The following is a conversation between Jake Porway, the founder and  Executive Director of DataKind and Denver Frederick, host of The Business of Giving on AM 970 The Answer in New York City. This transcript has been lightly edited for clarity.


Denver: There is growing consensus that the time to use data for social good has arrived. And as much as people acknowledge that, they often don’t get too excited by it because a messenger–frequently, a data scientist–can get us a little lost in the weeds of what they’re trying to do. But we have no such worries this evening because my next guest, aside from being an exceptionally gifted data scientist, is also an equally gifted messenger for this revolution. He is Jake Porway,  the founder and Executive Director of DataKind. Good evening, Jake, and welcome to The Business of Giving.

Jake: Good evening, Denver. Thanks for having me.

Denver: Before we dig in too deeply about what you do and how you do it… First, give our listeners a brief overview of the mission and objectives of DataKind.

Jake: DataKind–we seek to harness the power of data science in the service of humanity. And what that means is that we team up data scientists–these statisticians and programmers– to volunteer alongside social organizations—to co-create solutions that are going to maximize their impact.

Denver: Great! Before we get into the work that you do, let me start with the very basic question: What is data? What is data to you as a data scientist? What is data to someone working in the social sector?

Jake: That is such a great question, and one I love talking about because most people that we talk with hear the word “data” and think of spreadsheets… or the matrix… or anonymous zeros and ones. It’s a really foreign, impersonal kind of thing. It’s unfortunate that people think of it that way because data is actually so much more exciting!  Data is really anything that is digitized, and it’s anything that is recorded. So, that means: not just your spreadsheets!  But that’s all of the information coming off our cellphones, about everything we do there!  It’s about everything on our laptops– like what position the marker is right now in this podcast.  That’s digital information!  That’s data!   It’s satellite imagery; it’s all your Fitbit data. And so we’re facing this arrow coming into you now– where data that you used to have to go out and hunt…it used to have to be a scientist to go out and record data about the world with your tools and your microscope…  Now, it comes burbling up out of every activity that we do that’s digital!  And so there’s now just this endless trove of digital information about our activities and lives that could show us new patterns about ourselves, about our world, and about our communities.

What Plato basically said was: People make their decisions based on emotions, upbringing, and knowledge.  And what I love about that is I think he put knowledge third for a reason. There are so many things that go into our decision-making; there are so many emotions and so many fears, that if it only took data, we wouldn’t have a debate about climate change!

Denver: Well, I want to take this a little bit further, Jake. You talk about data for humans,  and you also talk about data for machines. What is the distinction you’re trying to make here?

Jake: Well, I think it’s a really important distinction to make because when people hear “data, ” they often tend to think of the kind of marketing report shown to a big boardroom, with a pie chart or a graph of information going up and to the right.  I think so many people use data to try to change someone’s mind to say: “Hey, let’s look. Here’s what sales are doing. And so therefore you should make a different decision about how you’re selling. “  Or “Here’s what’s going on in the world, and we’re going to show you through the data.”  And that’s very important.

Obviously, we use data as a way to communicate what’s going on in the world all the time. But the really new kind of fundamental opportunity that we have is that with the outbreak of computing…of low-cost computing…the fact that anyone can get a laptop or has access to cellphone technology means that you can also feed that data.  Not just to try to make a human make a different decision, but it can be fed into computers that can mine this data, explore it to find new trends, to understand things that we haven’t necessarily seen before. They either allow us to learn something new or do something at scale in millions of computers  across the globe. And just to drive that distinction home, a lot of people,  I think, believe that when you show someone data, they’ll do something different.  

There’s a great quote by Plato that I’m going to misquote because I don’t have it in front of me. What Plato basically said was:  People make their decisions based on emotions, upbringing, and knowledge.  And what I love about that is I think he put knowledge third for a reason. There are so many things that go into our decision-making; there are so many emotions and so many fears, that if it only took data, we wouldn’t have a debate about climate change!  We wouldn’t have a debate over vaccines and autism. There are a lot of human emotions that go into making a decision, so data is just one piece of that. That’s the data for humans piece.

The data for machines piece can tell you: “Hey, if you’ve already made a decision–let’s say, for example, that you do agree climate change is manmade, and we need to do something about it–computers could help you crunch over terabytes of data to understand what is the optimal intervention?  Or, if we do rebuild sea walls, what are the simulations and implications that could occur from that? So it’s less about using data to change somebody’s mind because now you’re facing behavior change and all of the political, ego, and social issues that go with that. And instead we’re saying: “Where are the opportunities to use data and technology to help us see worlds and inform our decisions in ways that we’ve never seen before?”  Not change our minds, but change our actions!”

Denver: Let me pick up on “using data to change people’s minds,” because that really speaks to confirmation bias.  People tend to pay attention to data that confirms their view of the world, and either ignore, or even denounce data that does not. If Fox News comes out with a poll, my Progressive friends would say: “Well, of course it shows that; what would you expect from Fox?” And if the New York Times does the same thing, my Conservative friends say: “Of course the New York Times is going to indicate that.” So the question is, how can data be used in a fact-based, well-designed, objective way to really inform people to make different decisions?

