[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Personality traits of (Debian) developers

also sprach Gabriella Coleman <biella@gmail.com> [2007.09.25.1415 +0100]:
> intelligence to the wider community. It is chapter four of my
> dissertation, which I will pass along to you if you would like.

I have your entire dissertation; thanks for the pointer!

> Also there is a long literature on hacking which notes the
> semi-obessive traits of hackers, which I would say also pertains
> to DD's. Turkle and Levy are two folks that come to mind.

Yeah, I am aware of those. What I am trying to do is actually argue
my way out of having to analyse other groups as well, so therefore
I am trying to narrow down the subject group a bit more than
"Hackers" :)

I guess the same could be said about Benkler: I am not talking about
motivation, not about why a DD does what he does, but rather why
s/he would do change the way s/he is doing something. Thus, this is
innovation research, the question about diffusions and adoptions.

There are plenty of factors which play a role in this, and I have
started to collect these *from the point of view of a Debian
developer*: http://phd.martin-krafft.net/wiki/factors/ . I intend to
also announce this list soon and let others comment, then possibly
also running a Delphi method on it.

If I don't limit myself to Debian developers, then I could spend
another 5 years collecting factors. I already feel like after
several weeks on this, I am far from the goal...

While I'm at it, here's my current plan, just in case anyone's
interested and/or wants to comment:

A framework to assess diffusions can be used to "compute"
(qualitatively) the likelihood of a speedy adoption of a tool in
a subject community, given attributes of the tool, the community,
the way it was spread, etc.

Given such a framework (to be selected, e.g. Rogers), which I use as
a starting point, and a diffusion of a tool in Debian (e.g.
debhelper), I want to modify the framework in such a way as to
minimise the discrepancy between projected adoption rate and actual
adoption rate (which closely relates to success of the tool,
I guess), using developer traits to steer and back up the

After several iterations and case studies, I hope to be able to have
condensed a framework which can be used to predict (and thus
engineer) the adoption rate of a tool in the Debian developer

I could go and look at other communities (e.g. Apache, kernel,
Plone, FreeBSD, etc.), but it'd be hard to deal with the bias, given
how immersed I am in the Debian community and how little I know the
others. I'd prefer to leave application of my framework to other
projects up for future research.

Comments welcome,

Martin F. Krafft        <mailto:martin.krafft@ul.ie>
Ph.D. student           http://phd.martin-krafft.net
Lero - the Irish Software Engineering Research Centre
University of Limerick, Ireland

Attachment: digital_signature_gpg.asc
Description: Digital signature (see http://martin-krafft.net/gpg/)

Reply to: