forecasting ability that the individual agents don’t actually possess.
In institutional game theory, an institution is defined as the equilibrium of a game with
self-confirming beliefs (Aoki, 2001). Within the institutional framework, participants join or
contribute resources in the expectation that other parties will enact their respective roles. If all
behave as the others expect, everyoneâ€™s initial beliefs are confirmed: the pattern of action then
becomes a self-perpetuating institution. When the participants in the institution are rational
actors, one of their self-confirming beliefs must be, â€œI am better off participating in this
institutional arrangement than withdrawing from it.â€ On this view, a stable nexus of contracts,
a solvent firm, and an active open collaborative innovation project are all special cases of
After this, Baldwin and von Hippel do some astute analysis of how changes in communication costs can make open-source collaboration viable. This is fine and worthy stuff; OK, so I got there first, but my arguments were qualitative and relatively informal. These two have have the analytical toolkit to be neoclassical where I was Austrian, and the degree of rigor they bring in is valuable.
But it seems to me that they are missing two other important effects: (1) rational evaluation of the collaborative game itself can substitute for forecasting the cost and value of particular design efforts, and (2) there is a sort of group-selection effect by which the relative success of different games changes individuals’ perceptions of the value of the one they are in. Where I am pointing here is towards nothing less than an economic analysis of the meaning and value of cultural loyalty.
Let us suppose that I, as an individual programmer, am exposed to the results of open source cooperation and find them beautiful and good. Because I want to be part of the culture that produces such things, I look for an open-source project to contribute to. I send a patch and experience the personal reward of seeing that it is incorporated in the project. Seeking that again, I contribute to this project, and perhaps to others.
I told a psychological story in the preceding paragraph, but I think there is an economic one underneath it. When the individual evaluates the products of the open-source culture are beautiful and good, he is learning confidence that his participation in the institution of open source is likely to bring him results he values. This confidence relieves him of the need to do a de novo payoff analysis for each bit of design effort he might invest. In effect, the manifest rationality of the entire game (judged by its products) reduces his decision costs with respect to any transaction in it. The effect is strictly parallel to the classical Coasian analysis of how firms lower decision and transaction costs for insiders.
With regard to my second point, institutions do not exist in isolation. Like firms, they have competitors. Competition occurs not just between individuals but between rival games, because institutions are selected for the ability to attract individuals to play their roles. Individuals judge institutions not only by their productive output in isolation but by their differences in productive output. Again, evaluation of these differences may, by reinforcing the beliefs associated with a winning institution, lower the individual’s decision costs about contributing design effort within it.
History matters, too. When an individual observes that an institution — such as an individual open-source project, or the entire open-source culture — has a long history of producing outputs of the sort the individual desires, that implies that it has competed successfully against rivals. This provides rational support for joining it, which again reduces the decision costs associated with any individual go/no-go choice about design effort.
There is yet another level: the individual knows that all the incentives operating on him also operate symmetrically on the other members of the institution. This edges us back towards the reciprocity incentives in the situation. I didn’t bring it up to rehash those but to point out that all these effects combine to reduce the overhead of making confident rational decisions about the payoff of individual design efforts from being somewhere in the knowledge-problem stratosphere to a level actual human beings can actually cope with.
Or, in other words: Yes, communication costs matter a lot. But so do loyalty, history, and trust.