The Devternity Conference, 2023 and the Erasure of Women’s Labour

Ishita Ghosh and Vidya Subramanian*

In the field of technology where developments are taking place at a rapid pace, lack of diversity, inclusivity and gender disparities can exacerbate the issue of discrimination and bias due to inequitable participation of women. In this piece, the authors suggest the adoption of the Bechdel test-which was originally designed to gauge the representation of women in the areas of films and television-in the realm of technology to assess and offer a mechanism for gauging the extent of participation of women in the technological sphere.

Introduction

The organizer of a software developer conference ‘DevTernity’, Eduardo Sizovs was recently in news for concocting AI-generated profiles of fake female guest speakers to showcase diversity and attract greater participation from popular speakers and attendees alike. The organizers were forced to cancel the event as several prominent speakers pulled out on realising the deceit. When questioned on the reason behind the lack of diversity, Sizovs justified it by saying that it was a one-off incident.  However, Georgely Orosz’s- the coder who broke the story in his newsletter- investigation showed that it was a continuous practice that started way back in 2021 when the Conference went online due to COVID, right after someone noted the all-white and all-male line-up at the conference.

What is even more concerning is the fact that the Conference’s Co-organiser, Julia Kirsina, who was listed as a speaker for three consecutive years, from 2021 to 2023 had not delivered a single talk at Devternity. The final nail to the last edition of Devternity’s coffin was the discovery that Sizovs had likely been sockpuppeting—another name for controlling an account just like a sock puppet speaks in the voice of the sock puppeteer— the Conference’s Co-organiser, Julia Kirsina’s coding advice account on X/twitter from where these posts were being made. In essence, the inclusion of women speakers was only a hollow promise to counter the critique of netizens, speakers and participants alike to the majority of white, male speakers that constituted the default speaker panel at the Conference. In this blogpost, the authors try to highlight the extant issue of  inequitable participation of women in the field of technology and offer a benchmark test which can function as a tool to estimate the participation of women in tech in particular, and which is also extendable to other male dominated fields.

Devternity: The Exception or the Norm?

Devternity forms part of an overarching trend of not only the erasure of women’s contribution in STEM who take up credit for women’s work, but also the crowding out of women altogether by men who want to take credit(or give credit to other men) for women’s work, while finding covert, and often nefarious ways of avoiding backlash. Take for instance, the large number of women inventors from history to date, whose work has either been stolen by men, or credited to men to the extent that Margaret W. Rossiter coined a term called the “Matilda Effect” after American Suffragist and feminist critic Matilda J. Gage. Rossiter maintained how “women scientists who have been ignored, denied credit or otherwise dropped from sight that a sex-linked phenomenon seems to exist, as has been documented to be the case in other fields, such as medicine, art history and literary criticism.”

In Devternity itself, female participation was considered important, but only so far as it was mediated, filtered, and ultimately deleted in favour of male participation in the conference, to the extent that the moderators and owners undertook the absurdist route of replacing female participation altogether. Were the organizers unable to invite women speakers due to a dearth of women techies? As per Julia’s X account, out of 14 invited women speakers in 2021, 3 had confirmed and 2 had cancelled. Did the 3 include the fake speakers? From the changes made to the website after the organizers got called out for the lack of diversity, that seems to be the case. Moreover, given that most of the invited speakers at Devternity were from Big Tech companies (the GAFAM group), it is difficult to back up that claim when women techies make up between 29 to 45 percent of the total workforce at these companies. It could be argued that due to the preference for speakers higher up in the leadership hierarchy, the organisers found it difficult to find suitable candidates to invite as speakers, since gender diversity cannot be the only factor in shortlisting speakers. While there is not much clarity at the exact metrics that were being utilised by the Devternity organizers to shortlist and invite speakers, one thing is certain – Devternity undervalued the contributions of the majority of women techies in the field, and deemed them to be incapable of occupying the same stage as the rest of the male speakers.

This treatment meted out to women techies is unfortunately, not particular to Devternity or its organisers, but a symptom of a larger problem of patriarchal capitalism in tech. As reported by the vast majority of women in this field, tech suffers from a severe lack of diversity, equity and inclusion (DEI), with women facing greater hostility the farther up the ladder. The presence of uneven gender distribution in certain course disciplines and fields of study, particularly, science technology, engineering, and medicine in higher education is also an issue that necessitates deliberation. The reduced presence of women in fields affiliated with technology encourages “hegemonic presence of androcentric and sexist values” in knowledge accumulation and in the development of tech applications and products.

Globally, women are also found to be remunerated less than their male counterparts, are seen to be in charge of fewer senior positions and their lower engagement/participation in STEM fields result in fewer jobs. A case in point, the newly-reconstituted board of OpenAI , after the reinstatement of Sam Altman, consists of an all male lineup shedding light on the stark gender lop-sidedness of women in the field of Artificial Intelligence. Thus, due to the rampant sexism that women face in these fields, their penetration into greater participation is impeded, which in turn creates an environment where women are largely absent, and their ideas and concerns drowned out. While this in no way indicates that male ability and contributions are necessarily greater, but simply that they are accorded greater credence and weight.

It is essential to highlight the advantages and significance of allyship with women in the field of development of AI products and applications considering how extremely pivotal their impact on society is. Especially if there is a possibility that the outcomes generated by AI applications are observed to be tainted by the limitations of persons designing these. The training data fed into the algorithm may be under representative of the sample population, which can exacerbate the issue of bias. Obtaining representative data that conforms to the target population is often a difficult task and this process tends to generate skewed results and biased outcomes. A classic example is the furore caused by the recruitment algorithm employed by the tech giant Amazon a few years ago, where the algorithm showed an affinity towards preferring male candidates over women based on certain parameters inherent in the training data and the usage of certain word embeddings, preventing it from being gender-neutral.

