The problem: URLs, especially those based on a back-end database, can become long and unwieldy. For instance, http://www.amazon.co.uk/s/ref=nb_ss_w_h_?url=search-alias%3Daps&field-keywords=ben+jeapes&Go.x=0&Go.y=0&Go=Go gets you the complete works of me as known to Amazon, including
Online Information 93: 17th International Online Information Meeting Proceedings London 7-9 December 1993 and Online Information Hong Kong: The 2nd Asian Information Meeting, which resolutely refuse to sell out no matter how hard I try.
But it doesn't really trip off the tongue, does it?
Some magazines, like Focus, gladly use tinyurl already. Organisations with a more proprietorial attitude still prefer to use their own URLs, because no matter how unwieldy (say) the above URL is, it still points you at the Amazon site. A tinyurl URL could take you anywhere, and you wouldn't know until you got there.
Ben's solution: why can't the two be combined? Why can't there be a package sitting on a web server that encodes the URLs in something short and alphanumeric - just five characters long would give you a range of 60466176 entries, which might not be much by Amazon standards but will do fine for most of us - and puts them after the root URL? Thus in bencode the above could be rendered www.amazon.co.uk/abcde. Companies don't have to use a third party, they control their URLs and the customer goes straight to their web site.
This seems so obvious that I can't believe someone hasn't tried it, and that being so, it seems so useful that the only reason you don't see it everywhere is that it doesn't work. So, what's wrong with it?