Kirill: 00:00:00
This is episode number 413 with Director and co-Founder at Data-Pop Alliance, Emmanuel Letouzé.
Kirill: 00:00:12
Welcome to the SuperDataScience Podcast. My name is Kirill Eremenko, Data Science Coach and Lifestyle Entrepreneur. And each week we bring you inspiring people and ideas to help you build your successful career in data science. Thanks for being here today. And now let’s make the complex simple.
Kirill: 00:00:44
Welcome back to the SuperDataScience Podcast, everybody. Super excited to have you back here on the show. Today, we got a special guest calling in from New York, Emmanuel Letouzé, who is a man of an extreme number of variety of different backgrounds and skills. So Emmanuel is the Director and co-Founder of the Data-Pop Alliance, an organization, an NGO, so a not-for-profit organization where they create or use data and AI to help solve global challenges. And on top of that, he’s also a political cartoonist, something he’s been doing for the past 25 years.
Kirill: 00:01:24
So a huge combination of backgrounds, a very interesting worldview. And today, we’re going to discover some very cool topics in the space of geo-economics. So what exactly did we talk about? Well, today, you’ll hear about behaving ethically in all aspects of your life and why that is important in separating professional and personal, some philosophical ideas there. Data and economic politics in the world, changing the world with data, better data versus better decisions, and what the difference is between using data as a lens for measuring, and as a lever for impacting change.
Kirill: 00:02:02
You’ll find out about the 17 SDGs, or Sustainable Development Goals and where they came about from what they mean in terms of geo-economic development of countries. I found that part extremely fascinating. Then we also spoke about big data as a mindset, climate change, gender-based violence, COVID. We finished off with data and disaster resilience. So some very important topics, global topics, and coming from a person who has worked in the space for quite some time and has a lot of interesting perspectives. So I think let’s get started. And without further ado, I bring to you Emmanuel Letouzé, co-Founder and Director at Data-Pop Alliance.
Kirill: 00:02:54
Welcome back to the SuperDataScience Podcast, everybody. Super excited to have you back here on the show. Today, we’ve got a special guest calling in from New York. Emmanuel Letouzé, Letouzé, right?
Emmanuel: 00:03:06
Yes.
Kirill: 00:03:06
Welcome, Emmanuel.
Emmanuel: 00:03:06
Thank you very much. Thank you for having me.
Kirill: 00:03:10
Excited to have you, lots to talk about. It was very cool what we were talking about just now about personal and professional. And you’re now a father of three girls, right? So let’s maybe continue that discussion. How has that changed your life? How has that changed your perspective?
Emmanuel: 00:03:35
I mean, as all parents, it changes your day to day routine. Obviously, you’re quite tired. But then there are, of course, deeper changes. It changes your priorities, what matters. Also, your perspectives I think on life, you tend to be less self-centered, less egoistic, you think maybe like longer term, because you’re thinking about the kind of world that they’re going to live in. I would say especially having had initially two twin daughters 10 years ago, and now an eight month old, so girls, I think it’s also made me more aware of gender inequalities and the kind of challenges that they might face, and wanting to work more on gender.
Emmanuel: 00:04:38
For instance, we have projects with Data-Pop about gender-based violence, which have been promoted quite a bit. And so I mean, there’s no counterfactual. So I don’t know if I would have done that to the extent if I had not had daughters, but I mean, the fact that is I think it had an influence.
Kirill: 00:04:58
It’s very interesting how that ties in with your, I think, general kind of sentiment in life that you’re a very aware person. For example, I noticed that in one of your presentations, I think you flew from Europe to the U.S. And one of the first slides was like, well, I spent two tons of carbon emissions getting here, let’s see what we can actually do of value. That’s a really cool way of seeing things.
Emmanuel: 00:05:30
Yeah. I think there’s a thread in my life, which is that I don’t see the very strong, or very much of a distinction between the personal and the professional. And so when people say we have to act professionally, I don’t really know what that means. And so I think I prefer when people say you have to behave ethically, in all aspects of your life. Yeah, so when I’m traveling to a conference, or was before the pandemic, yeah, you have to take into account all the consequences of your actions. To be frank, I had many instances where I thought, is it really worth it? I’m burning two tons of CO2, and so is it worth it?
Kirill: 00:06:40
Yeah. It’s interesting how sometimes we do take actions we think are for good, but actually tend to create more harm. And the other way around, it reminds me of a story. When in Iceland, there was a volcanic eruption. I think it was like maybe 2010. I’m not sure exactly about the year. But because of that volcanic eruptions, all planes were grounded for several weeks in Europe. And in essence, that meant that the amount of carbon the planes would have emitted would have been more than the volcano. So it was a carbon neutral volcano. I saw some journalist presenting on that.
Emmanuel: 00:07:22
Yeah, I actually remember that, the cloud that was over a big part of Europe, as a result of that. Yes. I mean, especially now with COVID and what it has revealed, I think about the state of the world, and the structural fault lines, and the structural systemic inequalities, like racism, sexism, but this also like pandemic of poverty and inequality. I think it’s really a time where we have to reassess our ways of life, and I think it’s a good opportunity to do that and just pause for a little bit, think about what matters, the kind of world and societies that we want also, like our children to live in. So not flying around the globe is one of those small changes, but I think there are also deeper changes that we have to make.
Kirill: 00:08:25
Yeah, yeah. Absolutely. Is that why you started Data-Pop Alliance?
Emmanuel: 00:08:33
So when you say why, you mean sort of like to-
Kirill: 00:08:38
What was the motivation? Because it’s an NGO, right?
Emmanuel: 00:08:39
Yes.
Kirill: 00:08:42
It’s a not-for-profit organization, nongovernmental, why … And it looks like a big undertaking, and I would love for you to tell us more about it just now. But to start off with like what motivated you to start this organization.
