Here's somewhat of a thought experiment. It's mostly the result of my experience clicking on a driving directions link: although I normally use MapQuest or MSN maps (and thus they remember a number of addresses that I commonly use), that link took me to Yahoo maps instead.
That seems a common and simple enough experience, but it appeared to me as quite similar to the one-click subscription problem. The generalization is about binding a piece of data on a web page with a service provider from the user's choice.
That piece of data could be an RSS feed (a feed: url), an email address (a mailto: link), a search box, an address to a physical location, or a reference to a book. In turn the service is the aggregator, the email client, a search engin, a mapping/direction service or a bookstore/catalog.
There are multiple problems with the current system:
- Many hardcoded service providers all over the web: a Google search box, a MapQuest link or a Bloglines subscription icon. This is inconvenient for the user whose experience is better at his preferred provider, because of the quality and personalization of the service.
- URI schemes like feed or mailto are mostly taylored to work with desktop-based service providers. That's not the case for all schemes and browsers though: IE can be configured to handle mailto urls using the Hotmail online service for example.
- New URI schemes are not very well handled: their usability is best if you're browser already knows about the service or scheme and if already have a provider installed.