(Summer 2014) Chelsea Barabas, G
Chelsea Barabas (G, Comparative Media Studies)
Chelsea spent her summer in San Francisco, where she worked with an organization called CODE2040 to design and study interventions that support pathways into the tech sector for Black and Latino/a college students. During her time with CODE2040 Chelsea conducted fieldwork to understand how informal learning experiences and emerging hiring practices translate into meaningful career opportunities for minority software developers. The goal of this work is to identify novel strategies for bridging learning to upwardly mobile careers, particularly for individuals who face significant barriers to accessing our current institutions of higher education.
Blog 4: Reflections
This year Google released statistics on the demographic make-up of it’s employees, revealing how largely white and male their employment base actually is. The numbers weren’t great, but Google’s transparency led to a domino effect in the industry as several other prominent tech companies in Silicon Valley subsequently came forth with their diversity numbers. Unsurprisingly, the numbers across the board pointed to a widespread underrepresentation of women and people of color within tech.
For many years, tech pundits have proclaimed the democratizing potential of free and open online resources, which enable anyone to develop the skills necessary to find success and opportunity in the industry. However, this characterization is beginning to change in light of the startling homogeneity of tech’s workforce. The release of these numbers has sparked a growing interest in problematizing popular notions of meritocracy in tech.
As the conversation around diversity in tech has grown, I have become increasingly interested in understanding how the industry understands and implements concepts of objectivity and merit within their workforce. With the support of PSC, I was able to work with an organization called CODE2040, whose flagship program is to connect highly qualified Black and Latino/a college students with engineering internships in Silicon Valley. Through my work with CODE2040 I sought to explore the following questions:
1. How does the industry diagnose tech’s “diversity problem?”
2. What are the pathways that underrepresented individuals take in pursuit of careers in tech?
3. How do these pathways map onto the metrics and processes used to identify qualified employees in the industry?
In other words, I was really interested in understanding who/what are the gatekeepers to opportunity in the industry. I was particularly excited to work with CODE2040 because they have intentionally positioned themselves as a gatekeeper organization for minority talent. Over the last three years they have focused on building out a network across the country to recruit qualified Black and Latino students. This network is now starting to pay off, as more companies approach CODE2040 in the hopes of hiring one of their Summer Fellows. My work with CODE2040 gave me the opportunity to conduct interviews and observations with their summer fellows, CEOs and head recruiters of technology companies, and industry leaders who are driving the conversation about diversity in tech.
The Favorite Diagnosis: Leaky Educational Pipeline
By and large, the most popular diagnosis of tech’s limited diversity is the educational pipeline problem: for a myriad of reasons, women and minorities lack the resources and supports necessary to develop marketable technical skills.
As such, most leaders described their role in supporting diversity as auxiliary, by donating time and money to educational coding programs that specifically target girls and urban youth. This explanation echoes an age-old narrative about the essential role that education plays in addressing issues of inequality and upwards mobility in society. While there is certainly a lot of value to be gained from strengthening the educational pipeline into STEM careers, this explanation stops short of acknowledging the leaky gaps further down the pipeline, the gaps between workforce preparation and economic opportunity. It fails to explain why roughly half of the minority college students who obtain degrees in computer science go on to work in industries other than tech. And it conveniently shifts responsibility from problematic industry practices to education reform.
However, a large body of research provides insights into the stubborn biases that continue to persist in our labor market. Work in the fields of psychology, sociology and cognitive science have unearthed the heuristics, or rules of thumb, we use in order to simplify our messy, chaotic world in order to make decisions. While these heuristics may facilitate daily decision-making, they are inherently limited. In the context of hiring a new employee, the heuristics can be as arbitrary as what zip code you live in or the racial identity associated with your name. While these rules of thumb may be effective in facilitating decision making to a certain degree, they clearly have the potential of systematically offering opportunity for some groups more than others.
When I interviewed CEOs and recruiters this summer, many of them expressed frustration at their inability to get women and people of color to apply to the jobs their company posted. They assumed that there simply weren’t enough females and minorities with the necessary skills to apply. However, I wanted to push past this assumption and dig into the heuristics that they deploy in their hiring process.This would enable me to understand how their practices were connected to certain gatekeepers at the top of the hiring funnel, shaping who even applies in the first place. What I found were recruitment practices that were very tightly coupled with specific gatekeepers: reliance on social networks of friends and colleagues, recruitment from elite universities and poaching top talent from well-known companies.
