As some will know, I wrote about the future of Google for the January 2005 issue of Technology Review.
In that article, I argued that the search market will become enormously larger and more diverse than presently, encompassing many forms of personal, internal corporate, for-sale proprietary, and public data stored in a wide variety of systems, ranging from PCs to iPods to corporate servers.
Then I asserted that Googles leadership position remains fragile, given the absence of barriers to switching and an impending challenge from Microsoft. In particular, the emerging requirement for search interoperability across many data formats and systems, and the diversity of search-related innovation, imply that search engines must become standardized platforms with open interfaces.
I concluded that unless Google created such a platform and made it generally accessible by providing Application Program Interfaces (APIs) to its search engine, it would both lose much of its potential market and render itself vulnerable to a Microsoft attack.
Conversely a proprietary but open architecture based upon public APIs would enable Google to attract more users, create interoperability across many systems, and generate switching costs.
In part, the article proved to be prescient. On March 17, Google announced an initiative to create developer communities based upon APIs, some of them newly released. While Google still has not made available an API for its core search engine services, it may be moving in this direction.
The article also generated several criticisms. One of these, which with some ironic license I have called the New Age Theory, runs as follows. With the Internet, we have entered a new and better era, one in which openness will triumph over narrow, proprietary self-interest. Because the Internet is a vast commons, the greatest fortunes will be made by those who are most open. Moreover large users, by now fully educated about the dangers of lockin, no longer tolerate proprietary control.
Alas, while this vision of Eden is very seductive, the evidence doesnt completely support it. On the Web, most users and advertisers are dispersed and unorganized, and large websites behave accordingly. Proprietary control is alive and well on the Web; indeed, the only thing more proprietary than a proprietary standard is a closed system with no interfaces at all, which is essentially what Google has traditionally operated (as have Yahoo and AOL). Amazon and EBay both have, in essence, open but proprietary APIs. None of the most successful firms on the Web have nonproprietary architectures that others can freely clone.