Women in the Free/Libre Open Source Software Development

Abstract

Free/libre open source software (FLOSS) has become a prominent phenomenon in the ICT field and the wider public domain for the past years. However, according to a FLOSS survey on FLOSS developers in 2002, “women do not play a role in the [FLOSS] development; only 1.1% of the FLOSS sample is female.” (Ghosh, Glott, Krieger, & Robles, 2002). In the mainstream research on FLOSS communities, many researchers also overlook different processes of community-building and diverse experiences of members, and presume a stereotyped male-dominated “hacker community” (e.g., Levy, 1984; Raymond, 2001; Himanen, 2001; Thomas, 2002). Moreover, issues around gender inequality are often ignored and/or muted in the pile of FLOSS studies. Female programmers often are rejected ex/implicitly from the software labour market (Levesque & Wilson 2004). The requirements of female users are not respected and consulted either (European Commission, 2001). This feature is opposite to the FLOSS ideal world where users should be equally treated and embraced (op. cit.). While many researchers endeavour to understand the FLOSS development, few found a gender-biased situation problematic. In short, women are almost invisible in current FLOSS-related literature. Most policies targeting at advocating FLOSS are also gender blind. Thus, this essay highlights the need for increased action to address imbalances between women’s and men’s access to and participation in the FLOSS development in cultural (e.g., chauvinistic and/or gender-biased languages in discussions on mailing lists or in documentations), economic (e.g., unequal salary levels for women and men), political (e.g., male-dominated advocacy environment) and technical (e.g., unbalanced students gender in technical tutorials) spheres. On the other hand, it also emphasises the powerful potential of FLOSS as a vehicle for advancing gender equality in software expertise. FLOSS helps transport knowledge and experience of software engineering through distributing source code together with the binary code almost without any limit. Many FLOSS licences such as the General Public Licence (GPL) also facilitates the flow of information and knowledge. In other words, if appropriately harnessed, FLOSS stands to meaningfully contribute to and mutually reinforce the advancement of effective, more expedited solutions to bridging the gender digital divide. In the end, this article points out that while women in more advanced countries have a better chance of upgrading their ICT skills and knowledge through participating in the FLOSS development, the opportunity is less available for women in the developing world. It is worth noting that although the gender issues raised in this article are widespread, they should not be considered as universally indifferent. Regional specificities in gender agenda in software engineering should be addressed distinctly (UNDP/UNIFEM, 2004).

Full text

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
Gendered Work Culture in Free/Libre Open Source Software
Development
Yu-Wei Lin, University of Stirling, UK
Email: y uwei.lin@stir.ac.uk
Matthijs den Besten, Montpellier Business School
E-mail : m.den-besten@montpellier-bs.com
Abstract
This article adopts a feminist perspective to examine masculine work culture in the development of
free/libre open source software. The authors draw on a case study of ‘the Heidi Bug’ discovered
during the development of the Mozilla Firefox web browser to examine how 'gendered talk' was
(en)-acted to facilitate 'bricolage' in an online work environment. Such gendered talks contain
cultural references familiar to male developers. Though seemingly innocuous, such acts could be
seen as a performance of gender that simply reflects the hegemonic heterosexual masculine
culture manifested in online virtual work space. The virtual work space therefore can be exclusive
to those who shared the cultural references. Although it may not necessarily be ignorance or
insensitivity of male developers, a more gender-balanced, women-friendly and inclusive work place
certainly would benefit from a more diverse environment. This paper highlights the gendered
aspect of software development through examining the language use and mainstream 'bricolage'
practice, and establishes a compelling ground for enlarging the talent pool to include more women
and integrating gender ethics (e.g., raising awareness of sensitive languages and design
approaches) into computer ethics education.
Keywords: free/libre open source software, open innovation, work cultures, online virtual
work space, heterohegemonic masculinity, humour, gendered talks,
Introduction
1

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
How different genders behave and communicate at work shape work cultures. Scholars have
looked into how work cultures can be gendered (e.g., McDowell 1997, 2009, Hall et al. 2007, Kelan
2008, Valentine et al., 2014, Warren 2015). For example, Hall et al. (2007) and Valentine et al.
(2004) investigate the embodied experience of female and male fire brigades. McDowell (1997)
looks into various ways of doing gender in the services sector. Warren (2015) studies the
masculinities of male workers in the context of surfboard-making. They all found that masculinities
and femininities are constructed through labour processes which are deeply emotional and
embodied.
Online virtual workplaces where people collaborate and communicate online has no exception to
such gendered work cultures. One can observe such gendering processes in the development of
free/libre open source software (FLOSS). FLOSS such as the Mozilla Firefox browser, is the result
of many incremental innovations contributed by people located in different places around the world.
Contributors communicate mainly online via email, on the Internet forums, mailing lists, Internet
Relay Chat (IRC), or on social media. Some of these channels are formal, used by organisations to
communicate work-related information, and others informal; some private, others public. While the
open characteristics of FLOSS, which include the freedoms to study, change, distribute and
redistribute source code, are warmly embraced, it has been argued that there are numerous
barriers stopping people from participating. Gender issue has been highlighted as one. For
example, the number of female contributors is extremely low, lower than the average in the ICT
sector (see Ghosh et al. 2002, Nafus et al. 2006a). Gendered or even sexist languages have been
observed on the communication channels mentioned above. While the 'open' nature of FLOSS has
generated a highly valued online collaboration culture and network effect (Benkler 2006), such
male-dominated networks have produced a culture predicated on male sociability and stabilised as
masculinist space (Lin 2005, Nafus et al. 2006a, 2006b, Reagle 2013, Valentine et al. 2014).
Barriers for women to take part in FLOSS development have been discussed (see e.g., Lin 2005).
For example, in terms of work culture and work-life balance, it is difficult for many women to juggle
different responsibilities in life while doing a job that requires long-hour commitment to coding.
Many women also do not have the skills and knowledge of some home-baked solutions circulated
only in a closed male-centric geeky free/open source software community; most female developers
seem to gain their knowledge from formal school curriculum (Lin 2005). Additionally, many have
reported experiences of discriminating, abusive, sexist languages used online and/or offline (e.g.
the RTFM culture), the lack of 'mentors' and role models, the male-dominated competitive
meritocratic culture. The term “brogrammer” was also coined (Fores 2012, Hicks 2013, Kumar
2

