From harassment allegations at Google to revelations of the biases encoded in artificial intelligence algorithms, Silicon Valley’s sexism has been thrust into the public eye. Just last month, MIT Media Lab revealed that Amazon’s AI facial recognition software has trouble identifying female and darker-skinned faces. In October, Reuters revealed how AI recruiting technology tends to favor male candidates, since it is developed and tested using men’s resumes.
This all raises a central question: Where are the women?
Actually, they were initially at the forefront of the industry, back when technologist jobs were considered menial, akin to typists. But as the industry became profitable, male executives developed hiring criteria and workplace cultures that sidelined women. So instead of a space that empowered women, the Internet’s business structures made it a sphere that reinforced masculine biases and patriarchal norms.
The tech industry’s male bent was by no means preordained. The group of programmers selected to work on the U.S. military’s first computer in 1946 was more than 50 percent women. One of the most celebrated early engineers was Grace Hopper, a Navy admiral whose programming enabled the United States to model the impact of atomic bombs. It was also a woman, Margaret Hamilton, who led the coding team that charted Apollo 11’s path to the moon. The conditions were far from glamorous, but the female programmers were known for their meticulous work ethic and attention to detail: Hopper famously identified the first-ever computer “bug,” tracing a glitch back to a moth trapped in a relay wire.
As the industry grew more lucrative, engineering jobs became higher-status and better paid. In 1950, the United States had only two electronic computers; by the late 1960s, electronic computers formed a $20 billion industry. With the sector rapidly expanding, companies were hiring quickly, but were unsure of what qualities to look for in their employees since coding was such a new skill. In “A History of Computing in the Twentieth Century,” IBM researcher John Backus writes: “Programming in the 1950s was a black art.”
In an effort to demystify that art, software company System Development Corp. (SDC) contracted psychologists William Cannon and Dallis Perry to create an aptitude assessment for optimal programmers. Cannon and Perry interviewed 1,400 engineers — 1,200 of them men — and developed a “vocational interest scale,” a personality profile to predict the best potential programmers.
Unsurprisingly given their male-dominated test group, Cannon and Perry’s assessment disproportionately identified men as the ideal candidates for engineering jobs. In particular, the test tended to eliminate extroverts and people who have empathy for others. Cannon and Perry’s paper concluded that typical programmers “don’t like people,” forming today’s now pervasive stereotype of a nerdy, anti-social coder.
Because of the sway SDC had in the sector — it claimed it “trained the industry” — its programmer profile came to shape industry demographics for decades. It was a cycle: The hiring process favored men, so men became overrepresented in technology companies, feeding popular perceptions of engineering as a masculine domain. Beginning in 1984, the number of female computer science majors began to decline steadily, from 36 percent to 18 percent today.
This stereotype of the ideal coder as innately genius rather than hard-working and well-trained has remained powerful in the tech industry. A 2015 study in Science found that computer science, more than other fields, places a premium on inborn brilliance, something considered a disproportionately male trait. Today, the field of U.S. software engineers is 80 percent men.
Explicit sexism has exacerbated these assumptions to create a hostile workplace for women. Beginning in 1973 at the University of Southern California, entry-level computer science classes used a nude image of Playboy centerfold model Lena Soderberg to teach engineers how to turn physical photographs into digital bits (the original jpegs). This became standard practice for computer science departments worldwide: Soderberg’s picture was the most widely used photo in image-processing research, and the model was dubbed “the first lady of the Internet,” a position she maintained as recently as 2015.
Recent research indicates tech’s misogynist culture stubbornly persists. In 2015, a group of female investors and executives surveyed 200 senior-level women working in the technology sector. Sixty percent of the respondents reported unwanted sexual advances in the office.
Tech’s leaders have long positioned themselves and their industry as a disruptive force, a group of mavericks reshaping the way we live, work and connect. Yet when it comes to gender roles and representation of women, the industry has mostly tended to reinforce our status quo.
Clearly, that won’t change until women are better represented and respected in the industry. Organizations like Girls Who Code are doing critical work to boost female representation among engineers. But given the decades of male leadership in the sector, is it possible to transform the underlying masculine sense of purpose and ethics driving technological development?
Maybe. Consider for example, the work done by VNS Matrix, which was created in 1991. An Australia-based collective of artists, its founding documents called on technologists to “remap cyberculture with a feminist bent.”
VNS Matrix emerged as part of the cyberfeminist movement, a branch of third-wave global feminism that grew at the end of the 20th century. Cyberfeminists saw the growth of the dot-com bubble and worried that tech, as an epicenter of money and power, would become even more overtly misogynist. They urged technologists to rethink their perspective on the Internet’s role in society and create devices explicitly intended to disrupt patriarchal norms. Members of VNS Matrix, for example, built a video game in which all players were genderless and invited to battle the sexist “Big Daddy Mainframe.”
Most important to cyberfeminist thinkers, especially Sadie Plant in her seminal work, “Zeros and Ones,” was that engineers reject the antisocial stereotype and consider the social impact of their work, its effects on human bodies and relationships.
The cyberfeminist movement remained on the fringes of the tech sector. Media professor Helen Hester argues that members did not adequately translate their theoretical findings into solutions for the most urgent technological threats, like sexual harassment on social media. Today, though, initiatives are beginning to translate cyberfeminist theory into concrete action steps for technologists.
One example is the Algorithmic Justice League, which is committed to exposing algorithmic bias and ensuring accountability to vulnerable communities during the design and deployment of coded systems. AJL is demanding that coders consider the gendered and racialized nature of their work from the earliest phases of product and code creation. Effectively, they want engineers to examine how their technologies will impact human bodies, especially women of color.
AJL’s mission, and that of other cyberfeminist initiatives, isn’t simple. Silicon Valley’s sexism is deeply entrenched. But ignoring the core principles that shape technological development isn’t an option. Technology has become too central to our lives for its underlying ethics to be treated cavalierly. What worldview lies at Silicon Valley’s core? Are its leaders seeking to reinforce age-old gender norms, or disrupt them? Until we examine these questions, we’ll address the symptoms of technological patriarchy, but not its roots.
correction: An earlier version of this article misidentified the software firm System Development Corp. as the Saigon Development Corp.