There are many challenges faced by civil society and its institutions. This discussion highlights just a few of those issues and focuses specifically on challenges to the responsible and effective use of data in the sector.
In the social sector, there is a higher bar for success than there is in other sectors. There is a greater need for fairness, accountability and transparency when working with vulnerable populations, in social services, and in domains such as criminal justice. The cost of failure is higher, and incrementalism can be more damaging than beneficial.
When I was chapter leader at DataKindSF, we would often get inquiries from local universities and schools; they were looking for “social good” projects through which students could practice basic skills. They hoped to use engagements with local nonprofits and charities as a way of showing real world experience on resumes, typically when applying for jobs at “real” companies.
While I sought ways to support good intentions and opportunities for exposure, I also saw this as frustratingly backwards. For profit jobs, in most cases, are simpler, more compartmentalized, and come with structure, training and mentoring opportunities. They utilize data that are more trustworthy and more complete, and apply them in problem spaces that have simple causal chains that are easy to understand.
Perhaps, instead, we should see an internship at a for-profit corporation as a way to learn basic skills in preparation for a job at a non profit?
Challenges and issues that may cause a data initiative to fail:
- Funding runs out, and the team is unable to secure new funding.
- Team members leave the project. This type of instability can be caused by lack of career options for the team, poor or insufficient leadership, or simply because students graduate. Bus factor risk can exacerbate the impact of team turnover.
- Leaders lose interest. Inventive structures and motivation are skewed towards innovation and building, especially for leaders in academic institutions. Many leaders aren’t interested or well-suited to managing teams, sustaining operations.
- Takeover: The organization or project is absorbed, usually by a funder or sponsor. This can be subtle, as when board seats are captured to redirect resources.
- The absence or instability of legal structure can be the downfall of initiatives. For example, when the initiative has no clear owner, or when there are problems in attaining desired legal status and fiscal sponsorship.
I’m intentionally and dangerously grouping under the term “civil society” a plurality of organizations and initiatives that are completely different from each other, and that face very different challenges.
Data tools in academic contexts and universities Many tools to address social, climate and environmental challenges are created by academic teams. These teams have strong domain knowledge, understanding of the relevant datasets and models, and their work is connected with other research teams through networks and peer review. They are innovative and constantly breaking new conceptual ground.
However, most such teams:
- Lack proficiencies to translate proven ideas into effective tools and products, skills like user-centered design and software development. To make these tools sustainable, they need experience in developing organizational structure, business strategy, and team building.
- Are not incentivized to acquire such skills. They are often strongly hierarchical teams, with leaders that are motivated by novel work, peer recognition and publishing research. While many teams strive to develop tools with real and lasting impact, they are still constrained by these incentive structures in early phases and often even after they spin-out into independent organizations.
- Put in terms of life cycle of risks academic teams tend to focus first on technical risk, and secondarily on impact risk, and few of them are equipped to think about adoption risk and scale risk.
Some of the ways in which we can improve success rate of data products in these contexts is through more informed grant-making, “innovation lab” programs that specialize in civil society data products, expanding career opportunities for students and post-docs, and by expanding the scope of on-campus support like the tech transfer office.
Data systems for journalists, activists and for civil discourse There is increasing awareness of issues in this space, due to work by Edward Snowden and the Freedom of the Press Foundation, and global movements like Black Lives Matter and the Arab Spring. These highlight the ways in which digital technologies intersect with protest and free democratic voice, and underscore the nature of the threat models for these actions. These conditions highlight the need for independent and trustworthy technical infrastructure for civil society data tools ( Pillar Three.
Journalists are also creating new types of data products. Consider
- The COVID Tracking Project, which emerged as an independent and vital source of information about the pandemic at a time when the CDC and other government institutions were unable to lead and when public trust in them was low. Started by reporters at The Atlantic and run largely by volunteers, this was a quintessential civil society effort until the Biden administration was able to transition. CTP helped create standards, provided recommendations, and provided data at a time when COVID dashboards were viewed by the public more often than the weather.
- The emergence of data journalism. About much more than building interactive visualizations, this growth represents the role that data is playing in political conversations and in guiding public opinion.
- Journalists and activists are driving the generation of new datasets to increase transparency, awareness and direct action; data about infrastructure and water supply, energy sources, carbon emissions, climate impact, property evictions, and more.
Ethical issues in data systems
When chased blindly and without understanding, data technologies have been shown to be deeply problematic. They can reinforce power imbalances, exacerbate biases and fallacies in assessing value, and they may make invalid assumptions, especially in socially complex domains. Over time, they can make permanent and opaque societal and racial prejudice.
There are many organizations doing great work in understanding and highlighting these issues, such as Data & Society and Data for Black Lives. In addition to the resources at those websites, notable resources include:
- “Algorithms of oppression” by Safiya Noble. (Book).
- “Digital democracy, analogue politics” by Nanjala Nyabola. (Book).
- What Algorithmic Injustice Looks Like in Real Life, ProPublica, May 24, 2016. A well written article highlighting these issues in criminal justice.
Meeting these challenges
In building sector capacity in the sector, these are the types of challenges we can and must address.
Nodes on the graph represent articles; try clicking on them. Experimental. May not work on mobile.