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
2014) to label the rather misogynist behaviour that male hackers sometimes exhibit when they
socialize. All of these contribute to a women-unfriendly environment (ibid.).
Numerous actions have been taken by larger FLOSS communities and ICT companies to improve
the gender gap. For example, the largest GNU/Linux distribution Debian has created the Debian
Woman project1 to encourage the participation of more female members and provide mentoring.
Other FLOSS projects such as the GNOME, KDE and Mozilla have also initiated similar groups
e.g., GNOME's Outreachy project2, the Womoz project3. Multinational IT companies such as
Google that use, develop and share FLOSS have also sponsored activities and opportunities to
bridge the gender gap e.g., Google-Sponsored Women Techmakers4 and the 'Made with Code'
initiative5. Also, non-profit companies such as Code First: Girls6 delivered training to young women
across the UK to increase the number of women in the tech industry.
Despite these efforts and the consensus of the imbalanced gender distribution in the field, the
questions about how such gendered environments emerge, and shape software development
processes, experiences of developers, and consequently the quality of work remain largely
unanswered. The above mentioned initiatives tend to focus on skill training that is visible on the
surface of this lack of gender diversity, rather than changing (sexist) behaviours that are based on
deeper but often invisible cultural differences. More thorough academic debates and scholarly
research that aims to identify hidden causes for the gender imbalance will therefore strengthen the
urgency of the need for diversifying the talent pool.
Following this line, this article will examine the gendered work cultures in the FLOSS development
through understanding how communications between FLOSS developers take place in an online
space. Social meanings and societal values are embedded in languages. ‘Language is the most
intense and stubborn fortress of sexist assumptions’ (Sontag 1973: 186). Analysing languages
therefore can deconstruct the linguistic devices created to facilitate gendered cultures in a certain
place (online and offline) and helps understand how virtual masculinist work space is produced and
stabilised through the repetition of certain communication behaviours (Valentine et al. 2014).
Through analysing the communications on public mailing lists (the language used specifically), this
1 https://www.debian.org/women/
2 https://gnome.org/opw/
3 http://www.womoz.org/
4 https://www.womentechmakers.com/
5 https://www.madewithcode.com/
6 http://www.codefirstgirls.org.uk/
3

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
article helps understand how developers behave and communicate, and how their communications
between developers reflect laddish culture and reproduce it in virtual workplaces.
Bug reporting, debugging and bricolage
The most effective way of understanding how a gendered online work environment is created is to
look into everyday narratives and common work practices. Bug reporting and bug fixing are
essential tasks in software development, but they are more than signaling software malfunctions
and fixing them. Bug reporting and bug fixing involve a process of identifying, raising, discussing
and resolving problems. It is a practice that requires both cognitive and communicative skills, and a
process that is highly socio-technical, subjective as well as interpersonal (den Besten and Dalle
2014).
In the FLOSS development, the identification and resolution of problems is commonly carried out
with the help of so-called ‘bug tracking systems’. As the name suggests, these systems allow for
recording, reporting, tracking, triaging and resolution (‘patching’) bugs. Reporting is an activity that
engages a lot of people, ranging from aspiring developers to computer-literate end-users
(Crowston and Howison, 2006). The sentence ‘given enough eyeballs, all bugs are shallow”
(Raymond 1998) highlights that an open bug reporting environment is seen as central to the
success of FLOSS. Yet, as Villa (2003) points out, it takes time and effort to deal with all the
reported bugs. To come up with a solution to the problem that the bug report identifies is not a
straightforward process. Although there are defined steps to take, communications, decision
making and problem solving can take place in many forms and in many directions.
The bug needs to be confirmed first, then verified (if the problem reported can be replicated), rated
(evaluating the importance and difficulty of the problem at hand), and finally, people with the
appropriate roles, responsibilities and expertise will be appointed to work on the problem. In the
debugging process, in order to solve the problem, all parties with an interest in its resolution need
to identify tools and resources available from different materials and sources. This is what Villa
(2003) calls ‘triage’ or what Cunha (2005) terms ‘bricolage’7. These FLOSS developers are
‘bricoleur’, ‘someone who uses 'the means at hand’ (Levi-Strauss 1966), using ‘instruments he
finds at his disposition around him... or to try several of them at once, even if their form and their
7 The concept is subject to a lively debate (see e.g., Perkmann and Spicer 2014, Baker and Reed 2005).
4

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
origin are heterogeneous’ (Derrida 1966), ‘using or adapting material which is at hand …
improvising, putting together found or second-hand items to create a look or emsemble’ (Barnard
1996: 192).
But all these activities, ranging from identifying, classifying something as a bug, to choosing a tool
to perform a task, could be political. As Borgman (2003) notes, ‘Multiple types of problems exist, as
do multiple types of knowledge that may contribute to solving them.’ (Borgman 2003: 99). It is this
cognitive and problem-identifying and problem-solving process where one may observe different
gender cultures reflected. Where to borrow ideas and concepts and tools from, and what strategies
to use to gather resources is also profoundly rooted in someone’ s cultural backgrounds,
influenced by someone’s social circles. Bricolage, from this perspective, can be political and
gendered.
Methodology and methods
This article uses the development of the Mozilla Firefox web browser as a case study to
demonstrate the gendered bricolage process. To study how bug-reporting and bug-fixing is
performed and played out, what (strategic) communications are involved to facilitate the work, we
conducted a case study of the 'Heidi Bug' drawing on the narratives in the bug tracking system
Bugzilla. We collected all the narratives in relation to the Heldi Bug and conducting an analysis on
the narratives. We started by looking at how questions are raised and phrased on the bug tracking
system, a common tool that has long been employed by developers to communicate with one
another in an open and distributed environment to manage the process of bug reporting, triage and
patching in the development and maintenance of software projects. An online forum discussion on
the [womoz] mailing list was instigated to understand how such gendered language is perceived by
female developers.
Unlike proprietary software development, problem solving in the FLOSS development is
transparent. A formal (and sometimes also informal) record of the problem solving process is kept,
which is archived and, mostly, publicly accessible. For instance, most FLOSS projects make use of
so-called bug tracking systems to keep a record of problems identified and features requested with
respect to their projects and of the efforts that have been made to resolve them. As we already
indicated, the process of identifying, reporting, discussing, and resolving of problems resembles a
5