Jake: Well, I say we are notoriously bad as humans in making decisions around data. There’s a quote I love that says: “If you torture the data enough, it will confess.” That means to me:  even two seemingly impartial folks could take the same set of data… depending on what they want to see in it or show…they can weave any story they want. And I think this is the human condition.

And this is really the reason that the field of statistics was born.  People recognize that we have this tendency to cherry-pick data that confirms what we already believe.  And so all the methodologies around statistics and statistic remodeling are built with the intention of removing that human fact. If we were to, for example, give half a group of people a drug and not treat the other half with the drug: do the differences actually not just feel like the drug works?… Can we say, statistically, it’s very likely that repeatibly, this has an effect. And so I think the answer to this, although it may sound a little dry is to increase the use of not just data literacy, but statistical literacy. Having people understand that bias is there, and these tools of statistics may help us fight back against it.

I say we are notoriously bad as humans in making decisions around data.  There’s a quote I love that says: “If you torture the data enough, it will confess.” That means to me:  even two seemingly impartial folks could take the same set of data… depending on what they want to see in it or show…they can weave any story they want. And I think this is the human condition.

Denver: Let me pick up on DataKind in your initial statement.  You are overwhelmed with requests from nonprofit organizations seeking the kind of assistance that you provide. Jake, how do you go about selecting which projects your organization is going to be engaged in?  And, how do you go about matching up these pro bono data scientists to work on a particular project?

Jake: It’s a great question, and I should say that when we’re talking about the projects that we work on: DataKind’s mission is to help nonprofit and social organizations maximize their impact. So,  if you are a group that is trying to reduce homelessness, we want to see: where can data and algorithms help you reduce  homelessness even more and make your operations more effective?…   See things you couldn’t have ever seen before that are going to supercharge you? With that in mind, when we look for organizations, we’re looking for a special kind of organization. On the surface, we want someone who, of course, has a really good theory of change. They need to be able to show us why it is their activities presumably do, say, reduce homelessness. Because only then can we step back and say: “Well, here’s how data and algorithms are going to make that even better.”  If there’s a shaky path to success, it can be harder for us to figure out where to jump in.So there’s that.  

But the other thing that’s been so critical for finding the right organization has to do with culture. Like we said at the top of the show, data is dry.  It can be scary. We haven’t talked about this–data can be weaponized against people– showing why they shouldn’t get funding, or why they’re not doing well. There are privacy concerns. And so,  we’ve really looked for organizations that want to embrace data as a resource, as a utility, as a tool to give them extra foresight into what’s going to happen… or extra efficiencies, and that usually requires someone at the top having that kind of entrepreneurial, innovative spirit, saying, “Look, I want to do this… to use data to drive decisions, not just to drive what I already know, and I’m willing to face the hard truth.  If you show this isn’t working, or if we try something that doesn’t work:  Great!  That’s a learning lesson to me.”

Denver: Nothing more difficult, I think, than creating a data-driven culture… which isn’t one already.

Jake: Yeah!

Denver: And I find so many organizations that like to use data to basically support what they have already have decided to do.

Jake: Absolutely.

Denver: And it’s just a tail wagging the dog. You see it all the time. So, let’s say you pick one of these projects, based on your criteria. You get your pro bono, data scientists to work. You do this in three phases:  Your first phase is going to be project scoping, Then project execution. Then the third phase will be wrap up and evaluation. Walk us through this process, and tell us about how long each piece of it takes.

Jake: Sure! So I think if anyone has been through a process of building a solution with a partner– like a tax solution, or even as a consulting job, you can probably imagine what these phases look like. I know you had Jim Fruchterman on the show from Benetech, so I’m  sure maybe he talked a bit about this.  When you’re building anything for someone, you want to understand their problem and scope the problem.  You want to build the thing, and then see if it worked.

Where I think the nuances come in with data science, is that we face a couple of interesting challenges, highlighted in each of these phases. So, in project scoping, what we’re looking for is to understand:  where can algorithms come to bear on the organization’s work?  And the biggest challenge we face–with the explosion of data since 2010, or the IPhone coming out in 2007– is that a lot of folks have difficulty understanding exactly what it is they could use. So the scoping phase we’re talking about–is really a big dialogue. We’ll go in with an organization and say: “Just tell us about your mission; let’s not talk about the data. Let’s see what the challenges are.  There’s really a great kind of one-two…. dance that goes on.

Denver: Little tension… back and forth.

Jake: Yeah! But it’s a creative…

Denver: A creative tension. A good one, a healthy one!

Jake: Yeah! We’ll bring a data scientist in with a nonprofit. A classic set up will go like this: The nonprofit will say: “Hey, we got this data set.  Now what?” And the data scientist will go: “Well, what are your biggest challenges?” They’ll say: “Well, we have trouble figuring out that of the constituents who come to our homeless shelters, some 50% just disappear!  Why is that?” The data scientist may say: “Well, you know, if we took that data about who’s coming in the door, and we also combine that with some information that the city makes available about voucher programs, we might be able to predict who’s going to fall off the system.” And this is usually an “Aha” moment, right? The nonprofit may say: “ Whoa, whoa, you could do that? I’ve just been using data to show my funders and to put a prospectus together. I can actually use this. If you could do that, have I got an idea for you!” And then you just see this wonderful synergy– that they’re talking about what data is available, what the challenge is, and what can be done with the technology. So I focus–I know you mention the three phases, but really that scoping…

Denver: That project phase, that scoping is the most important, and probably the most difficult as well.

