If you just give your core product away for free then users will only switch if they need the extra features of the paid plan. You can still use 'free' for marketing, because total conversion rates with a free plan will probably be higher, just don't expect these 'free' customers to switch if they don't have to.
In the meantime, though, my main Google account has 20gb for $5/year, and for the equivalent-lowest of 50gb/$100 I can get 400gb, which will also store my emails and photos. I can back up any email account on my machine by dragging the messages to a gmail folder, and waiting a little while. Docs now allows 1gb any-kind file uploads. I can send my photos to Picasa (the reason I have the extra space).
I'm a massive word-of-mouth spreader of Dropbox though, especially as I'm the Geek of a few families + their neighbors. I've probably gotten them more business than what I cost to host.
This way, people who don't bookmark very much can live with 1000, and people who eventually will use more than 1000 need to subscribe...
I think it's a very good plan if your product is structured in a way that a user will eventually hit the limit, but I think Dropbox was not a very good example. If I hit the limit, I'll just delete a few files I don't need. I see your point, though, and I think you're mostly right.
NB: Truth be told, I run batch files via task scheduler for backing up certain folder to dropbox, but I doubt the typical consumer does.
Dropbox is using my money to subsidize the free users, I'm not getting 10$ of value out of it.