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
process of bricolage, we study the communication and logs on Bugzilla bug tracking system to
investigate the gender politics embedded in this process. The threads on the Bugzilla system offers
a detailed account of problem-solving activities with documented time, dates and narratives. They
exemplify how programmers perceive, classify and frame a problem and how they come up with
solutions to it. Through narrative analysis, we identify signs of sexist languages in the bug
description or in the subsequent discussion and look for indicators of the effect that the
heterohegemonice pose (Frank 1987) has had on the success or failure of the process of bricolage
that accompanied the bug.
We then shared our preliminary finding from the analysis on the [womoz] mailing list (a mailing list
for female Mozilla developers) in 2013 to understand how this gendered event was perceived by
the members of the Women and Mozilla Project8 who were more sympathetic towards gender
ethics. After circulating a message on the [womoz] mailing list, we received responses from six
developers who started a discussion around heterohegemonic jokes, language and the
brogramming culture. This online discussion with the members on the [WoMoz] mailing list
revealed that people have different opinions towards what is considered as sexist, but men and
women like different types of jokes.
Bugzilla
The Mozilla foundation, which drives the Firefox browser development, uses Bugzilla as their bug
tracking system to organise information flows in a process of bricolage. In Bugzilla, bug reports are
assigned a status, which indicates the progress that has been made with the resolution of the bug.
Bugs reported by outsiders will have the status “unconfirmed” and only insiders with the so-called
“can-confirm privilege” can turn this status into “new”. Next, someone will be asked to take
ownership of the bug. If this person accepts the bug turns from “new” into “assigned”. After that the
bug may or may not be “fixed” and once it has been fixed solution will be verified before the bug
report is closed (cf. Masmoudi et al. 2009).
It is important to note that not all bugs can be fixed. Sometimes, a bug is orphaned (remains
unclassified, unanswered, unresolved) or abandoned. One of the reasons for abandoning a bug is
that it reports a known issue on which people are already working. In that case, it will be declared a
“duplicate.” Alternatively it may be decided that the problem reported is not a real problem and in
8http://www.womoz.org/
6

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
that case the bug will be closed with a line saying “works for me.” Apart from changing the status of
bugs, another important activity of the people participating in the bug resolution process on
Mozilla’s Bugzilla is to map out the mostly technical dependencies among bugs. A bug is said to
“depend on” another bug if the resolution of the former is only possible after the issue identified in
the latter has been resolved. Inversely, a bug is said to “block” another bug if the resolution of the
latter is only possible after the issue identified in the former has been resolved. Together the
dependencies among bugs form a bug report network (Sandunsky et al., 2004). Hence, by
identifying dependencies, participants contributing to the discussion forum associated with a bug
work together to resolve this bug in a network. Once verified and confirmed, the bug becomes a
shared concern for the people working collectively and collaboratively on it.
Since software systems are complexly networked and connected, related bugs are all relevant and
as such included in this network. Besides, bugs are classified according to the priority that should
be given to them in relation to other bugs. In particular, that can be nominated to be added to a list
of bugs to be resolved before a certain milestone is reached (typically a future release of the
software). In short, the log of past actions and comments about 'bugs' and 'issues' allows us
strategic communication and bricolage takes place.
The Heidi bug
“Heidi” is the alias of a bug reported by Christopher Aillon in 2002.9 Typically bugs that are reported
are assigned a unique identification number (in this case 121084). Aliases are sometimes provided
to allow people to easily distinguish and locate the bug report. In this case “Heidi” refers to the
German model Heidi Klum, whose association with the bug is to illustrate the problem of images
failing to be reloaded on certain occasions. The bug description contains presuppositions that are
indicative of the adoption of a heterohegemonic pose.
Here, we provide a chronological description of the event to demonstrate how a process of
bricolage was performed. We draw on the bug report data of the Heidi bug, compare it to the
processes of bug resolution in two other related bugs, which address similar issues.
The first is bug 98890, a bug reported by Guillaume Filion several months before Aillon reported
his. Even though it was reported earlier, this bug was eventually classified as a duplicate of the
 !"#
$%&'()" *+,-
"
7

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
Heidi bug (rather than vice versa). The second is bug 93015. This bug is related to the Heidi bug
via bug 83774. It was worth noting that the people working on bug 93015 had already resolved
many issues described in the Heidi bug, and that Jim Dunn, who was then assigned the bug, made
the connection with bug 83774 and managed to recruit Boris Zbarsky to devote his attention to the
remaining issues and fix the bug.
Figure 1 is a section of the bug report network around the Heidi bug, which provides an overview of
the complex communications between developers through a bug tracking system. Each bug report
is associated with a forum for asynchronous communication among developers. Dependency
relations are noted in the bug reports as soon as they have been identified.
Figure 2 is a partial bug report network that shows the dependency between the Heidi bug (report
number 121084 in bold) and other related bugs we studied in this paper. The dates of reporting are
indicated in brackets below the bug identifiers for each single report. Some bug reports are
mentioned as ‘collections’ and the dates of reporting are shown as range. Double-headed arrows
are used to show duplication of bug reports, and single-headed arrows are used to show
dependency. The Heidi-bug has 60 duplicates, and it directly depends on the resolution of the
report (89419) and blocks 6 other reports.
Gendered talk as a strategy for bricolage
Let us compare two bug reports: the summary that Christopher Aillon initially gave for his problem
was “Heidi Klum doesn't like being compared to the cache.” In contrast, Guillaume Filion chose to
describe essentially the same issue as “The URL of the image is showed instead of the image.”
While Filion’s summary was more informative, Aillon’s had a humorous twist.
Judging by the names (ID of contributors), among the 69 people participating in the discussion on
the Heidi bug, only one was female.
Among the 11 people discussing Filion’s bug report, no one appeared to be female. Note that Aillon
gradually added informative content of the bug summary after he engaged fellow developers. He
first reported the bug in January 2002. In April 2002 he added the explanation “Images requested
twice -> "The image cannot be displayed, because it contains errors." message”. In June, the joke
about Heidi was removed from the summary and replaced by the alias “Heidi” for the bug.
8