Emmanuel: 00:08:59
Yeah. I mean, of course, I could talk for hours for the nuances and all the different threads, but we don’t have that time, and probably few people have that interest. So, I mean in a nutshell, I think there were like short-term determinants or drivers and there were like deeper determinants and drivers. And so the short-term ones were that, so in 2011, like 2010, 2011 so these were the very early years of what’s become to be called a data revolution, of course, which we are now in and your podcast is definitely one of the both like your signals and example of what’s happened over the past decade.
Emmanuel: 00:09:54
And so I was working … I had worked for the UN before as an economist, which is more like my background, and then I was working at UN Global Pulse, where I wrote a white paper on the so-called big data for development, and so I had already entered that space, so to speak, and started thinking about how big data, data science could basically help improve, change the world. And then we might get to the deeper root causes or drivers of why I wanted to do that.
Emmanuel: 00:10:30
But at any rate, yeah, I felt a bit frustrated at times by the UN bureaucracy, the fact that it’s very hard to talk about politics, and political economy at the UN, because you have to deal with member states. And so you have to walk a fine line. And so I wanted to have … It was clear to me that I thought data was going to be this defining factor or feature of at least the next couple of decades, that it was going to be very powerful, that it was going to be very disruptive. I wanted to do something about and with it.
Emmanuel: 00:11:11
And then from there, I wonder, okay, what can I do? I’m pretty independent, in terms of personality. I don’t like authority very much. I want to say I have an authority issue sometimes. And so I thought, okay, I have to create something of and on my own. I didn’t want it to be for profit. I didn’t want to create a startup. And so yeah, I decided to create an academic minded NGO, so with colleagues, partners, people in my network at MIT and Harvard and the Overseas Development Institute, and I wanted to do different things. I wanted to do trainings, I wanted to do policy advocacy, policy advice, research, of course.
Emmanuel: 00:12:04
And so that was in 2013. Yeah, that was how it began with just those ingredients. And just to add on that, the very beginning of Data-Pop Alliance, it actually started or I had the idea after a discussion with Kenneth Cukier, the data editor of The Economist, at a conference in Los Angeles. After the conference, we talked, we went to a bar, and we had a pretty long discussion. And he asked me, like, what do you want to do in life? And I told him a little bit about this idea and he just said, “Just go for it.” For me, this was really like the sparkle.
Kirill: 00:12:53
Awesome. That’s very cool. And a bold move as well, like moving from your field of economics to data, which you clearly saw potential in, but quite different. And you managed to get backing by quite a few large universities like Harvard and MIT. Tell us that story. How did you get them on board?
Emmanuel: 00:13:15
So yeah, I mean, there’s always been different threads, I would say, in my life. Yeah, I tend to think and act a lot along the lines of threads and thinking in terms of also like systems terms and network terms. So first, it wasn’t that much of a change, because my background as you mentioned, is in development economics and demography, in particular. I have a master’s in economics, another master’s in development economics and a PhD in demography. So those are like quite quantitative fields to start with.
Emmanuel: 00:13:55
So I was always interested in statistics and data and numbers. And actually, my first job, which we might get to was working on official statistics for the French government in Hanoi, Vietnam. So with the National Statistics Office of Vietnam, so there was always a bit of a thread. And we started working, for instance, on data mining in Vietnam in 2002, which I had never heard of.
Kirill: 00:14:23
Wow.
Emmanuel: 00:14:25
And so over the years, I built a network of people. I mean, I just met people while being at the UN, while then working at Global Pulse, in this like nascent field of computational social science and data science. And so for example, I met or in particular, I met with Professor Alex Pentland at MIT, who’s one of the, let’s say, godfathers of the space of the field of computational social science. Then I also met with people like Patrick Vink at the Harvard Humanitarian Initiative. I had also been working with people at ODI in the UK on developing mechanics.
Emmanuel: 00:15:15
So when I decided to create Data-Pop Alliance, I thought, okay, I need partners, I need people who can also help me, vouch for me. And so I mean, just one quick last example, actually, one day, I wrote an email to send Pentland at MIT, and I said, this is what I would like to do. I had met him maybe two or three times. But he’s super busy and gets lots of requests. So I just told him pretty cold email, “Can I come up or actually down, up from New York, to Boston to Cambridge to talk about this idea?”
Emmanuel: 00:15:54
And he said, “Sure.” And so we met for five, 10 minutes on the sixth floor of the MIT Media Lab, and I explained what I wanted to do. And I said, “Do you want to be involved, help?” And he said, “Sure.” I said, “What does it mean?” “Well, just tell me what you need, et cetera.” And so then he became the academy director. And so I don’t know, maybe it was a bit of a bet for him, because he thought I have nothing to lose. And this guy seems pretty reasonable. And so yeah, that’s how it started.
Kirill: 00:16:31
Yeah. That’s awesome. Taking some bit of risks, and also reaching out to contacts. That’s good advice.
Kirill: 00:16:43
Are you subscribed to the Data Science Insider? Personally, I love the Data Science Insider. It is something that we created, so I’m biased, but I do get a lot of value out of it. Data Science Insider, if you don’t know, is a free, absolutely free newsletter, which we send out into your inbox every Friday, very easy to subscribe to, go to www.superdatascience.com/dsi.
Kirill: 00:17:06
And what do we put together there? Well, our team goes through the most important updates over the past week or maybe several weeks, and finds the news related to data science and artificial intelligence. You can get swamped with all the news, even if you filter it down to just AI and data science. And that’s why our team does this work for you. Our team goes through all this news and finds the top five, simply five articles that you will find interesting for your personal and professional growth.
Kirill: 00:17:36
They are then summarized, put into one email, and at a click of a button, you can access them, look through the summaries. You don’t even have to go and read the whole article, you can just read the summary and be up to speed with what’s going on in the world. And if you’re interested in what exactly is happening in detail, then you can click the link and read the original article itself.