For years, recruiters had been using these methods to find reliable, skilled employees. However, in recent times these methods have been inadequate – Silicon Valley social networks are tapped dry, salaries are climbing from relentless poaching, and pressure is mounting for companies to diversify their employee population.
Amidst these developments, a new set of gatekeepers has emerged who are eager to develop more efficient and far-reaching methods for recruiting talent. These companies are developing algorithmic recommendation systems that scrape data from the Internet in order to passively construct profiles of potential employee candidates. These systems go beyond static resume databases like Linkedin in order to identify people who haven’t necessarily opted-in to participate. The makers of these algorithmic recommendations claim that their services will enable companiesto extend their reach beyond the insular and homogenous tech community in Silicon Valley. Furthermore, they hope to decouple initial evaluation of competency from university degrees. Since much of the in-demand skills in tech are not traditionally taught in formal computer science programs, a lot of practical industry learning happens via online channels. These algorithmic developers hope to capture online learning in order to provide a higher resolution profile of an individual’s skill set.
Perhaps most interestingly, some thinkers in this space are interested in using big data methods to broaden and redefine the metrics that are relevant during the recruitment process. For example, it may be possible to evaluate factors such as an individual’s learning mindset and motivation based on their behavior online. These characteristics have historically been very difficult to capture using metrics like a university degree or resume. Some researchers are exploring ways to gauge these attributes through observation of people’s behavior online.
However, there are clearly some risks involved with the emergence of these new gatekeepers. Increasingly, researchers at the intersection of big data and social practice are urging us to push past the notion that “the numbers speak for themselves” and explore the limitations and biases that are baked into these algorithms. Nick Seaver has called for the study of “algorithmic systems,” citing the need to understand the cultural details that shape the technical formation, implementation and iteration of algorithmic decision making processes. Alistair Croll has warned that algorithms will become the battleground for 21st civil rights issues, particularly as they relate to decisions that systematically deny opportunity to specific groups of people.
As I come back to campus to start a new semester at MIT, my mind is swimming with the rich experiences I had this summer. Not only was I able to explore my initial research questions, but I also discovered domains of interest that I was not aware of before doing my fieldwork. Of particular interest for me moving forward is exploring the ethical and legal implications of emerging practices in algorithmic recruitment. I am also eager to share my research with the broader community of organizations and advocates who are working to increase diversity within tech. I am in the process of developing a plan to distill and disseminate key insights with CODE2040 and their broader community. My hope is to push forward the conversation about opportunity and access in this growing sector by identifying widespread assumptions regarding the origins and potential solutions to homogeneity of the tech workforce. Thanks so much to the E. Eugene Carter Foundation, for funding my work through the PSC and making this experience possible!
Blog 3: Informal Sunday Gathering Yields Thoughtful Insights
This summer, part of my job has been to attend most of the supplementary programming events that CODE2040 organizes for their summer fellows every week. These include a wide variety of activities, such as professional development workshops, informal networking opportunities and job fairs. Each week the organization hosts 1-3 such events for the fellows in the evening. This means that, in addition to their full-time jobs, fellows are spending several hours a week doing CODE2040-related activities. That’s a lot of time!
In spite of their jam-packed schedules, a small group of fellows organized an informal gathering to talk more intimately about how themes of diversity in tech affect them on a personal level. This was a rare occasion in which all authority figures and talking heads were gone and the fellows were able to talk in a casual, frank way about topics that mattered the most to them. Although I’m not a fellow, they were kind enough to invite me along for the conversation. On one warm Sunday afternoon, we gathered on the floor of a friend’s apartment and had a wonderfully rich and varied discussion on the role that diversity plays in their academic and professional lives.
The initial premise of the conversation was to talk about diversity in higher education. Throughout the discussion several students talked about the ways that being a minority shaped their experiences studying computer science in school. For instance, one student explained the heightened pressure he felt when answering questions in class. As the only Latino in most of his STEM classes, he was afraid that he might be seen as a representative for Latino aptitude in computer science in general. As a result, he was hesitant to raise his hand unless he is very certain that his answer is correct. Another student talked about the challenges she faced as one of the only girls in most of her classes. She described how challenging she found it to enter conversations that were dominated by male students eager to demonstrate how much they knew about a given subject. At the same time, she preferred these situations to programs and experiences that were designed specifically for female students, because they were more representative of the real world challenges she would face out in the workplace.
