This article on good URL design has some good “don’ts” and many “do’s” but seems to leave the answer as “use a date.” e.g. http://domain.org/2007/09/11/abba
I get that. I really do. It makes sense if your only goal is ensuring a URL to a resource still works 2 years, 10 years or 100 years from now. Anything else can change but the date of creation can’t (and if it does then you are probably into Creative Accounting too.)
Of course if time travel finally happens then we are a bit screwed but hey, there will be more important problems (don’t sleep with your grandma OK) than URL structure.
A date still bugs me though. Great for dependability, poor for discoverability. Sure, it helps you find all resources created on a given day or month or year but how do you find all resources created by Bob in Management?
I often look at the URL structure in sites and am able to figure out other related articles by the subject in the URL. But subjects in URLs are bad for long-term URLs. I experienced that first hand.
So how does one design URLs that are dependable but also discoverable? I have a half-baked realisation in my head that there is a difference between a URL to a resource and a URL to a “listing” page e.g. http://flickr.com/photos/tags/http. Sadly I see that the first result in that page is http://flickr.com/photos/daneelariantho/1345898949/ which has the username in it. I’d think http://flickr.com/photos/1345898949/ would be better as usernames can/should change.
It is still rather hard designing good URLs. Hopefully my name never changes and I can retain this domain. But already I see a problem in that I have put “journal” into my URL. What happens when I grow sick of journal and change it to blog? Shouldn’t be part of the URL I guess.