Fortune Magazine Issue: October 12, 1998
Why We Don't Speak Esperanto
N. Gregory Mankiw
I have been wondering lately why I should teach my newborn son English. Everyone I know speaks English, but would Peter be better off learning a more sensible, mellifluous language, like maybe Italian? It is, I admit, a stupid question. But stupid questions can contain the seeds of great insights. This particular stupid question leads to the frontier of economic theory, as well as to the intellectual foundation for the government's case against Microsoft (a corporation that, I should disclose, has paid me to speak on its behalf in its current antitrust skirmish).
So humor me: Why are American children taught English? The answer is that everyone learns English because everyone else learns it. In this respect, language is a perfect example of what economic theorists call a network. In a network the benefit one person gets from using some good--in this case, English--depends on the number of other people using it. Other examples of networks include the telephone (the more people who have them, the more people I can reach with mine) and the VCR (the more people who have machines with the VHS format, the more tapes are made available in that format).
Networks fascinate economic theorists because they don't fit neatly into the standard model of how markets work. In most cases, economists are defenders of unfettered markets. People left to their own devices, we argue, will typically achieve an outcome that is good for society as a whole--the vaunted invisible hand.
In the case of networks, however, this laissez-faire logic doesn't seem to work. It is easy to imagine that people might get stuck with a bad network that once established is hard to replace. Parents deciding what language to teach their children, for instance, don't really have much choice. How else can we explain why the Japanese keep speaking Japanese when less complicated languages are available, or why we aren't all speaking Esperanto--the "ideal" language that was supposed to become the world language?
For a while proponents of the new economics of networks pointed to what seemed to be a compelling example of the problem--the QWERTY keyboard. As the story goes, this arrangement of letters was originally designed to prevent typists from jamming the keys on early typewriters. Despite the availability of superior designs, and the fact that jamming keys is no longer an issue, QWERTY remains the standard. This, theorists argued, was a network-driven market failure: People still type on this inefficient keyboard just because that's what everybody else does.
Yet when economic historians examined the case more closely, this story turned out to be less fact than fiction. The claim that other keyboard layouts substantially increased typing speeds was based on a study by August Dvorak, the designer of an alternative layout who had a financial stake in that conclusion. Other studies have shown that the QWERTY design is actually not so bad.
This debate over networks, keyboards, and market failure might seem like arcana only economists can love, but it is having a profound influence over public policy. Many academics who have written about the theory of networks have worked for the Justice Department and other federal agencies. A frequent claim is that computer operating systems are like languages: Once a standard becomes dominant, it is practically impossible for anyone to consider an alternative, even a better one. The only difference between English and Windows, the argument goes, is that English is free.
To some network theorists, these arguments justify government involvement in the software industry. Yet even these theorists are unsure what the government should do. If the government thinks it should set software standards, why not really make the world better and require Esperanto in schools? Moreover, networks do manage to evolve on their own--witness the changeover from LPs to cassettes to CDs. How this happens and how the government might lend a helping hand in the process remain open issues.
Despite our scant understanding of how networks work, expect this new theory to pop up not only in the Microsoft case now winding its way toward trial but in future antitrust cases as well. Is it possible, given the subtlety of the theory and the ambiguity of the evidence, that government policymakers will apply these ideas in a way that improves society as a whole? Or will any attempt to do so be overrun by ignorance, hubris, and politics as usual?
Now, there's a stupid question.
N. Gregory Mankiw is an economics professor at Harvard and the author of Principles of Economics.