We also discussed the challenges the fellows faced when explaining their participation in CODE2040 to their friends from school. Many of them found it worthwhile to frame CODE2040’s work within larger debates regarding the value of a diverse workforce in tech. One fellow described diversity in tech as an opportunity to gather a varied set of perspectives on a given subject. If you grew up surrounded by people from non-white cultures or had mostly female friends as a child, he argued, you will have a different perspective on “what makes sense” during the development of a new product. Another student talked about the increasing wealth and opportunity gap in America. He argued that paving the way for more minorities into tech was key to addressing problems of economic inequality.
However, one young woman argued that none of those rationales really mattered to people who didn’t already value diversity in the work place. She preferred incentive-based arguments that highlighted the economic potential of building products for minority consumer bases that are currently untapped in the tech market. Blacks and Latinos comprise a very large percentage of the US population. Yet, there are few technology products that are catered specifically to their lifestyles, interests and needs.
I mostly listened during the two and a half hours we were gathered there. For me, it was refreshing to hear the fellows talk so thoughtfully and frankly about the struggles they encounter on the path to promising careers in tech. Although I have spent a lot of time observing and talking to the fellows one-on-one, we have had very few opportunities to hold extended conversations as a group. This experience gave me an opportunity to see the fellows working through tough issues together, pushing one another to think critically about how their personal experiences fit into the bigger picture.
Blog 2: Hack 4 Diversity
This past weekend I attended CODE2040’s second annual Hack 4 Diversity, a weekend-long hackathon focused on addressing issues of diversity within tech. CODE2040’s summer fellows, along with some amazingly helpful and inspiring members of the broader community, spent the weekend working on projects to promote diversity in a sector that is still comprised largely by white males. Teams explored a broad range of challenges that minorities and women face along the path to entering the professional world of tech. While the by-products of the weekend are still in the early stages of development (hackathon style!), the ideas that inspired each teams’ work reflect some provocative and nuanced perspectives on “the diversity problem” in tech. Below I describe some of the great ideas that the teams developed over the weekend.
Transdocs is a translation/editing platform that enables individuals to translate, edit and share technical documentation in non-English languages online. A text file can be easily uploaded to the Transdocs site, where an initial translation will happen using a free translation API. The user can then improve the translation by clicking directly on the document in the web browser, making and saving edits for the benefit of future users. The idea was born from the team’s desire to make technical documentation more accessible to non-native English speakers who are interested in learning to code. As CODE2040 fellow Christian Rodriguez explained, at his home university in Puerto Rico there are many students who are intimidated out of studying computer science because it requires strong English language skills. The Transdocs team took on this challenge by creating a tool that would take people beyond basic google translations by promoting crowd-sourced translations. Transdocs addresses a really important issue within the programming community: how do we promote communities that aren’t centered around the English language? I’m curious to learn if there are any other projects that deal with making code documentation more accessible to non-native English populations.
This team built an app-based platform that lets you curate photos from professional experiences and index them using hashtags associated with particular skills and professional themes. Employers can then use the same hashtags as a search tool in order to find people with the professional background they need. This team sought to create an alternative to the traditional resume by reimagining the ways we could categorize and index what’s valuable or “marketable” on the job market. At the end of their demo on Sunday, the team explained that the name for their project (‘One Thousand’) was inspired by the phrase, “A picture is worth a thousand words.” The team wanted to explore novel ways of displaying information about one’s career beyond what’s typically found on a traditional resume.Their project reflects a broader trend going on in higher education right now as companies like Degreed explore ways to capture and validate learning that happens outside of a formal classroom setting. The concept of “micro-credentialing” has grown in popularity in recent years as more people engage with online learning resources in order to develop new competencies and skills in their field. If micro-credentialing platforms gain traction, they could provide alternative avenues to rewarding careers for students who can’t access our current institutions of higher education, but who engage in other types of rewarding professional development experiences.