Kirill: 00:17:55
I do that almost every week myself, I go through the articles. And sometimes I find something interesting, I dig into it. So if you’d like to get the updates of the week in your inbox, subscribe to the Data Science Insider, absolutely free at www.superdatascience.com/dsi, that’s www.superdatascience.com/dsi. And now let’s get back to this amazing episode.
Kirill: 00:18:16
Can you give us some examples of projects or initiatives in the Data-Pop Alliance, so we get a better feel for what exactly it is that you do?
Emmanuel: 00:18:29
Sure. So the overall tagline is changing the world with data. And so I think that gives a pretty good sense of the vision and the mission. So it’s pretty ambitious. It’s a quite political undertaking and endeavor. Some might say yeah, it’s a bit vague changing the world with data, but I still like it and I think it conveys to be inspirational. And so concretely, so we are structured around like three main work pillars, which we call diagnose, mobilize, and transform.
Emmanuel: 00:19:12
So diagnose is research, so empirical research, for example, now we have, as I mentioned previously, a pretty strong focus on which of course is not exclusive, but there is focus on gender-based violence in Latin America, in particular. So we’re starting, we’ve actually published some initial papers on the correlates of gender-based violence in Mexico City. And now we’re working to extend that in Bogota and San Paolo. And so this is funded by the German Cooperation Agency, so GIZ. To do that we use mobility data. We use official statistics. We use call data to shelters that we’ve had access to.
Emmanuel: 00:20:05
And so we try to combine different types of data sources. So called big data, mobility data. But also official statistics and other kinds of let’s say, like administrative records. So we’ve done also research on poverty. We’ve done research and research papers on inequality. All of those projects tend to be very focused on the Global South, and actually anchored in the Global South. So we have teams in about 10 countries now or staff in about 10 countries. So it’s very grounded.
Emmanuel: 00:20:48
We also do strategic evaluations on digital transformation, for example, and we’ve helped the European Commission do their evaluation of digitalization programs in Sub Saharan Africa. And currently, we’re working with WFP on digitalization. They just got the Nobel Prize for Peace. And so we’re pretty not proud, because we don’t have anything to do with it, but just absolutely excited to be working with them.
Emmanuel: 00:21:19
So the second thing we’ve done, or we do is mobilize. And so under mobilize, I think the best way to describe it is to think about, we want to build capacities, and communities. And so we want to build data capacities and communities. And I think a key concept here is that of data literacy, or literacy in the age of data. And so we want people to be more aware of the main tenets of the data revolution, what is done with their or our data, how it can be used, what kind of skills they need, how they can be advocates, and so we’ve done trainings in about 12 countries with the UN on the Sustainable Development Goals. So that’s one of the things we do.
Emmanuel: 00:22:13
The last thing we do is under the transform pillar, so we advise some partners, clients, bilateral agencies, on their data strategy on data sharing, like agreements, protocols, systems. We have also supported the Colombian government in designing Colombia’s first data strategy, national data strategy. In general, yeah, we try to develop also solutions. Another project that I might talk about is the open algorithms or OPAL project. Yeah, that’s pretty much what we’ve done.
Emmanuel: 00:23:01
And just last word in terms of I think the key features or the defining features, elements of Data-Pop, are that it’s pretty academic, academically grounded, but also quite, so it’s very locally grounded as well. It’s mostly in the Global South, and with Global South partners. And there is also a bit of advocacy and a bit of activism in our DNA.
Kirill: 00:23:32
Okay, wow. What do you mean by Global South?
Emmanuel: 00:23:35
Yeah, so maybe it’s a bit of a contentious term. Typically, what people mean, is countries that are non-OECD countries.
Kirill: 00:23:47
Okay.
Emmanuel: 00:23:50
But actually, there are exceptions. So for example, Mexico, and Colombia are both OECD countries. So it’s sort of like the rich countries club of sorts, there are about less than, fewer than 30 countries. And so we mean that you have 170 plus who are Global South. Yeah, so there isn’t a very strict like definition. And some people have used like, third world countries, but I think it’s a terrible concept of phrase that a lot of people in my field like development try to stay away from. Yeah, so it’s Latin American, the Caribbean, it’s Sub Saharan Africa. It’s the Middle East and North Africa region. And it’s like most parts of Asia.
Kirill: 00:24:51
Gotcha. Non-OECD. So Organization for Economic Co-operation and Development, right? That’s the OECD countries. Gotcha. All right. Very, very interesting. Yeah, it looks like … Is it the case that you chose to help these countries because the data can have a larger impact? Or is it because you saw the needs there that people need better, I don’t know, like better solutions to the global problems that they’re facing in those regions? What was the main driver behind that choice?
Emmanuel: 00:25:40
I mean, yeah, so there were personal drivers. And they’re also more rational, like practical drivers. I mean, I would start with the more practical, concrete drivers. So yes, actually, I do think that there are larger capacity gaps, and overall, like a larger or very large potential in these regions for using more better data to change, to design better public policies, but also more fundamentally, to change, to improve the power dynamics, power structures.
Emmanuel: 00:26:33
These tend to be countries where you have a lot of power grab by political and economic elites. In part, that comes from colonization in some parts of the world, that come from, like natural resources, like oil, or diamond or gold. So these also tend to be highly unequal countries, if you look at the Gini coefficients, for example. And so yeah, just thought that data could have a big impact, thereby improving, let’s say, human systems. So that’s one reason.
Emmanuel: 00:27:12
Another reason is that it’s also where you have the bulk of the world’s population. And so if you want to have an impact, I mean we’re talking probably 7 billion out of 8 billion, and also where you have the highest rates of demographic growth. Africa’s population, for example, so they are still in their demographic transition. So it’s growing very fast. So they were also yeah, that consideration. And then they face specific challenges that I’m personally very interested in. So poverty, high levels of inequality, the impact of environmental degradation, conflict and crime in different regions, migration.
