Fred Wilson is writing about Free-vs-Pay services.
My main comment is that there are cases where the dominant company has a for-pay service. Amazon's web services platform, Dropbox and Rackspace virtual servers are the ones I depend on the most.
What they have in common is that these systems provide a low-level function in which the user creates their own service. They are platforms, even though Dropbox has a fairly high-level entrypoint. I have been coming up with new applications for it constantly. They're not hard to implement, but they take your mind through some wonderful little twists. I use it mostly for server management, but like everyone else we use it in our family to share photos and we're converting the family scrapbooks slowly so they're in a Dropbox folder structure. It's low-level, but until this service came along there was no neat way to do it. And of course we pay for it, happily.
People who think Amazon and Rackspace are just for developers are wrong. EC2 for Poets is something any user who mastered MS-DOS could handle. It could be even easier with customization. I tell people that running a server is almost exactly as complicated as running a laptop. Slight differences like one goes in your knapsack and the other one is always on and always connected. But the software is identical to the software that runs laptops. People are surprised when I say that, but it's true.
Services must be for-pay when this kind of flexibility is required. I don't have any doubt that a for-pay version of Twitter would work, in the same model as Dropbox. I've often felt that Amazon should have a simple notification service that does everything that Twitter does and nothing more. Huge explosion of innovation would come from that, because there are so many developers who eat Amazon APIs for breakfast. I happen to be one of them.
I've gotten so much value from the Route 53 API, for example. It's made it possible for me to do things I only dreamt of before.
Twitter, on the other hand, inspired similar dreams, with an uneasy feeling because I was concerned they would do more or less what they ended up doing. Squeezing out a commercial platform and killing off the parts that I loved. I end up with something useful, the Twitter user interface, with none of the geek love. What if instead of hiring a marketing guy to run it they had hired an architect with a sweeping megalomaniac vision. Some truly great stuff would have happened. That opportunity still exists.
But for me to buy into a for-pay Twitter-like service, I would have to know the company pretty well. I don't have absolute faith in Amazon, Rackspace or Dropbox, I've had issues with two of the companies (Amazon and Dropbox) in the last couple of years. Non-trivial ones. But net-net I go ahead and build on their services. And I like the deal. I can't believe how little I pay for them, but I'm glad I do pay.
BTW, I'm not pointing to Fred's piece because I don't want to be in the chorus for his piece on Techmeme. I'd like to get equal billing. (Update: it didn't work, they put me in the chorus anyway. And of course they didn't link to my piece about User/VCs which imho is much more to the point than free-vs-pay. The issue is how we fund development in tech. We're only getting a slice of it supported by investment, so of course, that's the part that works.)