openstack-doc-core team mailing list archive

[Anne Gentle <anne@xxxxxxxxxxxxx>]

This is an excellent paper... I wanted to share with you all.
http://blog.ninlabs.com/2012/05/crowd-documentation/

Quite a find for its relevance to OpenStack APIs! They studied Android, GWT, and
the Java programming language. In their words "a large established
API, a new and specialized API, and a less active API. The questions
they studied:

RQ1. Can we rely on the crowd to discuss an entire API on Stack
OverFlow? (not the entire but a good portion)
RQ1.1 Are APIs widely covered?
RQ1.2 What is discussed and what is not dis-
cussed? Are those API elements infre-
quently discussed also infrequently used in
practice?
RQ1.3 How fast is the crowd at covering an
entire API?

RQ2. What are the dynamics of a successful API community on Stack OverFlow?
RQ2.1 Who contributes?
RQ2.2 How does the crowd contribute?
RQ2.3 How many code samples does the crowd
provide?

Conclusion:
We have shown several sources of evidence that crowd
documentation exists as a viable process that can emerge
from social media sites, such as Stack OverFlow, for creating
software documentation. Documentation can emerge in
the form of questions and answers that feature many code
examples and discussions about using API classes and meth-
ods. The authors that contribute these items take distinct
roles in curating and maintaining the quality of questions
and answers.

I don't have other conclusions for OpenStack APIs but want to study
their paper further and the tool they made. This is highly relevant
research though.

Anne