This team’s project was inspired by Debbie McGhee’s experiments in measuring implicit bias. In her research, McGhee asks people to make associations between two sets of images and positive/negative attributes in order to capture the hidden biases people have about certain groups of people. The Bias Analytics team wanted to create a similar experience that would enable recruiters and HR personnel to measure and understand their own biases in terms of race, gender and educational background. They created a site that pushes the user to make rapid decisions between two resumes with comparable credentials, but which varied along one of the three variables mentioned above. After a rapid series of comparisons, the site would provide you with a sense of your own biases. This reminds me of a number of other projects within the Center for Civic Media, such as Open Gender Tracker, which collects metrics that enable newsrooms to gain a better understanding of gender diversity within news publications. The promise of these applications is that they provide the baseline information necessary for people to become aware of their own biases and take initial steps to implementing more fair decisions and practices in their work.
TechFindr and Hacknight
TechFindr and Hacknight were two projects that sought to support a learner’s burgeoning interest in computer programming within communities where few people are involved in the tech industry. The Techfindr team created a mobile application designed to help people identify tech events in their local area. The project was inspired by the experiences of a CODE2040 fellow named Fidelis, an MIT undergrad from Zimbabwe who said that, though he was always interested in the STEM fields, he knew very little about computer science before he came to the States for school. His team wanted to facilitate the discovery of coding events for people who may not as easily come across such experiences within the communities they live. Similarly, the Hacknight team created a social platform that enabled kids to find collaborators for DIY projects in their neighborhood (like building a spaceship!). Even in neighborhoods where there are few other kids interested in, say, doing an Arduino hack, the app could be used to alert adults to a child’s interests and then used to create supports for that child to explore her interests more. As one of the fellows on the Hacknight team explained, “If you don’t have a pre-existing network, you don’t know how to find out more about these things.” Hacknight and Techfindr were built to address this challenge.
Dynamic was created to customize resumes for a particular job opening. This tool was specifically built for users who don’t have many people in their network who can help them craft and revise a winning resume. The team also created an optional, “anonymize” function which would allow someone to hide aspects of their identity, such as their race, gender, neighborhood, etc. on their resume. There is well-documented evidence that removing such information from resumes can decrease certain forms of bias during the recruitment process. This idea sparked some interesting debate at the hackathon. “I want people to know who I am. I’m proud of my name,” mentioned one CODE2040 fellow to me during the presentation. Others discussed the risks and alternatives to “white-washing” resumes: was this a helpful tactic for getting one’s foot in the door? Or was this a work-around to a much more fundamental problem that needs to be addressed head on? How do we balance assimilation with authenticity in these kinds of situations? I am interested in learning about more tactics that people have used to address bias during the recruitment and hiring process. Please share!
As you can see, theses teams touched on some of the most complex and challenging issues related to diversity within tech. Their projects reflect some of the awesome insights that we can gain from listening to minority students who are currently in the process of paving their own careers into tech. If you are interested in learning more about these projects or teams, please let me know and I will try to put you in touch with a the fellows responsible for creating this awesome work.
Blog 1: Overview of Summer work
A couple of months ago, I received a Summer Fellowship grant from MIT’s Public Service Center to fund a project in collaboration with CODE2040, a nonprofit organization that creates career pathways into the technology sector for underrepresented minorities. Recently there has been a growing concern over the lack of diversity within tech. While 57 percent of the professional workforce is comprised of women, they hold only 25 percent of the occupations in computer and engineering occupations. The numbers get worse when we zero in on the startup space, where 87 percent of founders of internet companies are white. Asians comprise 12 percent of the population, leaving only 1 percent for Blacks, Latinos, Native Americans and other groups. Prominent minorities within tech have argued that the lack of diversity creates a “boys club” mentality on the job, which perpetuates an inhospitable environment for employees who don’t fit the hoodie-wearing white male stereotype of Silicon Valley.
I will be working with CODE2040 to research educational interventions designed to increase access to careers in tech for underrepresented groups. This topic is particularly interesting given the surge in interest in open online educational resources, such as MOOCS (massive open online courses), which some have proclaimed will revolutionize higher education. The poster child of this movement has been Battushig Myanganbayar, the Mongolian teenager who was admitted to MIT last year after acing a MOOC on circuits and electronics from the university. MOOC enthusiasts cite Battushig’s story as proof of the democratizing potential of online educational resources. For them, Battushig represents the powerful potential of making quality educational content widely available online, providing opportunities for the best and the brightest from around the world to learn and be recognized for their talents.
