Most Internet pundits use television as the dominant metaphor when explaining the Net: words like "channels", "shows", and "eyeballs" feature prominent in many analyses; ideas like push technology and ever-increasing multimedia flash are given credence despite being losers in usability terms. Television is used as a metaphor for two reasons: it is indeed the most powerful medium of the past, and it's all most media and advertising executives know how to deal with.
Well, most Internet pundits are wrong. The Web is not like TV. Most fundamentally, the Web is a user-driven narrow-casting medium utilizing low bandwidth with high flexibility, whereas television is a broadcast mass-medium utilizing high bandwidth with little flexibility. Because of the lack of flexibility and customization TV has to rely on production values to capture an audience. The Web will have much lower production values for the next many years; the only hope is to play up the strengths of the Web, which are the ways in which it is different from TV. The Web can be used to integrate the Internet and television to give added value to TV viewers, but such systems will not realize the true potential of the Web.
I believe that the telephone is a more fruitful metaphor for thinking about the Web. This is not to say that the Web will be exactly like the telephone: let's hope not, since telephony has horrible usability characteristics. Also, no single metaphor can explain all aspects of a phenomenon as powerful as the Web: Jerry Michalski shows how seven other metaphors can be useful for analyzing the Web (in an article that is no longer on the Web as of June 2002 - too bad; it was good and would still have been worth reading five years later) .
Telephony is fundamentally a narrowcast medium: 1-to-1 communication along a rather low-fidelity channel. The fact that telephones are so popular despite their poor audio quality is evidence that content is king, even here: what matters is who you talk to and what they say, not whether they sound exactly the same as if they were in the same room as you. Also, videophones have been failures so far, partly because higher fidelity is often unwanted (you can place an important business call to a client while you are very fresh out of the shower). The Web will also remain cursed with low bandwidth for the foreseeable future, so it would be better if we simply accepted this fact and designed sites that worked because of their content and not their production values.
Being a 1-to-1 medium is a characteristic of the Web and a reason the traditional advertising model will fail on the Web. Mass marketing is inappropriate when you can use customized value-added relationship marketing. Rather than blasting banner ads all over the place, companies should build customer value into their own websites and make it easier and more attractive for people to do business right on the sites. Direct marketing is a much better model than TV commercials: what matters is not how many people see your stuff ("eyeballs") but how many react on it (to contact you, to buy, or whatever else you want from them).
Unfortunately, the telephone metaphor also leads to telemarketers who invariably call during dinner with irrelevant offers. The Internet has spam, and the faster we get rid of it, the better. It might be possible to borrow from regulatory ideas in the telephone world (many countries outlaw most types of unsolicited calls), but it could also be that the Internet will pioneer solutions (like filtering and "you-have-to-pay-me-to-get-through" tokens) that could be brought back to help the telephone world.
Another fundamental property of telephones is that calls are initiated by the user at exactly the time they want to. Obviously, this is only true for the caller, whereas the callee is inconvenienced by getting a call as a "push" medium. Consider which of the two parties is better off and you will see why pull is better than push. On the Web, of course, the second party to the conversation is a computer, so getting hits at an inconvenient time doesn't matter (assuming that a sufficiently powerful server is used).
In general, the telephone is interactive, as is the Web. It's not a matter of giving people a huge, packaged chunk of content (as a TV show); instead, a telephone call is a back-and-forth exchange between two parties, where each turn depends on the information received from the other end.
A final difference is that everybody is a publisher on the telephone. On TV, of course, only a select few get to go on the air. Even though most interest is focused on the large websites, I think the Web as a whole will derive more value from the combined effect of millions of smaller specialized sites (even though each single one will obviously have smaller revenues than a big site).
In thinking about your website or about new Web technologies, I encourage you to think in terms of telephony. Metaphors are dangerous if taken too far, but they are helpful in finding perspectives and analogies that can take us farther than the immediate surface characteristics of our work.