When URLs go bad
30 November 2007, 22:16
We all know what makes a good URL. Its got to be hackable, meaningful, short, and most importantly its got to never, ever, change. But what happens when people get it wrong? <voice type="tvhost">Join me in a journey around the world of bad URLs...
http://www.ft.com/cms/s/0/4dc9d2ba-9f2e-11dc-8031-0000779fd2ac.html?nclick_check=1
CMS? Why do I care that a page is generated from a CMS? Does that identifier really need to be 36 characters long? Are the FT expecting to generate 36^36 (or more?) articles/pages in its lifetime? Just six characters (6^36 = 2.1bn) would probably cover it. You could push the boat out, go wild, and do seven (7^36 = 78.3bn).
http://prettymuchanyblog.com/archive/2007/12/31/the_blog_post_title/
This one has been hacking me off for a while. Many blogging systems put the permalinks to posts in an archive directory by default. To me, archiving implies something of historical interest that has been set aside from the main bulk of the posts. The URL could be used as an identifier of what we're actually expecting - a blog post, so why not make the link yourblog.com/post/2007/12/31/the_blog_post_title/ ?
BBC URLs are usually of the beautiful form bbc.co.uk/[whatever], but this disgraceful effort breaks all the rules (I've had to break it in half to fit the page).
- Stupid domain name - nobody cares where your server is!
- Technology specific - cgi-bin, cfg, php (twice!), can you fit any more technology identifiers in there?
- Ridiculous number of variables in the querystring, are they really all needed?
Why this URL can't be the following, we'll probably never know: http://www.bbc.co.uk/iplayer/support/finding_programmes/
http://www.livedepartureboards.co.uk/ldb/sumdep.aspx?T=DMK&S=VIC
National Rail Enquiries are renowned for having a crap website, even with Matt Somerville seemingly giving them free consultancy and showing them how to do it properly with the beautiful traintimes.org.uk. Checkout this URL porn: http://traintimes.org.uk/VIC/DMK, for the same page as above.
So, it seems what the National Rail people have tried to do is make a shortcut to get live departure information by registering and using livedepartureboards.co.uk. All very well, I know I like to check the train times before I walk out of the office into the rain. Problem is they then appear to have bodged it onto their nationarail.co.uk site. The result is repeating what site your on (the "ldp" directory - which I assume means Live Departure Boards) and a messed up interface. Users have been, rightly, drilled to realise that a domain name represents one site, which makes the "National Rail Enquiries" branding seem strange. Even worse clicking on "Home" takes you to the NRE homepage. The LDB homepage is a crap HTML redirect.
http://www-935.ibm.com/services/us/gbs/bus/html/bcs_financialmgmt.html?re=bcs_home
http://h71028.www7.hp.com/enterprise/cache/19054-0-0-225-121.html
Are the www-936 and h71028.www7 parts really necessary? It would be much better if these IT giants put these systems behind some clever load balancing system so that everything gets served from www.[companyname].com. Much less confusing, much more friendly.
RSS Feed
about
Jason is a web developer living in London, working for Google and Ferrago Ltd.
links
me
- Flickr (photos)
- Facebook (social network)
- del.icio.us (bookmarks)
- last.fm (music)
- Linked In (business social network)
- Twitter (microblogging)
- Google Reader (bookmarks)
work
- Ferrago Ltd (company)
- play (company's product)
- MetaWeather (company's product)
colophon
- Built in ASP.net using a Mac and TextMate
- Hosted on Windows, with IIS & MySQL
- Activity feed by FriendFeed