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
It appears that Aillon with his Heidi bug managed to gain more attention than Filion with his earlier
description of the same problem. Part of the reason that Aillon succeeded might be because he
made the problem look more interesting. At the same time, from the Heidi bug history, it was clear
that Aillon and others working on the bug went into great lengths to recruit people for his problem
(discussing it on IRC, adding people to the CC-list, establishing dependencies), at any rate more
than what Filion did. Moreover, Aillon was also proactive. For a long time, the bug failed to get full
attention from core developers. In March 2002 Aillon complained about the lack of attention the
issue received:
<sl8r> matti: i'm saying that at least for me this is a highly visible bug. i can't believe that
nobody else is seeing it. there are no dupes etc.
<sl8r> caillon: yeah... and i can't nominate it for anything, i'm not "sufficiently
empowered" ... *sigh*
After this complaint, David Hallowell noted a month later that the problem also appeared on
Windows and after that people started to identify duplicates. All this activity convinced Stuart
Parmenter to change the target milestone from the assignment “future” that he had given in
February to the more concrete “mozilla1.1alpha” in April. In July, Jason Bassford, who had been
working on Filion’s bug, noticed the Heidi bug and pointed out that it was a duplicate of Filion’s.
Christopher Aillon, however, did not appreciate this action:
Dude, don't dupe this bug to the other just because it has a lower number. This one has
more people on it that care about the bug. It's been triaged already and people know how
to find this bug and not the other. If the other is the same as this, then dupe that one to this
one. Re-opening. (The gendered/broish terms in the quotations are italicized by the
authors.)
The fact that Filion had even less success in getting attention could be due to the more neutral
description that he chose, but just as well it could be because fewer efforts were made to make the
bug known to Mozilla developers. However, this episode exemplifies how strategic communication
(humour in this case) is brought into a bricolage process to solve a problem, and reflects certain
male-centered values, though implicitly. After identifying a bug, the reporter, instead of providing a
plain description of the fault, provided an anecdote containing a heterohegemonic presupposition
9

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
in his pitch of the problem to attract more attention. The light-hearted tone gave the bug report
much currency to flow in the busy and complex bug tracking system. However, while successfully
securing the attention (resources) from fellow developers on this particularly bug, the content of
this bug report reflects the heterohegemonic value that is prevailing in mainstream society. For
example, the use of ‘Dude’, and a heterosexual male gaze on a highly sexualised and objectified
female celebrity model, and the rather patronising tone throughout the message. The gendering
process started when the bug reporter assumed fellow developers would share his worldview or
background belief (assuming that the female blond model is a symbol of beauty and a pleasant
image on a computer monitor). This presupposition was verified (thereby exacerbating the already
gendered situation) when much needed resources (attention from fellow developers) were
allocated. It was confirmed that the presupposition was mutually known by the reporter and the
addressee; both parties considered such an utterance appropriate in context. The well-received
bug report reaffirms that such a brogrammer's view and culture is accepted and shared.
There had been other signs of hegerohegemonic culture. It turns out that the source code to which
the patches were applied in bugs 83774 and 93105 was called “libpr0n”. A Mozilla forum gave the
following explanation for this name:
“The main goal of the library is to render pornographic images in an efficient way. Plus, the name
"imglib2" is boring.”10
As developers are aware that the Bugzilla bug reports are public, the casual use of gendered
cultural references might be indicators of bigger underlying issues. In this particular case of the so-
called 'Heidi Bug', there had been both positive and negative effects on the capacity of the bug to
attract attention and resources. The way through which the “Heidi bug” and other bugs closely
related to it were fixed seems to suggest that the 'small and casual talk' used to 'frame' a question
is not trivial at all. In fact, the adoption of a heterosexual masculine pose in a specific context might
help bricolage. But, the very same strategy may not be replicable in other contexts (as it has not
been seen elsewhere). Some fellow developers remarked that the gendered attitude and
behaviours are not welcome for resolving a bug.
Responses from the Mozilla community
./(.0./1'2/!)2/.1
10

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
As a way of reflecting and reciprocating, the researchers shared the preliminary analysis on the
[womoz] mailing list in autumn 2013. [Womoz] is a sub-community of Mozilla established to
encourage women to participate in the development of the FLOSS Mozilla Firefox browser.
Discussing this case study on [womoz] therefore could enact a much-needed reflexivity and
reciprocity into online research (Morrow et al. 2015). The researchers decide to report the remarks
from the community verbatim and at length to reflect the kind of dynamics and conversation that
people would have in a virtual work place.
Few members had heard of the so-called 'Heidi Bug' (or the 'Heidi Klum Bug'). There were short
responses describing the bug as “'old and quite simple' - (i.e. technically uninteresting, and a bug
reported 10 years ago cannot be regulated by the community guidelines11 where sexist languages
are not allowed)”.
One thought 'Heidi Bug' is humorous but problematic:
An informative title can be easy for people to write off "oh I see from the summary this
doesn't matter to me" whereas a humorous title can get people to read more of the bug,
and once they're in the bug, they can't help but try and solve the problem.
I honestly can't think of any other examples like this where the bug is humorous but it's a
real issue. The other ones I can think of are humorous entirely. (MC1)
Another said:
The reporter of the bug gave an URL to an adult site to demo the bug. It's disputable, given
that the bug was apparently very frequent. There was audience to the bug and it was
solved. Could not say if the sexist jokes helped or not.
The expected results made me laugh though it can be considered as sexist. But it's not
really that bad (at least he offered lunch), it could have gone worse (especially in the
comments). (FA1)
11 The issue is the use of a sexist joke to draw attention to an actual technical issue - seems likely to violate some sec-
tion of our community participation guidelines:
https://www.mozilla.org/en-US/about/policies/participation/ but this bug is more than 10 years old, so the
policy was not applicable, at the time.
11