Emmanuel: 00:28:03
And then there are more so like increasingly geopolitical, or political questions about data sovereignty, the fact that, for example, Sub Saharan Africa is becoming a bit of a hotspot for competition between U.S., China, Europe, but also African actors, when it comes to connectivity, data centers, et cetera. So I think these are also like very, very interesting regions. And of course, last but not least, things around gender equality, gender-based violence. So there’s a lot that interests me.
Emmanuel: 00:28:45
I would just say and then we can talk more about it if you want, but on a personal level, I spend a lot of time growing up traveling with my parents first and then for work. Yeah, I did part of my studies in Senegal, for example. So I always had a sort of the bug or the drive to actually work in these regions.
Kirill: 00:29:14
Yeah, no, I completely agree. It’s important, like if you have the opportunity to expose yourself to different cultures and see how the world is different and what other problems they’re facing there. I think this would be a good segue to your UN paper, Big Data for Development. I believe you wrote two papers for the United Nations. I only read the abstract of one, I think the 2019 paper. You talk about some very interesting concepts about how data is being used currently to address some of the world’s problems, but is it really being used to address our world’s problems or is it used just to measure them?
Kirill: 00:30:00
We haven’t spoken about this, but you do cartoons for explaining concepts, and what some of those cartoons you use to explain data. And in this particular report, I think on page like, on the one of the first pages, there’s a cartoon illustrating the whole situation. Let me just read out what it says. This is page three, and somebody is speaking at like a big assembly saying, “The data revolution is here. The good news is we can now measure your poverty levels at amazing levels of geographic granularity in real time. The bad news is we still can’t do anything about it.”
Kirill: 00:30:39
I found that very interesting. I found it also close to my heart, because for a lot of our courses, in what we teach, we use the World Bank data. There’s a lot of geo-demographic indicators, poverty, mortality rate, I don’t know like electricity consumption, GDP, GDP per capita. All these things, I’m used to working with them. And the way you put this into your UN paper was interesting that, yes, we can get better and better at measuring these things. But will that actually improve the way of life of people and make changes?
Kirill: 00:31:21
You have a really cool quote here. You said, “Politics often trump statistics, in shaping policies, as you can have all the data in the world, but at the end of the day, it might come down to politics.” So could you tell us in a nutshell, what are your thoughts on, are we using data and AI to just measure things? And does that imply automatically that things are going to change? Or do we need to use them somehow differently in order to bring about this change? And if so, what are the ways that we can use data and AI more to help the world and help improve these geo-economic indicators?
Emmanuel: 00:32:06
Yeah. So yeah, likewise, basically this is, I was going to say my life, but it’s not my life. It’s my work. But as we’ve hinted or mentioned, I don’t see a very strong distinction between both. I think actually, often, we tend to be like walls, when there should be just the fences that you can easily jump back and from to between the personal and the professional. So, yes, and so that’s the big question for me. And it is this question of whether and how, so under which conditions measuring X actually helps affect that X? It’s a fascinating question.
Emmanuel: 00:32:59
I think what this cartoon was saying, so it is a cartoon. So the nice thing about cartoons is that you can be a bit simplistic. You can just look at one angle, one part of the problem. And here, I’m basically saying, well, we in the data science for good community, tend to be a bit complacent. We tend to have this simple vision of the world where we say, all that we or policymakers, and governments, and economic elites, and political elites need are better data. If only we or they well intended, kind of like Bismarckian leaders had better data, then surely we would make better decisions.
Emmanuel: 00:33:45
Now, what this cartoon is saying that, well, not really. If you look at the list of countries around the world, just as an example, and you ask yourself, why are there so much gender inequality? Why is there so much corruption? You cannot make a very convincing case that it’s because these governments and political leaders don’t have the data. But it doesn’t mean it’s just mostly more often because they don’t have an interest, because they are well, they’re fine. They’re doing well, they don’t want things to change.
Emmanuel: 00:34:28
And so I was a bit frustrated indeed by this narrative of the data revolution. And now the question is, okay, so what do we do? How can data matter more? How can measurement matter? How can you have those feedback loops? Which you actually have in an AI system, like an AI system is able to learn from feedback, so from measurements, and it’s able to do that very well. Like, driverless cars were very bad 10 years ago, most people thought that it would never work. Google Translate was very bad 10 years ago, and now it’s gotten better, because actually these systems are able to learn from data.
Emmanuel: 00:35:16
So you do have the feedback happening. And the big question for me is how, as societies, can we actually learn and adapt, and change our behaviors and policies, but also individual behaviors and collective behaviors, on the basis of data? And so you get to things like, it’s quite close to evidence-based policymaking, but I think it’s broader and deeper than that.
Emmanuel: 00:35:49
The key for me is to change incentives, to educate people, but all of us having a data culture, and being interested in the impacts of our actions as the data tells us that they impact the world. And so there’s also a lot of psychological work that has to be done. So that we look at the data, we think, “Okay, what is driving this? What is the data telling us? How can I adjust accordingly” And ultimately, we can talk more about it. I think it’s very, very political. It’s about power structure. It’s about power dynamics. It’s about who has access, and who is controlling this new resource of the century, which is data.
Kirill: 00:36:51
Absolutely. And you also in the same paper, you talk about how with the power of big data, we can replace the way we measure a lot of these geo-economic indicators. For instance, you have these three tiers. You talk about first one, for instance, first year would be something like, I imagine, like I don’t know, population, or poverty and things like that that are very well defined and are usually already measured. But they’re measured through surveys and other things like that. Whereas they can now be measured through certain behaviors online or certain sensors in the real world and things like that.
Kirill: 00:37:44
There’s also two more tiers. Tier two is indicators that are conceptually clear, and having some national established methodology and standards available, but data is not regularly produced by countries. And tier three is now internationally established methodology or standards are yet available for this indicator. So like AI will actually and data, big data will allow us to collect information, for instance, I don’t think there’s economic indicator. I think you put this in the paper as well that an economic indicator that talks about interconnectedness of society, in a certain region.
