The Human side of Data

UNDP Accelerator Labs
9 min readMay 4, 2021

By Mirko Ebelshaeuser, Innovation Programme Specialist, UNDP Accelerator Labs. Twitter: @MirkoEb

A little while ago Gina Lucarelli put out there what not many people in the development sector dare to say: the Sustainable Development Goals are off track. While confronting us with a rather grim testimony of recent events — we lost 25 years of development efforts in 25 weeks; and the achievement of the Agenda 2030 needs the combined change in behavior of almost 8 billion people on the planet — there was also a glimmer of hope. The re-emergence of distributed action and collective problem solving seen in many levels of societies around the globe lets us hope that we still have a chance. Put in a more technical term Collective Intelligence, or the hive mind, might be an emergent way forward to “hedge against uncertainty. And that’s a good thing because relentless uncertainty may be the only sure thing on the road to 2030.”

Can the hive mind save us? Yes, no, maybe, I don’t know — can you repeat the question?

What is this hive, or collective intelligence? Something that I´ve been tinkering with a lot lately. Collective intelligence in the most simple sense means “becoming smarter together” and in more elaborate terms the “utilization and interplay of people, data and technology” to achieve impact. In doing so, collective intelligence aims at enabling us to better understand the challenges & solutions we work on; empowering collective action and decision-making; gaining new insights to complex systems; and being more inclusive in reaching the people furthest behind by diversifying, increasing, and empowering the people we work with.

Ok, enough development jargon.

Central to understanding the power collective intelligence holds for development is the value people and ordinary citizens add in solving complex challenges. Ethnographic insights, local customs, knowledge about everyday life, or even the contextual understanding of cause and effect have all far too long been neglected in development. Skeptical? Sure, but more often than not official statistics, census data, or static data sheets (often outdated by the rapid changes of current environments) are the basis on which decisions are made and realities affected — which at times are misleading or painting an inconclusive picture on the situation on the ground.

Of aircrafts and algorithms

One of my favorite examples of “data almost gone wrong” is the story of Abraham Wald, a Hungarian mathematician and US emigrant, and the US-Navy during WW2. The Navy, eager to find out where they would need to strengthen aircraft armor to ensure the safe return of their aircrafts, analyzed the patterns of bullet holes across their aircrafts with the subsequent conclusion that the wings and central body would need reinforcement (where most bullet holes were). Except, Abraham Wald disagreed, arguing that the nose, engines, and mid-body area would need reinforcement (the parts with the least bullet holes), because all the aircrafts that were shot in these parts didn’t make it home. The Navy didn’t look at the whole sample, but only at their survivors.

Image Source:

Ok, WW2 is a bit far back in history and a lot has happened on the data-side-of-things since. Today we have Artificial Intelligence, Machine Learning, more advanced data models, complex analytics, and algorithms that assist us in making decisions (often faster than we ever could) but also these are far from perfect (yet). Recently, we saw how easily a Tesla can be tricked into acceleration when a piece of tape is put on a speed sign or how OpenAI´s vision system can be fooled with simple handwritten notes placed over objects. Beyond the state of technology, we also see a sort of renaissance of the human factor when Google asked the crowd to assist in training its picture recognition AI or in countering the spread of COVID-19 misinformation on social media blending automated and human review.

What’s the point you may ask? Twofold: 1) while the presented data in Walds’ case seemed logical, it was actually a contextual reading that proved the right answer; and 2) while modern availability of data and analytical models become more advanced, human check and balance is needed to ground us in reality. Human expertise adds context and reality — two important characteristics we need within development and core to collective intelligence — which data alone can´t do.

Unearthing (waste) realities

Photo by Markus Spiske on Unsplash

When we at the UNDP Accelerator Labs speak about Collective Intelligence, we not only speak about it as one of our core methodologies but more about a philosophy that guides how we try to reimagine development for the 21st century — in a way that centers human context and reality.

One far-reaching and highly contextual challenge we have been working on lately is sustainable waste management. Waste and its interconnected drivers of behavior, governance, technology, environment, and health pose many challenges to sustainable development. The issue of waste is impacting almost all of the 17 SDGs and is a far-reaching problem that impacts people alike. Current pathways for improving solid waste management place the burden mostly on government systems and its public service entities, but the generation of waste is a collective activity. A perfect case for the hive?!

By applying a collective intelligence lens, we not only look at the available data and statistics to understand the magnitude of the problem and derive solutions, but we ground these in the realities of, and with the people affected. We asked:

  • What are the shortcomings of public waste collection practices? How are these addressed?
  • Where do data gaps persist? What do these gaps tell us?
  • Who is contributing or part of the waste value chain? What is their role?

