[ademarre]

You can trick people, and possibly even search engines. I've wondered if blackhat SEOs could abuse such URLs to discredit content on competitors' sites.

I believe it can be a negative signal when sites stuff too many keywords in their URLs, especially if those keywords aren't relevant to the page's content. A server accepting arbitrary URLs is in a way blindly sanctioning loaded URLs.

Granted, Google's algorithms are surely very sophisticated in this regard, but fighting web spam is hard.

[MartijnHoutman]

Surely, a correct canonical URL will prevent this from happening.