Very interesting proposal here... though I think having the consumer do an http request to get the "link" from HEAD is a bit too much to ask, for a few reasons:

* It takes bandwidth

* It's (presumably) before a pageload, and is blocking... the latency to load the original URL to get the "link" gets added to the consumer's pageload time.

* Now you have to deal with "timeouts"

Ideally, "link" should be used by the browser, and not the web service. By the way, other such auto-discovery systems are OpenSearch (first proposed by Amazon, and is pretty widely adopted) and FavIcon.

At any rate, I think tinyurl, and bit.ly, etc, are a pretty fast and easy solution at this point. They should improve their services by including a "title" attribute to the link they give you, which says the URL and/or page title it's going to.