@cackhanded is right. class is used on visible content.
Thus HTML authors & designers learn & use class already.
@snookca @Malarkey, #microformats prefer simplicity, familiarity, & learnability.
For #microformats2, we aimed to make them even simpler than classic #microformats, as well as more compatible with existing class values.
In addition the data-* attributes are meant for private page-specific meaning, and explicitly not for shared or cross-site data.
http://microformats.org/wiki/HTML5#data_attributes
There were attempts to use new attributes in HTML instead, with both RDFa and microdata, and the result was more complex designs using *multiple* new attributes, each, which were more work to learn, and harder for authors to use:
https://indiewebcamp.com/irc/2014-06-21/line/1403415993000
And as @cackhanded says, easy parsing too, especially with the microformats2 parsing specification and now numerous open source libraries in nearly every language:
http://microformats.org/wiki/microformats2#Parsers