Kirill: 00:38:21
But at the same time, with the power of data, where we would be able to measure that how interconnected a society is in Region A versus in Region B? And then how is that impacting other indicators? And how can we improve that? So it’s very interesting how data can help modify how we measure these things, probably increase the frequency, accuracy of measuring these things, and also increase the scope of the different indicators that tell us about how healthy a certain population is.
Emmanuel: 00:38:57
Yeah. Yes, and yeah, that’s also quite fascinating. So what you’re referring to are the so-called Sustainable Development Goals indicators. So those are those 17 goals that so 193 governments, or we tend to call them countries, but they’re governments, so agreed to in 2015. And this is in the framework of what’s called the 2030 sustainable development agenda. And so in the 17 goals, you have things like gender equality, poverty, urban development, et cetera, like peace, security. So a whole range of global goals.
Emmanuel: 00:39:40
There are seven indicators. Those seven indicators are indeed grouped into those three categories of the ones that we know how to measure well, the ones that we sort of know how to, they’re called like SDG tier one, SDG tier two, SDG tier three. And indeed, as you mentioned, I think there is the … I think big data, AI approaches are relevant to all three kinds of indicators. So for example, we could measure poverty, which is an SDG one indicator, we know how to measure poverty. We’ve done it for many decades. But it could improve their frequency, et cetera. And so it’s still relevant for SDG one.
Emmanuel: 00:40:21
For SDG three, indeed, I think this is where you have the most experimentation, innovation, and I think scope for impact of big data and AI. So things like social cohesion, and social capital, and different forms of social capital, which are referred to as like bonding, and bridging social capital, trust between governments, and between people and governments. So those are things that are typically not very well measured, and they’re hard to measure, because they’re very psychological, probably behavioral. And in these data, so you can think of both records like how people call each other, which groups call each other? How people talk about each other on Facebook, for example.
Emmanuel: 00:41:16
So political polarization, so if people say on Facebook, “Oh, I had to give a bribe to a police officer.” Those are some SDG 16, in particular, and so there’s clearly here a big role for those kinds of innovation. I will just say that this is really hard work we’re trying to do. We’ve done some pilots in countries like Botswana, we’re doing sentiment analysis currently in places like Equatorial Guinea, like Togo, like Lebanon, Jordan. But we’re still in the infancy, I think of those approaches. But it’s a gradual progress, like process, and I think we’ll get there at some point.
Emmanuel: 00:42:03
So it’s going to help measure. But that only gets to one piece of the puzzle, which is using data to better measure human processes. So it gets us back to the initial discussion. Okay, if you have all these great measures, how is this going to change anything? And here, and then I’ll stop there or here for now. Here, I think we get to the difference between the different two main roles of data, which is one data as a lens, or lenses on the world, so you can see the world through data in very imperfect ways. So you can think of the Plato like cave analogy that data is a reflection of the world and sometimes it’s a poor reflection, but it’s sort of like, almost the best that we get. We see the world through data, data as a lens or as lenses.
Emmanuel: 00:43:04
And then there is data as a lever, as a lever for change, like how can we latch on, use data and these indicators to actually have an impact, to actually change policies, to change regulations, to change our behaviors? Part of the answer is in better measurement, but not only.
Kirill: 00:43:31
Okay, so what can we do to this lever part? I understand what you’re saying. I just wanted to, like get to the meaty part, how do we use data as a lever? I only know about, okay, use data as a lens, even like in a business sense, or in a government sense, okay, get some information, get some insights, and then the business decision makers make the decisions, but how can they actually use data as a lever? Do you have any like, I don’t know, examples of that?
Emmanuel: 00:44:05
So let’s say, yeah, conceptually, I think there are different ways, but also concretely, there are very simple ways in which data is and has been used as a lever. I mean, right now with COVID, for example, you can see that some decisions in … I would just take like friends as an example. I’m not saying it’s the best example. But just as an example, well, there were data that showed that it was the first wave. They were models that showed that so many people were going to die in the absence of some measures. There were mobility models, et cetera. They were so used, using cell phone data, like including cell phone data from [inaudible 00:44:47], for instance.
Emmanuel: 00:44:47
The government said, okay, if we don’t do a lockdown, a very strict lockdown, then it’s going to be a public health and economic and social disaster. We’re going to lock down the country for about two months. So that’s a very clear example where data is used as a lever. And there are many, many other such examples. So that’s really like the entail, I would say, of the measurement, like theory of change or causal change measure some things. And you say, on the basis of that, I’m going to make that decision.
Emmanuel: 00:45:23
What is interesting here is, what are the requirements for this to happen? So you need people who trust data, you need people who trust science, you need people who trust the government. In France and other countries, a lot of people just agreed to be locked down and it wasn’t by coercion, for the most part. It was just because there was this shared belief that indeed, it was the best thing to do. And so people had some good level of trust. And so you need those ingredients actually for data to be used as a lever.
Emmanuel: 00:46:00
I think, also a very interesting part about data and is that it’s also sort of like a language. And it’s also a culture. For instance, Andreas Weigend, who used to work at Amazon, and who’s a pretty famous data scientist, referred to big data as a mindset, the mindset to turn mess into meaning. I think that when we talk about data culture, or data literacy, it’s clearly like one of those ingredients. But I think it also points to the fact that you can actually bring people together to talk about certain issues using data almost as an excuse.
Emmanuel: 00:46:52
So you can do capacity building workshops, you can do community engagement workshops, where you bring the private sector, you bring the official statistics office, you bring the NGOs, you bring startups, you bring civil society organizations, and people actually talk. And you have this exchange of idea, you create trust, you create a shared understanding. I think it matters. I think, then it actually helps create, or trigger partnerships, collaborations. This is where I think data is also important. But it’s not the measurement channel.
