So if your objection is that my language implies the concept “solely”, then yes, let’s change my wording. Here is my new statement, changed words in italics:

“I do not want my music retrieval and discovery algorithms to be dominated by by the millions of individual posting (and click) links in order to help determine which musicians and songs offer content of value.”

And I proffer a rephrasal of Whitman’s statement:

“If we let our user experience be dominated by collaborative filtering to discover music, the popular artists would eat the unknowns alive.”

Does that change the fundamental argument? No. It should be more clear than ever that I still do not want Google to end up doing to music what they did to the web. What they did to the web is rely more on collaborative filtering signals (mass query-click graphs, PageRank, etc.) than on individual signals (what?! still no relevance feedback?!) And so my web search experience is dominated by collaborative filtering. I don’t want my music search experience to be dominated by collaborative filtering. I want more attention to be paid to individual signals, and less to mass signals. I’m not saying Last.fm is perfect and couldn’t be improved. But it is a strong step away from the web-like user experience.

You’re correct; it’s not just that it’s collaborative filtering. It’s how you apply it. The way Google applies it is non-exploratory; it creates a single line, a non-interactive ranked list. The way a company like Last.fm applies it is more exploratory. At Last.fm, you can pivot around, starting with a band and going not only to similar bands, but to similar listeners. Should you select a particular listener, you can the get an “expert” recommendation on other bands and songs (rather than a collaboratively filtered recommendation) based on just that one listener. What Last.fm is still missing (though I know folks who work there and are aware and thinking about these issues) is a way of adding yet another pivot point based on content methods (signal analysis, etc.) So that part is coming. But at least Last.fm doesn’t try to make all your decisions for you, and present you with a single, linear list around which it is difficult to pivot and explore. It interactively gives the user a choice.

So I still do not want music search to be done the way web search is done: A single ranked list, with little ability for interaction, pivoting, exploration, feedback, etc. I still do not want Google to do to music what they did to the web.

2. Music distribution is currently almost totally based on popularity or money. Which one depends on your flavor of cynicism.