On Fri, 22 Nov 2002, C Falconer wrote:

> Cool idea - but the user doesn't know what site the link is to and can't
> judge the usefulness of the link.
>
> http://makeashorterlink.com/?O2AD32682
> http://makeashorterlink.com/?P57D22682
> http://makeashorterlink.com/?V2BD22682
>
> Which of these is useful, which is strange, and which is just plain
> wrong?
...

Yes, the problem is obvious, it is a little bit like dialling a mobile
phone number without knowing to whom it belongs. However, they can
only create these short links by exploiting the hugely excessive
redundancy in "normal" URLs. By removing most of the redundancy, and
still coding it unambiguously, they get to the shorter links. The loss
of human intelligibility is almost unavoidable if you have a simple
piece of software do this.

Certainly, a human theoretically could do this much nicer, like change

www.canterbury.ac.nz/whateverunnecessarilylongfilenametheymightthinkof.html

[Nothing against UC! I just chose it because someone from UC brought
it up]

to      www.canterbury.ac.nz/toolongname.html    which is much
shorter, but has similar semantics of the filename. BUT, this would
require that you first check if that filename does not already exist,
and then, it would have to be the UC server itself who would have to
provide it, and not some external service. Of course, they could still
make

shorturls.com/uc.ac.nz/toolongname.html from it at an external service
called shorturls.com . But the translation of filenames, maintaining
the semantics of the name, and this under the boundary condition that
ambiguities are strictly forbidden, is anything else than trivial and
would at least require some fancy AI thing.

I do not say it is impossible, but it might cost a lot. The thing they
have done lacks beauty, but seems to be comparatively simple and
straightforward.

Cheers,

Helmut.

+----------------+
| Helmut Walle   |
| [EMAIL PROTECTED] |
+----------------+

Reply via email to