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
Another more detailed account was provided:
I can't say in my years developing code that I have come across other examples to give
you, but I'll give you my thoughts on this, what I have, and I hope it helps.
While I do notice that the 'representative bug' is not the oldest one [1], as is convention(but
this convention is by no means an invariant of bugzilla), but the 'funny' one about Heidi. It is
possible that the offensive nature[2] of the test image & humorous text made it more memo-
rable than the original bug.
It is possible the Heidi bug was more memorable than the original bug because it could be
viewed as a computer science in-joke. Computer graphics/image issues often need images
of figures, especially figure& face for test cases (dynamic range, multiple types of curved
lines, etc). While I personally had no love of seeing Playboy centerfolds used casually in my
university graphics classes[3], it is something that is still baked into the field's history, so
'Heidi Klum as a modern day centerfold' may be a sly reference/homage to the use of cen-
terfold images for testing dating back to the 1960s.
The closest I have encountered personally might be not gender but sexual orientation
based[4], and there was no particularly special reaction reaction when we triaged the bug.
I also notice that the Heidi bug has 59 duplicates, which would make it a highly visible
problem for developers. In my experience, often reported, highly visible, and thus
memorable bugs do acquire names. However, the name is usually associated with a real
person, often the bug finder or the bug writer. Examples such as 'the mjrosenb memorial
bug' or 'the katz bug' come to mind. One could argue that the bug takes on the gender of
the person associated with the bug, but that might be a bit of a stretch?
I suspect that may be a topic where there is more unconscious bias than overt bias is in
play, and you may need statistics on a large data set to prove it, but I'm afraid I do not know
how one would extract that kind of nuance from bugzilla.
I suppose we could file a graphics issue using a saucy image of a man for a test case and
another with a gender-neutral-meme and compare the results.Would that make a good
case study? We just turned on apzc on metrofx, so I've got graphics issues to spare right
now.
12

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
[1] first bug filed was https://bugzilla.mozilla.org/show_bug.cgi?id=98890
[2] bug reporter notes offensive image https://bugzilla.mozilla.org/show_bug.cgi?
id=121084#c2. As of this writing, the original image link 404s so I can't tell how much re-
semblance it bears to the iconic test images like [3]
[3] http://en.wikipedia.org/wiki/Lenna
[4] https://bugzilla.mozilla.org/show_bug.cgi?id=903953#c2
(AMN1)
However, some disagreed that this way of framing the bug is 'sexist' and inappropriate:
I know we all have different feelings here, but I am uncomfortable with calling a straight
male who likes to look at women "sexist." I know I sure like looking at men (well and women
too, we're gorgeous ladies!) I'm totally comfortable with calling it inappropriate in x,y,z situa-
tions because it *does* offend some people. (MC2)
To follow up on an earlier comment (not quite in thread, apologies) I'm not saying its not
okay for men to find women attractive - I find it objectifying and inappropriate in this *con-
text* though. That's just me. Anyway. It's a really interesting conversation. (LS1)
It also seems that the Mozilla community has developed quite a smooth working relationship so de-
velopers know each other. So a more experienced respondent, instead of accusing their male
counterparts, rationalised:
I actually know Chris and would be happy to ask him if he remembers how he ended up
choosing this example. One thing that occurred to me is that we test against the top 50 or
100 or whatever popular sites, and many of these are porn. So if we see a bug that refers to
a porn site it could simply be that it was part of the set of test sites and that's why the error
was caught there. If that is happening and that is the case this is a chance for us to give ad-
vice on whether or not we think the bug should be reproduced on a non-porn site, and then
filed that way. (MC2)
13

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
Here is the comment in the bug that explains why the older bug was duped to the newer
bug -https://bugzilla.mozilla.org/show_bug.cgi?id=121084#c81
This happens, you can see the dupe wasn't found for a long time, and the relevant people
had seen the newer bug but not the older bug. Think that a developer notices something,
files a bug, pings his/her developer friends to help out on it - that bug has obviously got in-
stant attention and work started on it. (MC3)
I agree with Majken (now that she explained why the image could have come from an adult
site...). I laughed at the idea of Chris expecting to have dinner with Heidi and so on. It's just
the kind of thing I would say, totally quirky humor...
There was no further comment on the kind of picture or Heidi herself in the comments.
Do you know that, in french, "bug" and "problem" are male words whereas "solution" is a
female word?
E.g. : "J'ai un bug!" "J'ai trouvé la solution."
Think of that the next time you file a bug :)
(FA1)
But this discussion has raised some attention on developers's bug-reporting behaviours:
I agree with the point of this, is this something that happens more than once? How often, is
it the same people?
(MC2)
The "Heidi" bug looks to me as though the original reporter of the bug just didn't bother to
find a non-adult example since he felt comfortable posting adult content (and his
accompanying 'humorous' STR) in a Mozilla bug report. This isn't directly sexism but it's
certainly an example of privilege and comfort of being the straight guy in a primary straight
guy space. So if the intent of bringing up this thread was to discuss whether bugs like this
might alienate potential community members in an open source project I would say that
(besides it being an old and resolved issue) yes, it might be and it's very likely to come up
(see the bug above) on occasion in a project.
I like how Liz dealt with the bug above, a recent example, in that she called it out for what it
was and since there was no _real_ issue, the bug has remained closed. We don't monitor
14

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
bugzilla for stuff like this, or at least not in any official capacity, but don't think that there's
not unintentional leveraging of privilege and power imbalances playing out in our bug
tracker sometimes that might accidentally scare off or set the wrong atmosphere for a
potential contributor.
Good things to think about when filing our own bugs, and when helping others file theirs.
(LB1)
Gendered language and its hidden meaning
Valentine et al. (2014) emphatically argue that
‘the public expression of the most blatant forms of gender prejudice, sexism persists and is
manifest in subtle ways. As a consequence, it can be difficult to name and challenge with
the effect that patriarchy as a power structure which systematically (re)produces gender
inequalities is obscured by its ordinariness.' (p. 401)
In an environment where distributed collaboration takes place, it is even more challenging to detect
invisible sexism hidden in technical languages, and to manage the relationships between actors
from different backgrounds. To facilitate the collaboration, some languages have been selectively
or strategically used to motivate or enact participation. However, some of these languages also re-
flect a hegemonic masculine culture that dominates in the computing field. Seemingly innocuous
jokes actually reflect male-centric position and may lead to invisible sexism (Bemiller and Schnei-
der 2010, Shifman and Lemish 2010a, 2010b).
In light of the responses received from the Mozilla community, most of the respondents agreed that
the title of the 'Heidi Bug' report was humourous, and was a form of strategic communication.
While few considered it sexist, some did notice the cultural meaning hidden in the technical
discussion and the subtle framing. The cultural reference employed in the 'Heidi Bug' report
explicitly reflects the brogrammers's interests, beliefs and aesthetic values. The attention received
suggest a shared brogramming culture supporting such a value system. Had it been a woman
reporting the same bug joking about problems of seeing half-naked men, would it be treated in the
same way in a brogrammer community prevailing with heterosexual masculine culture, the kind of
15