Emmanuel: 00:47:36
It changes behaviors, dynamics. It creates trust that they use the trust, it creates trust between these ecosystems, which tend to be quite fragmented. I’ll just give like one example. Like one quick example, also, like from France. So there was a commission about climate change that was put together where some people like citizens were almost like, randomly selected to come up with 100 plus, 150 plus measures about climate change. And what was very interesting is that, at the beginning, people were very polarized as are our societies.
Emmanuel: 00:48:24
When they started talking on the basis of facts, when they started exchanging ideas, perspectives, then you can get to like this is where a consensus like it comes to emerge. I think I really see a big role for data and facts to actually create those kinds of consensus, of compromises in very polarized societies.
Kirill: 00:48:57
Wow, very, very interesting. I wanted to ask you, have you seen The Social Dilemma on Netflix?
Emmanuel: 00:49:03
No, I mean, I’ve seen it in the sense that I’ve seen it in my Netflix thread or recommendation, but no, I haven’t seen it.
Kirill: 00:49:13
I haven’t either, but it’s on my list, and I’ve heard a lot of things. In light of that movie coming out and raising this awareness of this issue, I also wanted to ask you, this actually was … I saw in one of your, I think, course that you were presenting. How can personal data be tapped into in ethical and safe ways? We talk about using data to model these economic indicators to come up with consensus or come up with consensus on policies on how to use data and things like that.
Kirill: 00:49:53
But a lot of it will require, or will be enhanced by the analysis or by the access to certain elements of personal data. How do we walk the line of using that to empower decision making and using data as a lens and as a lever, but at the same time not overstepping the boundary and allowing unethical things to happen because of this? What are your thoughts on that?
Emmanuel: 00:50:23
I mean, first of all, you said lever and I will see lever. So maybe we talked about learning, so maybe I just learned that.
Kirill: 00:50:30
I’m not sure myself. I think it depends on the country.
Emmanuel: 00:50:34
Okay. [crosstalk 00:50:35]. But it is pronounced lever. But yeah, at any rate, so yeah, so that’s also a very big question. And there are [inaudible 00:50:48] and some directions. I just want to take like 20 seconds to just provide one more argument actually for this lever argument.
Kirill: 00:51:05
Yeah, sure.
Emmanuel: 00:51:06
Yeah, I think it’s very important. I’m still trying to think about what it means, what it implies, what it requires to actually use it as a lever for change. I think one way is really along the lines of this commission on climate change, if you as we want to do in this project about gender-based violence in Mexico, Colombia and Brazil, if you organize focus groups, or discussion groups with women, but also men about this project, so you try to take their perspective, et cetera, I think that’s going to have an effect above and beyond the effect of the model or wherever the predictive model or the insights that we’re going to glean and yield from the data.
Emmanuel: 00:51:53
So I think it’s maybe like rehashing what I said about bringing different perspectives, like concretely, I think, for some of those women, it might be the first time that they actually get to talk about these experiences. And so, as a consequence, perhaps later, they will be more inclined to report cases of domestic violence. So there’s also, I think, an empowerment. There’s also like a psychological argument that I think is quite important when we think about how data can really matter, can really make a difference.
Emmanuel: 00:52:30
And ultimately, and I will stop there and then go back to your question. Ultimately, it’s really what is fascinating and promising is that it’s really down to us as individuals. It is our choice to stand up in the subway, to give up our seat or not. It is up to us to do all those things that collectively make up the world the way it is. It is up to us to decide if we buy everything on Amazon, or just what we really need. And if we don’t need, if we can go next door to buy something, then I think it’s better if we do that.
Emmanuel: 00:53:07
Now, to go back to your question about tapping into what’s called yet PII, personally identifiable information, or personal data, private data, so there’s different ways it’s called and actually a range. So in short, the way it’s done is typically by not sharing, not opening the data itself. So let’s call it like the raw data. Some say that is an oxymoron like raw data, but not opening the raw data, because it’s just too sensitive and there are commercial considerations, ethical, privacy considerations, but instead allowing some computation, some analysis to be performed on the data through a question and answer or through question and answer mechanisms.
Emmanuel: 00:54:07
So what we would be interested in, for example, is looking at mobility data, for instance, from cell phone activity. So those little sensors and trackers that we all carry on us. You would want to know things like what is the current population density or distribution in a given area? Is it changing by more than 10% from some baseline? Do you see massive movement of population? Do you see …
Emmanuel: 00:54:41
For example, in the case of COVID, recently, there was a study that showed that rich areas in the U.S., in urban regions were able to reduce their morbidity by more than 50% three days before poor areas. And so here’s data as a lens, it shows something about human ecosystems. We see something happening, and then it can also be data as a lever, because then you can say, “Well, why is that?” It is because poor areas, or poor people usually don’t have a choice. They have to go to work, a lot of them are essential workers, et cetera.
Emmanuel: 00:55:27
And so then you can take actions on the basis of that. But to do that, you don’t need to have access to the raw data. So you can just set some feature, you can just partner with the telecom companies and say, can you tell us if, when, where mobility as measured by some algorithms, according to some metric, where morbidity has declined by let’s say 50%. We can talk about the technology. Yeah, so these data are pseudonymized, so meaning like your phone number is changed, I mean, is replaced by a string of numbers and letters. So it’s not anonymization, it’s no longer called anonymization. It’s called pseudonymization, because it’s replaced by a pseudonym.
Emmanuel: 00:56:25
And then it’s also aggregated. And so the big question is, how can it be re-identified or not, and at what level? And so there’s a whole strand of research on that, but that’s the gist of it, this question and answer mechanism based on pseudonymized, aggregated data.
Kirill: 00:56:50
Gotcha. Pseudonymized. That’s very insightful. Thank you. Well, this has been so fast, we’re slowly getting to the end, actually. And I still have so many questions, but I’ll ask this one that I think will be interesting. What are your thoughts on how data can help with disaster resilience? We’re entering a world where there’s more and more changes happening fast, including, like fires in Australia, this COVID situation, and there’s going to be other disasters that strike. What are your thoughts on data and disaster resilience?
