[fallintothis]

Also, why does the URL need to be human-readable? URLs are for computers. The content is for the human.

Disagreed. I can't stand sites with ugly, incomprehensible URLs: even if they're just appending parameters ad infinitum, it's unnecessary drag on the process of editing, sharing, and typing the URL. E.g., something like

  http://www.amazon.com/Kindle-Books/b/ref=sa_menu_kbo0/187-1539918-5044028?_enc
  oding=UTF8&node=1286228011&pf_rd_m=ATVPDKIKX0DER&pf_rd_s=left-nav-1&pf_rd_r=05
  F7VG5R8T4MBDH7QQ6N&pf_rd_t=101&pf_rd_p=328655101&pf_rd_i=507846

is, near as I can tell, mostly equivalent to

  http://www.amazon.com/Kindle-Books/b/ref=sa_menu_kbo0/187-1539918-5044028?node
  =1286228011

and could probably be made even simpler. Of course sites need to carry some amount of incomprehensible information (YouTube's video IDs, reddit's story IDs, marco.org's blog post numbers, etc.), but shoving as much as you can into the URL isn't the way to do it. I type specific URLs like en.wikipedia.org/wiki/Topic and news.ycombinator.com/newest all the time. If the site author neglects to even have this sort of basic functionality, I tend to consider it a bad interface.

[blasdel]

The context of a hyperlink is meaningful, but the text of the corresponding URL is of little relevance.

'Clean' URLs tend to be a REST anti-pattern, as they are tied up with the notion that the ability to edit an known URL to refer to a different resource constitutes an API, much less a REST-ful one. HATEOAS Motherfuckers!