Underscores vs. hyphens in URLs

Question

I'm trying to come up with a page-naming convention so that each page in a web application I'm developing has the human-friendliest URL possible. The current URL structure of the web application is:

http://foo.example.com/webapp/pageName

The only part of the URL that I'll be changing is pageName. The application is a basic CRUD system and has a UI that consists mostly of forms and lists. There are both object- and task-oriented page names -- e.g. addEvent, eventManager, eventReview, imageUploader, editEvent, addItem, editItem, bulkAddItems, bulkEditItems, etc. Given those constraints, I'm weighing a few different URL styles for these page names:

1. The current camelCase convention, which produces names like those listed above. This is simply a vestige from prior development; the pages were named like variables, without much thought given to their UX impact.
2. Hyphenation, producing page names like:
   A. http://foo.example.com/webapp/add-event, http://foo.example.com/webapp/event-manager, or
   B. http://foo.example.com/webapp/Add-Event, http://foo.example.com/webapp/Event-Manager
3. Using underscores, producing page names like:
   A. http://foo.example.com/webapp/add_event, http://foo.example.com/webapp/event_manager, or
   B. http://foo.example.com/webapp/Add_Event, http://foo.example.com/webapp/Event_Manager

Stack Exchange and Smashing Magazine use a style somewhat similar to 2A, e.g.:
- http://ux.stackexchange.com/questions/9093/senior-usability-and-navigation
- http://www.smashingmagazine.com/2011/07/25/email-is-still-important-and-here-is-why/

Wikipedia uses style 3B, e.g.:
- http://en.wikipedia.org/wiki/Extreme_Programming

I'm leaning toward style 3B. It emphasizes the page name with capitalization, and the underscore seems like a clearer and more intuitive word-delimiter than the hyphen.

What do others think? How much do these different URL styles lend themselves better to UX depending on what type of web application (blog, Q&A site, encyclopedia, what have you) they're being used for?

Answer 1

If you're looking for the most SEO friendly URLs that are also human readable, then I would recommend using all lowercase, hyphenated URLs, as that is what Google recommends in their Webmaster tools documentation. However, if SEO doesn't matter for your web app (if, for instance, it all lives behind a login requirement), then you can use whatever deliminator you want.

I would strongly recommend that you avoid camel case URLs unless you are prepared you accept the URLs without proper capitalization as well... nothing is more frustrating to a user than being told to go to www.example.com/FooBar.html and getting a 404 when they instead type www.example.com/foobar.html (and yes, this really happens... some of the older systems at my workplace do this, and it's obnoxious!)

Answer 2

The parts of the URL below the root domain name are generally case sensitive on UNIX web servers and case insensitive on Windows web servers. Some unix webservers use the Apache mod_speling module to look for an alternative case mix if the given URL is not found.

To reduce likelihood of a '404 page not found' error you can create a 404 page yourself which can lower case the unfound URL and check for its existence and if that is found, redirect the page. That's a little feature no-one need ever know about.

But that little tweak is going to be much harder if you have CamelCase or MixeD_CasE in your URLs, because it easy to automatically lower case everything - not so easy to un-lowercase everything!

Go for 2A. Make life more flexible from day 1 and reduce chances of a headache later.