Emmanuel: 00:57:37
Yes, another fascinating topic. Actually, we worked a few years ago, as Data-Pop Alliance on a series of case studies and a report using big data for climate resilience with DfID, so the UK’s development agency. And it gets to things about complex systems. And it’s super interesting. I think I see like two main ways. One is more short-term, and in a sense, you could call this like more humanitarian response or emergency response, where you can try and predict the spread of a fire or the spread of an epidemic, or get sort of like digital smoke signals for conflict, or for crime, for hunger, and then you react as fast as possible.
Emmanuel: 00:58:43
It’s been used in the case of earthquakes, in the case of tsunamis. There, you have a lot of technical and scientific challenges about representativity, about sample bias, that you really need to be aware of. Just a super quick example, if you have an earthquake, and if you rely on cell phone activity to know where people are to send trucks, then it’s actually likely that if you send it to where you have more signals, like from SIM cards, then you’re going to send it to the wrong regions.
Kirill: 00:59:22
Why?
Emmanuel: 00:59:24
Because it’s actually where people were the least affected. The regions that were the most affected, like towers are down. They are under rubbles. They are unconscious. This is where statistics, basic statistics and knowledge, expertise come in, people should be able to tell the respondents who would look at a map of hotspots. No, like you’re getting a wrong signal, we’re going to make a very wrong decision because you don’t understand basic statistics. So that’s one way.
Emmanuel: 01:00:01
Another way I think is, let’s say yeah more in the development or societal development sphere, which is more upstream, and maybe like more long-term and more challenging, which is how can data help us change, adjust our behaviors, or collective decisions in ways that make those events less frequent, less violent, and to different degrees? So it can be about using less electricity. It can be about using our cars less. It can be about being less violent, less greedy.
Emmanuel: 01:00:49
I think a lot of that, a lot of what we see around the world comes down to this very simple human feeling or desire of greed. People tend or seem to be wanting to have more money, more power, more social and other forms of capital. One of my hopes is that by learning from data, by being more aware of the consequences of our actions, then we could change our behaviors and societies in ways that would make them more resilient.
Kirill: 01:01:30
Fantastic, I love it. Ties in data and consciousness, how data can help us on the path to becoming more conscious as a species. Some people argue that that shift in consciousness is absolutely necessary in order for our planet and our species to survive.
Emmanuel: 01:01:49
Yep, just one quick thing, very quick thing here.
Kirill: 01:01:53
Sure, sure.
Emmanuel: 01:01:55
I think, like really I said at the beginning, but I think it’s really maybe a once in a lifetime opportunity for us, including as data scientists, to really make a contribution, to really like change, like try and change or improve the world, and not in superficial ways. Sometimes there are trade-offs. It’s very political. I think sometimes I’m quite frustrated about the data for good movement, because it doesn’t really get to the deeper power dynamics and structures that underpin and preside over some of the outcomes.
Emmanuel: 01:02:40
And so, one example of that would be if you … There are projects right now that are partnering with very big banks, and I’m not going to name them because it’s not about the naming and shaming, but very big banks. They have a pretty, I think, shady record, when it comes to doing illegal things, ripping off clients with the very high fees, et cetera, whose CEOs make 30 plus million dollars a year.
Emmanuel: 01:03:13
And so they partner with those big banks. And they say, “Well, it’s going to help, because we have credit card transactions, and we’re going to understand the effect of COVID, et cetera.” But it doesn’t really get to the root causes. And it doesn’t really get to … Yeah, it’s a bit complacent I think is my point. There is always a fine line you have to work in the existing world. But I think we have to be a bit bolder.
Kirill: 01:03:47
Gotcha. Interesting. Yeah. Yeah. It’s a very interesting perspective. And I agree, we need to be conscious of how we approach these things. Okay, Emmanuel, a wrapping up question for you. So, from all your experience with economics, with data, with the UN, with the Data-Pop Alliance, from things you’ve seen, what do you think the future of data science holds in the next three to five years? What is coming our way in this space?
Emmanuel: 01:04:36
I mean, I would say some big trends, and then some big hopes.
Kirill: 01:04:41
Okay.
Emmanuel: 01:04:43
Big trends, I mean, increasingly. Yeah, so I think like FinTech, of course, is growing very fast. I think COVID will accelerate a lot of those trends. Likewise, I think we see a big move towards digital identities with all the risks and benefits of many people having actually dual identities. And so I think for data scientists, it will mean being aware of the surroundings of these mega technological trends, and not focus only on crunching data. And so increasingly, they will have, and we will have to think about architectures, distribute architectures to store and analyze data and connections between different industries, especially around personal data and digital IDs and digital transfers.
Emmanuel: 01:06:05
In terms of hopes, my hope is that more and more young people get into the field of data science, but I would say with a purpose. I hope that, for instance, if a very smart young woman, or student or man wants to get into data science, that maybe it will be controversial to say that, but maybe they won’t choose to work for the Amazon data science team as cool as it is, as much money as they can make. It’s very enticing to be asked if you want to join the Amazon data science team, just as an example. I hope that we need people with good brains, good hearts, good minds. There’s this quote that says, “Science without conscience is the soul’s perdition.” And so I think data science-
Kirill: 01:07:05
Source perdition, what does that mean?
Emmanuel: 01:07:08
The soul’s perdition.
Kirill: 01:07:10
Oh, soul’s, and what is perdition? I’m not familiar with that word.
Emmanuel: 01:07:17
Yeah. The loss of the soul.
Kirill: 01:07:19
Okay, gotcha. Thank you.
Emmanuel: 01:07:21
And so yeah, you could apply of course, the same or just add data in front of it, that data science without conscience is the soul’s perdition or the loss of the soul.
Kirill: 01:07:32
Yeah.