(Don´t) Burn baby burn

Photo by Moritz Bruder on Unsplash

Centering our exploratory efforts around the human side of data helped us to better understand the different realities the waste problem impacts around the globe.

For example, combining satellite imagery with GIS and crowdsourcing modalities enabled our Lab in Ukraine to map open burning activities in rural areas. At first, the Lab set out to understand the frequency and locations at which open burning occurs but soon realized that the picture is incomplete without knowing why burning occurs in the first place. Thus, they partnered up with community and grassroots groups to conduct multiple micro-sensemaking workshops with the aim to give the mapped patterns a more contextual reading.

While working closely with communities, the Lab learned that not only is open burning of waste linked to a deep-rooted cultural dimension of socializing (Yes, to gather around the fire and hang out), but also that many citizens don’t seem to be aware of the environmental and health issues associated with burning.

Our Lab in Lao PDR took a similar approach, albeit with a different twist. With the help of community volunteers and a combination of GIS mapping with satellite data, the Lab mapped open burning sides and collected sensory data on air quality, showing that inadequate infrastructure prohibits collection services from actually reaching communities off main roads. Working with citizens, they also found that organic waste attracts stray animals — and this drives burning instead of composting.

What the Labs discovered is what official data does not tell us: reading static data in context illustrates that burning of waste does not solely occur to get rid of waste but holds a social, societal, and security (stray animals) value — which can now be taken into account when planning a response. These contextual insights gain considerable importance given that in today’s information age there are far more statistics and data points than people competent to analyze them — what we need is human expertise in making sense of information (referring here to the Simpson’s Paradox and correlation reversal if you’re keen for more).


But there is more to the story. Looking at our Labs in Viet Nam and Paraguay, both set out to shed more light on the informal sector that has evolved in the absence of (adequate) public waste collection. The Labs have made it their mission to engage in this often misperceived and inaccessible part of the waste economy.

Actively working with informal waste collectors to better understand the dynamics and challenges faced in this supposedly “dark side of the economy” helped the Labs to map the geographical and economical dimensions of the sector at large. While official data sources on informal sectors were very scarce, the Labs were able to make the invisible visible by conducting micro-narrative interviews, tracking GPS routes, and shadowing the waste collectors.

Both Labs share similar learnings in their collective intelligence journey: the informal waste sector not only provides an income opportunity for a large number of informal waste workers, but it also contributes significantly to the waste management and resource efficiency of the country by collecting, sorting, trading and sometimes even processing of waste materials.

More specifically in Viet Nam, the Lab was also able to profile a typical waste worker in their country allowing for insights into the economic and societal hardship many of these workers are facing, although their efforts are covering a significant geographic area (>80%) and are largely contributing to the waste management system of Da Nang City (7.5–9%).

Also here, consulting the people closest to the ground and thus centering our efforts around human experience and expertise has helped the Labs to gain a more contextual and close-to-reality insight into an often inaccessible part of the economy. The Labs understood informality as a space of opportunity (as opposed to a space of predicament) and its actors as active contributors to solving the waste problem — thus making visible a system often not covered in official statistics and operating without official recognition.

Context and reality

The experiences of the Labs offer insight into what potential collective intelligence can hold when you place your bet on distributed and collective action of a variety of different (and unusual) partners. Centering around the human side of data yields close-to-reality insights into complex and often intertwined systems unearthing dependencies hidden in official information.

Recently, we featured our work in collective intelligence and waste during a co-creation workshop for the Global Festival of Action and had 1500 people registered, close to 500 people on the workshop, and many more in the live stream (recording here). (Wow!). Working with the crowd and opening up our practice to a wider audience demonstrates the growing demand of citizens and the public to be part of solving and adding their expertise to the problems that affect them.

We are determined to continue placing our bet on collective intelligence to help us put the Sustainable Development Goals back on track — and by putting emphasis on the human side of data, hopefully not armor the wrong parts of our plane.

What is your experience with Collective Intelligence? Have you already worked with or are you part of the hive? We’d love to hear from you.

If any of these learnings and reflections resonate with you — give us a shout!.

And if you are keen to learn more about our experience with Collective Intelligence, stay tuned for the launch of our new publication series, “Collective Intelligence for Sustainable Development: Getting Smarter Together,” including the accompanying report, “13 Stories from the Labs”, which details more of these pioneering approaches deployed by the Labs — from using big data to improve waste management in Lao PDR to combining multiple datasets to tackle gender-based violence in Mexico on May 13, 2021.

— — —
- Get in touch and join our conversations online via @UNDPAcclabs
- If you would like to collaborate with us, please email us at
- Find out more about other UNDP Accelerator Lab initiatives via



UNDP Accelerator Labs

Building the world’s largest learning network around development challenges. 91 Labs in 115 countries.