However, this framing of Battushig’s story fails to recognize the importance of a few key details. For instance, the principal of Battushig’s high school, Enkhmunkh Zurgaanjin, was the first Mongolian to ever graduate from MIT. Upon graduation in 2009, Mr. Zurgaanjin returned to Mongolia with a vision of strengthening the STEM education in his home country. In addition to hosting the MOOC at school, the principal purchased lab equipment and recruited a Stanford PhD candidate in electrical engineering to tutor a small cohort of students during the course. While there is no doubt that Battushig is an exceptionally bright pupil, these details are hugely important when we begin to think about the circumstances that shaped his ability to use these online resources to pursue his educational goals. However, these circumstances are frequently overlooked in the debates regarding the role of technology in the future of education. Focus tends to be placed on emerging platforms and open resources, rather than on the circumstances in which those technologies become transformative for students. The risk of this kind of discourse is that it promotes the illusion of an objective, merit-based system, in which talented and hardworking students rise to the top using resources free and available to all.
This kind of thinking has crept into the debate regarding diversity within tech as well, where pundits describe their field as a meritocracy in which anyone who puts in the time can make a dime. This sentiment is captured well in a statement made by Internet entrepreneur Jason Calacanis, who last year was quoted saying, “The tech and tech media world are meritocracies. To fall back to race as the reason why people don’t break out in our wonderful oasis of openness is to do a massive injustice to what we’ve fought so hard to create. It flies in the face of our core beliefs: 1. anyone can do it, 2. innovation can come from anywhere and 3. product rules.” Such statements are reinforced by books and newspaper articles which depict the lone entrepreneur/innovator who drops out of college to pursue “the next disruptive idea.” This too-cool-for-school narrative implies that raw talent is far more important than a piece of paper from a university, which can be appealing when one considers the mounting costs of attaining a college degree. As student debt continues to rise, there is a growing interest in the expansion of alternative pathways to promising careers via online educational resources. This is particularly true in the tech sector, where online learning platforms like Treehouse, Code Academy, and Udacity, advertise the promise of future prosperity for any hardworking individual looking to make a career change (https://www.youtube.com/watch?v=zeGT4UArCdc) .
Yet, this kind of talk gets uncomfortable when we think about just how white and male the tech sector is today. Hypothetically speaking, if anyone with interest and reliable internet access can learn to code within a short period of time, does that mean that students who are female, Black, or Latino/a simply lack interest or aptitude for these subjects? Of course not. A more productive starting point may be to map out the the educational pathways that individuals have taken in pursuit of full-time work in the tech industry. Only then can we understand the barriers and opportunities that shape who ultimately pursues careers in this field.
I’ll be spending my summer doing just that. In collaboration with CODE2040, I am investigating potential strategies to strengthen the educational pipeline for minorities and women pursuing careers in tech. I am particularly interested in the social, cultural and economic factors that shape one’s interest and ability to learn the skills necessary to work as a programmer or software engineer. Below are some of the key questions I am exploring:
Formal vs. Informal Education: What is the educational path that people tend to take in order to secure careers in tech? What role do experiences outside of a traditional classroom setting play in developing one’s skills? What are the social and economic factors that shape one’s ability to engage in semi-formal learning experiences, such as hackathons, online coding courses, etc?
Evolving importance of college degree: If universities are not where job-ready skill development occurs, what importance does a degree hold in today’s tech world? Could it be that degrees are important for other reasons, such as a reliable indication of one’s “ability to learn”?
Social connections: What role does one’s social networks, particularly their “weak ties,” play in getting a job in tech? What are the sites where these weak ties are formed and one’s reputation as a programmer are developed?
Power and the negotiation of value: How does the negotiation happen between learners, educational providers and employers about what’s important to know and what counts as a valid demonstration of competency in this sector? How is the power of credentialing authority shifting in these emerging eco-systems of informal learning? Who would stand to gain or lose if this system were to change?
My hope is that this research will provide key insights into the factors that we should consider when designing more vibrant alternatives for learners interested in careers in tech, but who face significant barriers to pursuing a traditional university degree. I look forward to sharing more of this work with you in the future.