Apr

28

Market Extent and Self-Selection

Filed Under Peer Production | 

In an earlier post (Peer-Production and the Extent of the Market), I had suggested that perhaps peer-production efforts did not necessarily represent a non-market phenomena, but may instead represent an adaptation to insufficient market extent. A weakness in that argument is the fact that market size is clearly not an issue for the notable peer-production efforts like the open-source Apache web server or Linux operating system. The April 2nd EconTalk Podcast has an entertaining discussion of specialization and market extent, with some ideas that may be useful in thinking about how the “top-tier” peer-production efforts differ from the rest.

A key point that postcast host Russ Roberts and guest Mike Munger raise is the distinction between David Ricardo’s and Adam Smith’s perspectives on specialization. The Ricardian view revolves around comparative advantage: Two entities (nations, individuals, etc…) will be better off if they specialize and trade, even if one is better than the other at all productive tasks. Smith’s view is more focused on opportunities for specialization based strictly on the size of the market. From the podcast summary:

When you have one hunter, he has to do every task for himself. When you have 100, all equally skilled, one hunter can open a service making breakfasts and lunches to go. All the hunters can be equally skilled at hunting and running a restaurant but when there are a lot of hunters, it becomes profitable to run a take-out food service that can’t be profitable when there are only a few. So 100 hunters are more than 100 times more productive than one hunter because of the economies of scale in making the breakfast and lunches, freeing up more time for the hunters to be productive in the field. Four times as much resources can mean more than four times as much output even without Ricardian reasons for specialization.

Two factors are important in this scenario. First, the market is large enough to support specialization. Second, the hunter who switches roles is able to discover the fact that he can successfully switch. He faces uncertainty that may only be remedied by experimentation. Experiments succeed if the market proves large enough and his skill and productivity are adequate. His customers must also overcome their half of the knowledge problem and discover that by buying lunch rather than making it themselves they free up time for more hunting.

It is sometimes argued that an advantage of peer production is that it makes it easier for individuals to self-identify for a task. This is certainly plausible, but it has always struck me as being matter of degree. In a free society, we all essentially self-select for our productive activities. This does not guarantee success, of course, as bad luck, bad choices, shifts in the structure of the economy, etc… may leave us in unsatisfying jobs (or worse). Blogs, wikis, open-source software projects, and other peer-production efforts are buffeted by the same forces, and differ primarily in that the stakes are generally much lower.

The self-selection argument, however, seems to be based on an essentially Ricardian model. As such, it assumes the presence of a large market: Not only is there enough demand for something that people will try to produce it, but enough people will try to produce it that comparative advantage among them will be revealed. In the world of open-source software, there is no doubt that this sometimes happens. Large projects like Apache and Linux are driven by extremely talented, and specialized, developers. The market for these projects is large and transparent.

Below the level of Apache, Linux, and the like are countless smaller niche efforts, undoubtedly comprising the bulk of the 147,000 projects currently registered on SourceForge. It seems plausible that at this level, developers are operating in a more Smithian world of specialization. In the worst case, they are the lone hunter in the example from the podcast, forced into self-sufficiency. Where market extent is greater but knowledge of the extent is limited, comparative advantage need not matter: A suboptimal implementation of a tool is better than nothing. Self-selection, in this case, may not matter with respect to skill, but only to initiative in actually trying.

Comments

Comments are closed.