Emmanuel: 01:07:33
And so my hope is that there will be more of a drive towards projects, and jobs activities, leveraging the very, very massive, impressive power of data science for higher purposes than maximizing the revenue of any given corporation.
Kirill: 01:08:01
Lovely. I love that, data science with a purpose. Emmanuel, that’s a lovely note to wrap up on, and a good parting thought for me to think about, for our listeners to think about as well. I want to thank you for coming on the show. It’s been a huge pleasure and great discussion. At the same time, before we …
Emmanuel: 01:08:24
Thank you very much.
Kirill: 01:08:24
Yeah, but before we go, I’d like to ask you, what’s the places our listeners can get in touch, contact you or maybe take part in some of the great work that you’re doing, or at least follow the progress?
Emmanuel: 01:08:41
Well, in two ways, one is just the typical online media that we have. So the website is Data-Pop Alliance, datapopalliance.org, in one word. And then of course, like Facebook, Twitter, et cetera. Also on LinkedIn, so under my name Emmanuel Letouzé. And then, of course, people can also send me an email, if they want. And so I’m happy to give my email which is my first initial and last name. So E-L-E-T-O-U-Z-E as in zebra, E-L-E-T-O-U-Z-E eletouze@datapopalliance.org. I’ll be happy to get any questions, comments, suggestions, and I’ll make sure to answer.
Kirill: 01:09:35
Awesome, thank you very much. And final question, what’s a book you would like to recommend?
Emmanuel: 01:09:42
I mean, the book that I just actually bought recently is … I think we got disconnected. So a book that I bought recently is called The Technology Trap. And then the subtitle is quite enticing. It’s Capital, Labor, and Power in the Age of Automation, by Carl Benedikt Frey from Oxford and so yeah, that’s what I’ve been starting. So I’m currently reading this.
Emmanuel: 01:10:17
And other than that, the book that I think was the most formative and important for me was The Great Transformation by Karl Polanyi, which was written in the 1940s. It’s a classic in political economy, develop and economics, The Great Transformation. I think, of course, it resonates today, because I think we are also in the midst of a great transformation, because of technology, COVID, and many other major trends.
Kirill: 01:10:53
Awesome, fantastic. So The Technology Trap by Carl Benedikt Frey and The Great Transformation by Karl Polanyi I think.
Emmanuel: 01:11:02
Polanyi, P-O-L-A-N-Y-I.
Kirill: 01:11:06
Polanyi.
Emmanuel: 01:11:06
Yeah.
Kirill: 01:11:07
Awesome.
Emmanuel: 01:11:07
Both are called Karl, but yeah it’s just-
Kirill: 01:11:10
What about, why did you pick up Technology Trap? What are you hoping to get out of this? What are you aiming to get out of this book?
Emmanuel: 01:11:18
So I really like history. I studied history and what really interested me in this book, reading the reviews, and some of the executive summaries that it looks at the Industrial Revolution, also, and it tries to glean lessons from the Industrial Revolution about the role of technology, and the very disruptive effect of technology. And so, in a nutshell, a lot of people lost their livelihoods, their jobs, during the Industrial Revolution.
Emmanuel: 01:11:58
It wasn’t for five or 10 years, like we think that this process of adjustment is quick. And yes, some people lose their jobs, but then we’re all better off, but it took generations. And so it was very, very disruptive. And so today, with automation, with the rise of AI, et cetera, people are concerned for privacy, there is a big pushback against technology. And we can just brush it off saying, oh, we’ll adjust. We’ve gone through that.
Emmanuel: 01:12:29
I think if we push too hard, and if we don’t do this the right way, I think there could be very big backlashes. Democracy, I think is on the line. We see political upheavals all around. And so I think it’s a bit of a … What I’m expecting to get is some pointers about some lessons of the past to help me think about and navigate these current and looming trade-offs, and try to find the right pathways.
Kirill: 01:13:09
Wow. That’s very, very cool. I like the conscious approach to picking a book. That’s a great example of that. Awesome. Well, there you go. Technology Trap, if anybody’s interested. On that note, Emmanuel, thanks so much for joining us today. It was a huge pleasure having you on the show.
Emmanuel: 01:13:25
Thank you very much. Yeah, likewise. And so yeah, good luck. And yeah, all of us, be well and safe, and I would say be kind.
Kirill: 01:13:40
So there you have it, everybody. I hope you enjoyed this conversation with Emmanuel as much as I did, and got lots of valuable insights and takeaways. There were so many interesting topics that Emmanuel could comment on from his experience from the work that he’s done, whether with the United Nations or through Data-Pop Alliance, or through other projects he’s been involved with. It’s exciting to see people pioneering or taking charge in spaces of global world problems, and using data to address them.
Kirill: 01:14:16
My personal favorite part was about measuring versus impacting. That was a really cool way of describing how data science can be used, whether it can be used just as a lens to view the world. But really, what’s the big purpose of viewing the world through that lens, the lens of data if you’re not going to do anything about it, if nothing’s going to change? So that second part, using data as a lever for change is an important thing to always keep in mind.
Kirill: 01:14:51
I also appreciated quite a lot what Emmanuel said at the end that we need to be using data consciously, and that data science or the saying goes, “Science without conscious is the soul’s perdition.” But also you could add data and make it data science without conscious is the soul’s perdition or the loss of the soul. So definitely something to think about.
Kirill: 01:15:20
And on that note, as usual, you can get the show notes for this episode at www.superdatascience.com/413. That’s www.superdatascience.com/413. You will find the transcript for this episode, any materials are mentioned on the show notes, and your links, and Emmanuel actually promised to send us a cartoon that he drew up to put up there, so you can check it out there as well. If you enjoyed this episode, then make sure to leave a review on iTunes or wherever you’re listening to this podcast to help spread the word about this podcast to the world, and let’s make it a better place together. On that note, I’ll see you next time, and until then, happy analyzing.
