Digging for yams: Achieving data science success in the social sector
No mandate is modest, and at Our Community, ours is to link arms with the Australian social sector and march into a data-savvy, effective future. We escalated that goal in 2019 by starting a tutorial series, consulting with not-for-profit organisations, and running meet-up events for the local data science community. This month, we’ll be keeping momentum when Melbourne’s first Datathon for Social Good kicks off. But with a year under our belt, it’s a suitable moment to step back and reflect on the broader challenges ahead.
The history of technology is stuffed with superlatives — grand promises of transformation and improved lives; some brilliantly realised, many forgotten. As the dust settles, the past decade will prove no different. Data science has hyperbole built into the name: a mélange of “data”, that über-commodity, vaunted as the lifeblood of business, with “science”, the modern religion.
Whether data science is a new field leveraging a firehose of data, or a decades-old practice rebranded, the question of interest is how much impact it will have.
We’ve pushed through the hype cycle rapidly. Most people accept that data science has become a catch-all for a set of skills and techniques that few simultaneously possess or employ. As testament, the Harvard Business Review, who infamously spruiked data science as the (sexy) work of unicorns, now champions a team-based operation. Rather than seek CVs that can contort to match a project’s needs, you begin by identifying talents that already exist within an organisation, and upskill or outsource the gaps.
A talent-driven approach suits organisations who don’t have the resources to helicopter in a wholly new team. But it points to the first challenge in our initiative to uplift not-for-profit organisations: even if they possess a well-structured and useful dataset, a lack of higher-level data capability, infrastructure, and governance can often put slack in the sails. In this scenario, conducting value-adding data science is not like plucking figs that dangle from low branches. We must get on our knees and dig for roots.
An organisation early in its data maturity has fewer options, but they can excavate what they have to identify smaller, still-valuable projects to execute. The labour may feel disproportionate for the initial return, but it will pay twice: much as sifting the soil encourages future plant production, getting stuck into a data project now will provide lessons for future work, and the mere willingness to do so is a necessary step in shifting company culture.
The second challenge has to do with an organisation’s reason for being. In a for-profit enterprise, the end-goal is eponymous, and the measure of a project is clear. In a social enterprise, things can be more nebulous. Social outcomes are manifold, often qualitative, and long-term. A robust pipeline and a valley of talent cannot help if the right data is not being collected. But impact measurement is a field unto itself. Doing that kind of work is outside of — and yet inextricable from — our immediate mandate of making the social sector data-savvy. The path out of this maze isn’t straightforward; it’s a paradox that fuels our ongoing meditation.
That’s the territory we negotiate as we pull the social sector towards data science. What of our aim to push data scientists towards purpose-first work?
Building community is no easy task, as anyone who has tried it can attest, but it’s an essential aspect of change-making. Besides the inherent logistical challenges, we’ve grappled with some unique wrinkles. For one thing, the catch-all nature of data science makes for a motley crew. It can be tricky to cater to the disparate skills and motivations of data natives while seeking to draw data neophytes with not-for-profit experience into the same room.
Then there’s the reality of the job market. Data science gigs are offered by firms that can afford and support those roles. Not only does this describe a minority of social sector organisations, such organisations form a minority of the market mix to begin with. Our aim, then, is less about creating a cohort who prioritise for-purpose work — though that’s commendable — and more about cultivating a social consciousness in the field that can be taken into commercial firms. We want to encourage pro bono work and inspire data scientists to take up social causes once they’ve established themselves.
It’ll be beneficial to have more professionals furnish a prosocial lens as we interrogate AI bias and similar ills. Technology can expose and introduce bias, as well as magnify its consequences. The conversations we’re promoting now may act as a surreptitious vaccine or alarm bell in the future. After all, what is often missing from the dialogue is that technology is an increasingly powerful lever to reduce the effects of human bias.
Rather than accept the status quo, data scientists can use this moment of scrutiny over their work as a catalyst to leverage their power for good.
Where to begin? Perhaps with people who have an understanding of the dangers. To that end, our mission is effectively a dual one: while we envisage a purpose-first data science workforce, it must also reflect the diversity of the people it serves. We must treat the engagement, involvement and leadership of women and under-represented groups as essential, not nice-to-have, in whatever we do. It’s the only way we will achieve equity in a world governed by data.
At Our Community, we’ve had a promising start. Our early consultations have been constructive and are already informing the next phase of our initiative. Our tutorials have reached dozens of not-for-profit organisations, and we cherish the kernel of social-good advocates that’s formed around our action-focused events. It can be messy in the dirt, but the rewards are self-evident.