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
“hegemonic form of masculinity prevalent in society, which determines who counts as a ‘real’ man”
(Frank 1987, Connell 1995, Kelan 2008).
Not only does the brogramming culture exist online, but also offline, for example at free/libre open
source software conferences, where male speakers outnumber female. While these tongue-in-
cheek jokes told in a speech given by a male speaker can be considered as harmless, they reflect
certain hegemonic masculine values dominated in society. The speakers tell these jokes to amuse
like-minded conference participants / delegates, but these jokes may not amuse women (or offend
them even). These can be seen as the kind of microaggressions or microinvalidations referred to
by Mitchell in her study of live experiences of women in geek communities (2016). Those who do
not share the sentiment are excluded from the ambient.
Our observation echoes Wikhamn and Knights (2013) who argue that hegemonic masculine
discourses are reproduced rather than challenged by open innovation. In their research, they found
that when masculine norms govern an open innovation setting, it is difficult to incorporate
alternative (feminine) discourses when acting within a strong masculine hegemony (ibid.).
By no means is this paper suggesting a censorship on humour from a heterosexual male
persopective; nor is this article arguing that there are gender differences in language use in terms
of dichotomies such as public vs private and informational vs affective (Talbot 2010). Through
focusing on the construction and performance of gender in discourse (ibid.), this paper merely
demostrates that languages or presuppositions or anecdotes cited for strategic communication can
reflect certain types of cultures and they dominate the stage / field / platform. The work cultures in
a distributed online environment can still be hegemonic masculine, as noted in Wikhamn and
Knights (2013), and we need to pay more attention to such patterns and langages used to ensure
that different voices (women, LGBT, disabled, other minorities) are not excluded or muted in these
environments. It also demonstrates that software development, like other types of works, are not
neutrally technical jobs, but full of gendered performance and gender politics (as argued in Kelan
2008). This existing body of works challenges heterohegemonic masculinity, important for achiev-
ing more inclusive, tolerant and equitable workplaces. In a similar vain, we are emphasising the
need to diversify the languages so as to have more inclusive work cultures on the internet. While
feminist scholars have challenged gendered jokes and humours, it seems feminist ethics have not
yet fully integrated into the computer science or software engineering curriculum. This is an effort
to raise awareness by discussing the discourse emerging from the everyday talk in distributed soft-
ware environments.
16

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
Conclusions: Software development as a gendered form of work
Herring (1996) powerfully argues that '“netiquette” (a combination of “network” and “etiquette”)
norms are both a moral and a political dimension, in that they are founded on systems of values
and judgments which may vary according to different groups of users.' (p. 115) Her research has
found that in computer-mediated communication, “it is typically the most powerful or dominant
group whose values take on a normative status” (ibid.). She further submits that “the gender differ-
ences in public discourse on the Internet are not randomly distributed across individuals, but rather
display a systematic pattern of distribution with male users as a group tending toward more adver-
sarial behavior and female users as a group tending toward more attenuated and supportive be-
haviors” (p. 137).
Many other feminist scholars have also questioned digital materials or artefacts are imbued with
traditional gender identities, norms, roles, symbols, meanings, and proposed gender-sensitive de-
sign approaches (Rommes et al 2001; Kember 2003; Bath 2007; Rommes et al. 2012). For exam-
ple, in her work on AI, Kember (2003) talks about predictable and stereotypical female forms (e.g.,
electronic assistants and robot nurses). In her study of the semantic web, Bath found modern infor-
mation systems and computational artefacts often are designed with “binary assumptions about
women and men [which] are not reflected [upon] or the (gender) politics of [a particular] domain is
ignored. Thus, the existing structural-symbolic gender order is inscribed into computational artifacts
and will be reproduced by [their] use.”12
This paper looks into how seemingly technical activities such as bug-reporting and bug-fixing are
actually gendered. As argued earlier, framing a problem (classifying something as “a matter” or “an
issue”) is a social process. In the context of bug resolution, it involves efforts to gain the attention
and interest of people who might have the key to solving the problem identified. Consequently, one
might interpret brogramming behavior as an attempt by the “bricoleur” to create a common ground
and make the bug that needs resolving seem fun and appealing. Unfortunately, what might look
appealing to some group will offend others and the repeated search for a common ground might
engender a process in which the purported interests of the group that is appealed to become more
and more homogenous as those who feel offended leave or refrain from entering. Fores (2012)
argues that it is common to observe misogynist behavior among hackers when they socialize.
Strategic communication involved in performing bricolage that contains presuppositions of
12 http://readwrite.com/2008/08/06/will_the_semantic_web_have_a_g
17

