How I would think of traditional publishing

This post about (yet) a(nother) bestselling author who is making more money self-publishing went up on Passive Voice last week. The thing that really piqued my interest was a comment by Edmond saying:

You have to wonder whether the career path of a writer will become.
Self publish => Get attention => sign publishing contract => build following => return to self publishing taking readership with you.

That was interesting to me in light of Amanda Hocking's experience--she's definitely building a new audience, as she had hoped, while at the same time keeping a substantial number of self-published titles available. Instead of following Edmond's cycle, however, she's multitasking--since she has so many titles out, she never had to cycle out of self-publishing. She can advance on all fronts.

It doesn't change my very, very strong opinion that if you are a new writer and you want to get published, you should self-publish. (Like, OMG, you should self-publish. Don't be a fool.) Traditional publishing seems to work best for people who basically don't depend on it--they've got self-published titles to fall back on if things go to hell, and they've got self-published titles poised to gain sales if things go really well. (This necessarily means that they don't sign traditional publishing contracts that restrict their ability to write and publish other books.)

I think the best way to think of traditional publishing is as something akin to a Next Step, like doing an audiobook or creating a store on your Web site. You don't start out by seeking a traditional publishing contract; you start out by self-publishing an e-book. Traditional publishers could help you reach people who you might not be able to reach on your own, like those who insist on buying hardcover books at the airport instead of downloading a Kindle app onto their phone, just like an audiobook could help you to reach a previously-inaccessible niche.

But it's an expansion of your distribution model; it's not your only distribution model. It's not even your first or fundamental distribution model. It's something you bolt on, not something you use as your foundation. And for God's sake, don't put all your eggs into it--the future does not look bright for those people.