Legal and Policy response to this problem has been to require mandatory quotas for women, with companies largely being left to decide how to implement or meet these targets, leading to a form of ‘tokenization.’ A 2011 French Law called Copé Zimmermann or the ‘French Gender Quota Law’; required all listed and non-listed companies with revenues or total assets of over EUR 50 million or employing over 500 persons for three consecutive years to reach a gender quota of 40 per cent by 2017, with a first step of 20 per cent by 2014. Similar quota-based legal requirements also exist in other EU countries such as Spain, Norway, Italy and Germany. A similar requirement also exists under the Indian Companies Act, 2013 wherein listed companies are required to have at least one woman member in their boards. Unfortunately, companies have resorted to hiring individuals who tow their line, or female family members to fulfil this requirement, leading to a lack of actual and active female participation. Quotas have led to male employees and potential recruits viewing, and treating women employees and other recruits with hostility for being under qualified and securing the job only due to the quota rather than of their own merit; rather than looking at them as corrective measures against the gender disparity in workplaces. Women are still not being provided a space for them to actively exercise their agency, and end up being only a mouthpiece for whatever the majoritarian male-dominated board decides. Is there a way to indeed increase women’s participation in tech? The answer, as opposed to a new legal or policy measure, might actually lie in popular culture.

It’s a Man’s World (In Tech and Beyond): Why we Need a Bechdel Test for Tech

While legal and policy measures are excellent for providing extrinsic stimuli that compels companies to change, what is often necessitated is a change of company and societal culture. Predominantly, leaders and owners in Big Tech do not recognise women’s physical and cognitive absence as a problem, leading to a culture that uses diversity just to score points with the public. Women who are in the tech arena are required to rise up to the bromance culture in tech, rather than celebrate the diversity in perspective they bring to the table.  In order to reinstate women the power that comes with their position, we urgently need a Bechdel test for determining the veracity of female participation. Before that, a short primer on  what is the Bechdel test.

In 1985, cartoonist Alison Bechdel laid out an informal “test” for measuring the active participation of women in films and television. The said test had three parts or “rules”: firstly, whether there were at least two women; secondly, who had a conversation with each other, and thirdly, about something other than men. While Bechdel had not intended it to be an actual test for measuring the participation of women, the test continues to be used to this day as a measure of the active participation of women. Much like us, other tech researchers and scientists, including Kate Crawford have noticed the glaring disparity between men and women in tech, and pondered whether it was indeed time for a Bechdel Test for tech.

A variant of the Bechdel test already exists for gauging the participation of women in coding in particular – the Bechdel coding test. In order to pass this modified test, a software development project needs to a) possess at least one function, b) written by a woman developer that c) seeks out another function coded by a fellow woman developer. For a field that is largely dominated by “tech bro” culture– that treats women as outsiders and demeans their contribution despite a parity in aptitude and skills- the Bechdel coding tests creates a relevant benchmark that acknowledges the gender imbalance by placing the focal point on women. However, this test is also largely limited to specific situations and industries. For example, the development of software using different coding languages by women developers on the team was insufficient to satisfy the test if the women developers did not “call” each others’ “functions”. Calling a function would essentially ensure “interaction” between the programs developed by each developer. Moreover, it also did not include the interactions between other roles within the team or organization that were led by women- such as visual designers, content designers and UI/UI developers, among others. As a result, many companies and teams may not qualify under this test, leaving a gap in ensuring women’s participation in tech across the board.

We suggest a modified Bechdel test, ‘The Bechdel for Tech test’ for providing a quantitative threshold for ensuring participation in tech herein, that serves to provide a relevant benchmark to assess the participation of women in diverse subfields and instances. The test has three main components:

a) In each discussion space- panel/meeting/team/conference-there must be representation of at least 50 % women; b) who work with- i.e. speak/contribute/build/ideate with each other (or other women within the team org.) in at least 30% of the speaking time or in 30% of the project documents; c) on an idea/innovation/concept/project related to development in the tech sector.

The test can be tweaked further within each specific sub-industry, based on the needs and requirements in a particular field. While the dynamics of the industry in question is indeed important in determining the quality of participation of women, the test is meant to provide a threshold, one that highlights the lacunae of formal equality in ensuring women’s participation. Let’s take the example of a Fintech company which consists of various teams such as software development, compliance and legal, UI/UX Research, Social Media, etc. The ‘modified bechdel test’ suggested above is to be applied not just at the macro level, that is to the overall organization to ensure coverage of cross-team interactions and contributions;  but also at a micro-level, within teams themselves, to ensure that in meetings, discussions and other work-related spaces, women are not only represented but provided a suitable space and atmosphere to develop, interact, and innovate.

Conclusion

In the field of technology, where developments are taking place at a rapid pace, lack of diversity and inclusivity can further perpetuate gender disparities. While law and policy are both important tools to embed formal mechanisms that ensure women are provided a seat at the table, it is ultimately institutional culture and attitudes that impact how minority employees are treated, and their talents utilised. Whether as participants in tech conferences, consumers of tech products or even employees working within an organization, the modified Bechdel Test suggested herein provides a litmus test to check whether companies, teams, projects and discussion panels are worth our time, attention and money.

*The authors are Academic Fellows at National Law School of India University, Bangalore, who have jointly co-authored this piece.