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
mainstream interests and beliefs reproduces heterohegemonic masculine cultures in society in the
virtual work places, as also argued in Wikhamn and Knights (2013).
While the authors are aware that the analysis of a bug done in this paper cannot be generalised to
the whole software development process, our analysis on the communications in a publicly acces-
sible bug tracking system does suggest that software development process can be highly gen-
dered. The gendering process started from the initial problematisation and framing of a bug (i.e.
cultural references cited) and the sense-making discussion processes even though the extent to
which they play a role remains unclear. Brogramming languages and behaviours are often made
hidden or invisible in the highly technical talks, such as the one examined in this paper. But such
seemingly pragmatic mindset of bricolage held by brogrammers in fact reflects the gender norms
prevailing in society (e.g., sexualisation and objectification of women bodies and static gender
roles).
This highlights the tensions and conflicting values in the FLOSS development: on the one hand,
the culture of 'openness' in FLOSS welcomes valuable practices for flexibly managing different re-
sources for problem framing, collective problem solving (e.g., teamworking, networking, and brico-
lage), but on the other hand, the prevailing 'brogramming culture' leads to an undesirable “gen-
dered arrangements of values [that] perpetuate dominance even in cases where no intimidation is
intended” (ibid. p. 137).
How to inject the findings from our study back to the apparatuses of computer ethics? Our work
contextualises bricolage in a male-dominated decentralised online work environment such as the
FLOSS communities. It bears a lesson for IT managers: the kind of bricolage observed in the
FLOSS development is a useful skill and practice, but on the other hand, gender should be
strengthened as an important area in computer ethics education to eliminate brogramming culture.
Whilst computer ethics have been developed into an academic and practical discipline, we need
more rigorous methods to integrate feminist ethics and gender-sensitive languages into computer
ethics education (Adam 2000; Adam and Ofori-Amanfo 2000). That said, we need to continue to
update the existing and invent new languages and vocabularies to delineate or develop a field of
inquiry in relation to gender so that we can engage in intellectually sound discussion and make
computer ethics education practically relevant to professionals working in information systems
18

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
(Stahl et al. 2014). Some community-based efforts of creating community ‘codes of conduct’ (e.g.,
the Debian Code of Conduct13) or a list of harrassment incidents (the Geek Feminism Wiki14), or a
statement illustrating a shared community identity (e.g., the Debian Diversity Statement15) may
also be a solution to documenting such undesired treatment. Lastly, conferences should increase
female speakers or reject those sexist jokes that please male delegates more than female. Proac-
tive actions must be taken by conference organisers or community coordinators in addition to some
affirmative action on paper.
To covercome the methodological limitation of this paper, future studies will include interview with
more developers about how they work, communicate, compose email messages when reporting
bugs to understand online FLOSS work cultures and environments more.
13 https://www.debian.org/code_of_conduct
14 http://geekfeminism.wikia.com/wiki/Online_harassment
15 https://www.debian.org/intro/diversity
19

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
References
Adam, A. (2000) 'Gender and computer ethics in the Internet age'. The CPSR (Computer
Professionals for Social Responsibility) Newsletter 18(1). URL
http://cpsr.org/prevsite/publications/newsletters/issues/2000/Winter2000/adam.html/ (retrieved
on 7th December 2014)
Adam, A. and Ofori-Amanfo, J. (2000) 'Does gender matter in computer ethics?' Ethics and
Information Technology 2(1): 37-47.
Baker, T. and Nelson, R. E. (2005) ‘Creating something from nothing: Resource construction
through entrepreneurial bricolage’. Administrative Science Quarterly 50(3): 329-366.
Barnard, Malcolm (1996) Fashion as Communication. London: Routledge.
Bath, Corinna (2007): "Social" Robots & "Emotional" Software Agents: Gendering Processes and
De-gendering Strategies for "Technologies in the Making" in: Isabel Zorn, Susanne Maaß, Els
Rommes, Carola Schirmer Heidi Schelhowe (Eds.): Gender Design IT. Construction and
Deconstruction of Information Society Technology. Wiesbaden, VS-Verlag, 53-63.
Bemiller, M. L. and Schneider, R. Z. (2010). 'It's not just a joke'. Sociological Spectrum 30(4): 459-
479.
Benkler, Y. (2006). The Wealth of Networks: How Social Production Transforms Markets and
Freedom. New Haven and London: Yale University Press.
den Besten, M. and Dalle, M. (2014) ‘Coordination by reassignment in the Firefox community’.
Proceedings of ECIS 2014, July, Tel Aviv.
Borgman, C. L. (2003). ‘Designing Digital Libraries for Usability’. In Ann P. Bishop, Nancy A. Van
House & Barbara P. Buttenfield (Eds) Digital Library Use: Social Practice in Design and
Evaluation. London: The MIT Press.
Connell, R. W. (1995). Masculinities. Cambridge: Polity.
Crowston, K., Howison, J., (2006). Hierarchy and centralization in free and open source software
team communications. Knowledge, Technology & Policy 18, 65–85.
Cunha, M.P. e, (2005). Bricolage in organizations (FEUNL Working Paper Series No. 474).
Universidade Nova de Lisboa, Faculdade de Economia.
Derrida, J (1966) Structure, Sign and Play in the Discourse of the Human Sciences and
Discussion. Presented at an international symposium, entitled ‘The Languages of Criticism and
the Sciences of Man’, at The Johns Hopkins Center, Baltimore, USA, on 18th – 21stOctober
1966.
20

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
Frank, B. (1987) ‘Hegemonic heterosexual masculinity’. Studies in Political Economy, 24, pp. 159-
170.
Fores, B., 2012. Brogrammers: When computer nerds become frat boys. Daily Caller.
Ghosh, R. A., Glott, R., Krieger, B., Robles, G. (2002). 'Free/Libre Open Source Software: Survey
and Study Part 4 Survey of Developers'. http://www.flossproject.org/report/FLOSS_Final4.pdf
(Retrieved on 30 June 2015).
Hall, Alex, Jenny Hockey, and Victoria Robinson. (2007). “Occupational Culture and the
Embodiment of Masculinity: Hairdressing, Estate Agency and Firefighting.” Gender, Work and
Organisation 14: 534– 551.
Herring, Susan (1996) 'Posting in a Different Voice: Gender and Ethics in Computer-mediated
Communication'. In Charles Ess (Ed.) Philosophical perspectives on computer-mediated com-
munication. New York: State University of New York Press. pp. 115-146.
$3!42/.15*6#)789$:;<)
783 ===31>4.53?@6??
Kelan, E. K. (2008). Emotions in a rational profession: the gendering of skills in ICT work. Gender,
Work & Organization, 15(1), 49-71.
Kember, S. (2003) Cyberfeminism and Artificial Life. London: Routledge.
Kumar, D. (2014). Disrupting the cultural capital of brogrammers. ACM Inroads, 5(3), 28-29.
Levi-Strauss, C. (1966) The Savage Mind. London: Weidenfeld and Nicolson.
Lin, Y.-W. (2005). Gender dimensions of FLOSS development. Mute Magazine, 2(1). URL:
http://www.metamute.org/editorial/artic les/gender-dimensions-floss-development (retrieved on
14 Nov 2012).
Masmoudi, H., Besten, M. den, Loupy, C. de, Dalle, J.-M. (2009). Peeling the Onion: The Words
and Actions that Distinguish Core from Periphery in Firefox Bug Reports, and How They Interact
Together, in: Crowston, K., Boldyreff, C. (Eds), Proceedings of the Fifth International Conference
on Open Source Systems.
McDowell, L. (1997). Capital Culture: Gender at Work in the City. Oxford: Blackwell.
McDowell, L. (2009). Working Bodies: Interactive Service Employment and Workplace Identities.
Oxford: Wiley-Blackwell.
Mitchell, L. G. (2016). Exploring the impact of gender on ‘online’ and ‘in person’ lived experiences
of women in geek communities. Master of Research Thesis. Univiersity of Glasgogw.
Morrow, O., Hawkins, R. and Kern, L. (2015). 'Feminist research in online spaces'. Gender, Place
& Culture: A Journal of Feminist Geography, 22(4): 526-543.
21

