Why Google Seemed Impossible

At the Xanadu project during the years 1989 through 1992, sometimes the question would come up about the feasibility of (what we would today call) search engines. At other times, we would discuss possible revenue models for the future world of hypertext. At that time, I argued (convincingly to many, I regret) that search engines were impossible, and that advertising was not a workable revenue model. In both cases, the core argument was simple, and, I felt, solid:

For a company to run a search engine, it would have to individually acquire disk space that was proportional to the total disk space used by the sum of all other players. If there were multiple competing search engines, they would each need to do this. This seemed obviously impractical. Although I knew storage would continue to get cheaper at an exponential rate, this didn't seem to me to invalidate the argument, since it would also be cheaper for everyone else.

When advertising was proposed as a revenue model, I argued for the inevitability and unsolvability of what is now known as "click fraud", as well as ad-blocking services such as today's Adblock. The one form of advertising I thought was sustainable was product placement ads, where the advertisment is woven into the content, such as the famous "Reeses Pieces" in the E.T. movie. (If I were being paid for this statement by Hershey or Steven Spielberg, which I'm not, then it would serve as a good example of itself.)

Some links

The following list of links were gathered to contribute to a discussion about how wikis could be improved, by adapting insights from Xanadu and related work. I'll leave the author unattributed until checking with him.


A short, easy-to-read article about how information would flow in a collaborative Xanadu medium, published in Unix Review in 1990:
Hypertext Publishing

A paper on the societal implications of the Xanadu feature set:
The Open Society and its Media

Impact of the Xanadu approach on the evolution of knowledge, by K Eric Drexler, inventor of the concept of nanotechnology:
Hypertext Publishing and the Evolution of Knowledge

A technically successful attempt to bring the bidirectional, fine-grain, overlapping links of Xanadu to the open Web. Though it generated a lot of interest and a lot of people
experimented with it, it was ultimately socially unsuccessful (too few enduring users to bother to maintain it) for ease-of-use reasons that I think can be overcome in the more controlled context of a Wiki with modern AJAX techniques:
CritLink: Public Web Annotation
CritLink Mediator

Automated mapping:

CritSuite: Support for Mediated Content on the Web

A more recent description of why we need Xanadu, especially since the Web has taken off, by Ted Nelson, inventor of the term "hypertext":
Xanalogical structure, needed now more than ever: parallel documents, deep links to content, deep versioning, and deep re-use

Are We Smarter Yet?

Like some early television pioneers, some early hypertext pioneers -- myself included -- were motivated by high hopes that the new media would make us smarter. That it would, in Doug Engelbart's famous phrase, "Augment the Human Intellect". Fifteen years into the web, I reflect on ways in which the modern world of hypertext has exceeded our hopes, on ways in which it disappoints, and on what steps to take from here.