Denver, I get that clocke thinks broadcast marketing doesn't work on the Web. I just think he's wrong, or rather, that the places where Gonzo works are more limited than they seem at first glance. Both you and Hernani have suggested that consumer packaged goods, e.g. Coke and Jell-O, somehow "don't count", but CPG advertising is the majority of all advertising, so a revolution that doesn't change CPG doesn't really change the advertising landscape as a whole.
The central dilemma of Gonzo, from my point of view, is scaling. You've suggested that every member of a 100 member micromarket is also a member of any number of other micromarkets. (This is the 'Small Words' structure, most rigorously described by Duncan Watts in a book of the same name, and made famous by Malcolm Gladwell in 'Tipping Point'). So, starting from one person in one micromarket and the idea that "any number" is, say, 10, you are one hop from 1,000 people ("my friend says..."), 2 hops from 10,000 people, (A friend of a friend says...") and so on.
Now its easy to see, a la the old Clairol ads ("...and she tells two friends...") that this is in theory a fantastic way to spread the word about something. It is also easy to see that the actual effect is much more limited, for 3 reasons:
1. Not everyone in a micromarket will become an evangalist
2. The ones that do will be of varying effectiveness
3. As the group grows, the number of overlaps, where you reach people who have already been converted, will grow
These three characteristics make the shape of word-of-mouth a logistic curve (like a stretched out letter 'S') rather than a simple ever-upwards curve of explosive growth. In other words, for any infectious spread (viruses, memes, protocols, whatever) there is always an upper boundary.
So, to reach a million people (and remember, there are products used by a billion people, so a million is setting the bar pretty low) you would have to gather *everyone* four hops from you. Now what are the chances that 1 million people are going to become passionate about something because you are, which is to say because their friend's friend's friend's friend likes it? Nil, of course -- you can't even get all thousand of your putative one-hop friends to agree on anything. And once you start introducing inefficiency and non-takeup into the equation, Gonzo starts looking to me like a pretty resource-intensive way (esp when counting time as a resource) to get a million customers.
So let me propose a hypothesis: There are no products other than software that have ever reached a mass audience without some use of a broadcast (which is to say one-way, one-to-many) medium.
The same question, put another way: if uni-directional models are so bad, how come "Gonzo Marketing" is a book?