Lin and den Besten, post-review manuscript accepted for publication in Gender, Work & Organization.
Nafus, D., Leach, J., Krieger, B. (2006a). FLOSSPOLS Report: Gender: Integrated Report of
Findings. URL: http://flosspols.org/deliverables/D16HTML/FLOSSPOLS-D16-
Gender_Integrated_Report_of_Findings.htm, 2012/11/14
Nafus, D., Leach, J., Krieger, B. (2006b). FLOSSPOLS Report: Gender: Policy Recommendations.
URL: http://flosspols.org/deliverables/D17HTML/FLOSSPOLS-D17-
Gender_Policy_Recommendations.htm, 2012/11/14
Perkmann, M and Spicer, A. (2014) ‘How emerging organizations take form: the role of imprinting
and values in organizational bricolage’. Organization Science 25(6): 1785-1806.
Raymond, E. S. (1998). 'The Cathedral and the Bazaar'. First Monday 3(3).
Reagle, J. (2013). “Free as in sexist?” Free culture and the gender gap. First Monday 18.
Rommes, E., Oost, E. van & Oudshoorn, N. (2001): Gender in the Design of the Digital City of
Amsterdam, In: Green, E. & Adam A: Virtual Gender. Technology, Consumption and Identity.
London: Routledge, 241-261.
Rommes, Els, Corinna Bath, Susanne Maaß (2012) 'Methods for Intervention. Gender Analysis
and Feminist Design of ICT'. Science, Technology & Human Values 37(6): 653-662.
Shifman, L. and Lemish, D. (2010a). 'Between feminism and fun(ny)mism: Analysing gender in
popular internet humour'. Information, Communication & Society 13(6): 870-891.
Shifman, L. and Lemish, D. (2010b). '”Mars and Venus” in virtual space: Post-feminist humor and
the internet'. Critical Studies in Media Communication 28(3): 253-273.
Sontag, Susan. (1973). “The Third World of Women.” Partisan Review 40: 186– 199.
Stahl, B. C., Eden, G., Jirotka, M. and Coeckelbergh, M. (2014) 'From computer ethics to
responsible research and innovation in ICT: The transition of reference discourses informing
ethics-related research in information systems'. Information & Management 51(6): 810-818.
Tablot, M. (2010). Language and Gender. 2Nd Edition. Polity Press.
Valentine, G., Jackson, L. and Mayblin, L. (2014) 'Ways of Seeing: Sexism the Forgotten
Prejudice?'. Gender, Place & Culture: A Journal of Feminist Geography, 21:4, 401-414, DOI:
10.1080/0966369X.2014.913007.
Villa, L. (2003). Large Free Software Projects and Bugzilla: Lessons from GNOME project QA, in:
Proceedings of the Linux Symposium. Ottawa, Canada.
Warren, A. (2015). 'Crafting masculinities: gender, culture and emotion at work in the surfboard
industry'. Gender, Place & Culture: A Journal of Feminist Geography
Wikhamn, B. R. & Knights, D. (2013). Open innovation, gender and the infiltration of masculine
discourses. International Journal of Gender and Entrepreneurship, 5(3), 275-297.
22

Heidi
83774 viewblocked
93015
66022 moz-broken 47475 100970 125768 125983 130265 146513 192204 193865 196380 196735
89419
48202 223142 CVE-2008-5012 458845

121084
[Heidi]
(Jan 2002)
121084
[Heidi]
(Jan 2002)
98890
(Sep 2001)
98890
(Sep 2001)
89419
(Jul 2001)
89419
(Jul 2001)
83774
(Jun 2001)
83774
(Jun 2001)
93015
(Jul 2001)
93015
(Jul 2001)
59 reports
(Feb 2002-02 –
Mar 2003)
59 reports
(Feb 2002-02 –
Mar 2003)
11 reports
(Sep 2001 –
Sep 2005)
11 reports
(Sep 2001 –
Sep 2005)
3 reports
(Mar 2006 –
Feb 2009)
3 reports
(Mar 2006 –
Feb 2009)
10 reports
(Feb 2001 –
Dec 2008)
10 reports
(Feb 2001 –
Dec 2008)
5 reports
(Mar 2000 –
Nov 2002)
5 reports
(Mar 2000 –
Nov 2002)
13 reports
(Jan 2003 –
Apr 2003)
13 reports
(Jan 2003 –
Apr 2003)
171895
(Oct 2002)
171895
(Oct 2002)
32 reports
(Jul 1999 –
Mar 2003)
32 reports
(Jul 1999 –
Mar 2003)
96926
(Aug 2001)
96926
(Aug 2001)
3 reports
(Jan 2001 –
Mar 2002)
3 reports
(Jan 2001 –
Mar 2002)
4 reports
(Feb 2002 –
Dec 2002)
4 reports
(Feb 2002 –
Dec 2002)
Feb 03Mar 03
Jul 02
Feb 03
Oct 02
Nov 01