Friday, October 9th, 2009
This is really awesome; I’ve long thought that applying usability studies to APIs like we do for user-interfaces can yield real results and make development easier.
A natural question is whether developers can actually use microdata in practice and understand it. On the WhatWG blog Ian Hickson writes about usability work Google has been doing around the HTML5 microdata proposal.
We first created three different variants based on the original microdata proposal:
- One based on what the spec said (documentation)
- One trying to put types in an explicit
itemtype=""attribute and moving “about” to
item="", and replacing
itemfor=""with just having multiple
item=""s with the same name (documentation)
- One trying to remove types altogether and using
itemas a boolean attribute. (documentation)
Each of the tests involved having participants read the documentation and try out various examples, comparing and contrasting the approaches to see how developers responded.
The conclusions are basically that the attribute name ‘about’ should be replaced with ‘itemid'; that participants were comfortable with the idea of types, renaming ‘item’ to ‘itemtype'; and that people were confused by the idea of scoping requiring some small changes there. Another result is that people don’t have any problems dealing with URLs as property names, and that the length due to using URLs didn’t bother people.
The blog post doesn’t have it, but it would be nice to have a final sample that incorporates all the changes to show everything.
Posted by Brad Neuberg at 7:30 am