Jake: Absolutely! It’s the genesis of the whole project. And one of the other challenges that I think is important to know is the world we live in today is incredibly decentralized. Organizations that are tackling big sector-wide challenges live in a community of people–say, you’re all tackling homelessness.  The data around that doesn’t necessarily live neatly contained within each organization.  There could be data about homelessness in one homelessness organization. There could be data  that the government has made publicly available about social services.. To really stretch this…. satellite imagery of people waiting in lines to get into homeless centers.  And all of that is data that could be used to bear on the problem. So there’s also that piece,  and, of course, the data scientists have walked away on Silicon Valley and on Wall Street, so they’re working at other organizations. So the scoping process has really become a weaving process.

The people in this room are the most high caliber folks I’ve met. There’s a machine learning expert from Google; there’s a data scientist from NASA. Oh man!  We’ve got 48 hours to ourselves.  What are we gonna build? We could build stuff that could change the world…

Denver: Well, let’s talk about those data scientists because the ones that work for you do so on a pro bono basis.  I’m sure you’ve heard:  What kind of substantive impact can these pro bono data scientists provide…with weekend hackathons… and a bunch of loosely coordinated activities?  What’s the secret to taking these somewhat unrelated bands of activity, and turning them into meaningful engagements that will really advance the work of the nonprofit organization?

Jake: It’s a great question… And I know many people hear  “volunteer pro bono”  and think low quality or low skill level.  But I have to say the people in this network are anything but. And actually before touching on the pro bono piece, I think it’s worth talking about what the heck a data scientist is.  If I were listening to this, what are you talking about, right? So because we mention there’s this new explosion in data and new explosion in computing, data scientists are the people who can pull all that data together and get computers to make sense of it. So they’re the people who write Netflix’s recommendation algorithm– that watches everything that people rate, everything they watch, and is able to crunch all that down and say: “Hey, you wanna watch “Notting Hill.”

Denver: Yeah. And they’re right some of the time.

Jake: Yeah! That’s right. Nothing’s perfect. And so these folks… they’re a very rare breed of both computer programmers and statistical modelers. And what that means is: they’re incredibly rare; they’re super expensive; and industry has figured out they’re valuable because they drive everything from Facebook to Twitter to Netflix to Spotify’s value. So they are virtually unattainable if you don’t have a lot of resources. So that’s a profile of the folks we’re talking about, and I might consider myself a data scientist based on my training.  

To talk about the pro bono piece of this, I have to tell the story of how DataKind came into existence. I was sitting around at my first hackathon…And  hackathons, for the uninitiated, are these weekend events where a bunch of technologists and data scientists get together and they just say: “Hey, what can we build? We got all this data; we got all these computers.  Let’s just build something cool!”

And I thought to myself, going into this room: “Wow, this is crazy.” The people in this room are the most high caliber folks I’ve met. There’s a machine learning expert from Google; there’s a data scientist from NASA. Oh man!  We’ve got 48 hours to ourselves.  What are we gonna build? We could build stuff that could change the world. We could build stuff beyond what our bosses or the government thinks we need… We can build amazing stuff in just 48 hours–just 48 hours!  

We built stuff that was super unfulfilling. It just depressed me so much, because what people built were… like apps to go find local deals.  Or someone made some app that was an improvement on helping you organize your iTunes library. And using this advanced technology, advanced machine learning and data science…

Denver: For what?

Jake: …for what? Yeah. I mean, look, don’t get me wrong. I love that my iTunes are  organized.  But in the vast problems of the world, that ranks pretty low to me. And I think this is largely because most data scientists skew young, skew wealthy, skew white male to be honest…

Denver: Yeah, true.

Jake: And their problems aren’t finding low-income housing. Thery’re finding cheap beer. So, there’s this incredible energy, incredibly high-skilled people getting together on their own volition to build pretty capable things in a short period of time. It felt to me that they just needed a channel to bring that towards social impact causes. What we found was so many of them wanted to be doing that work; it’s just those problems aren’t as readily available. You can’t just dream up a new solution to homelessness.  You have to spend a lot of time, as we mentioned, talking and working at it.

This is all to say that the profile of folks we have in our community, they are data scientists from Buzzfeed, Google. There are professors of data science and, sure, when they get together for a weekend, we don’t expect them to solve a massive problem.  But what we have found is that getting these data scientists together for what we call a “DataDive,” –which is like our version of a hackathon–We invite three social organizations with their challenges and their data sets to work together with these data scientists. Two really big things happen: the first is that you get a huge number of prototypes, research moments, really “A-ha!” moments for the social…

Denver: Yeah, and real fast too.

Jake: Oh, yeah! Crazily fast! Setting up the problem is half the battle. If you’ve got people coming in with the right challenge and the right data sets, in 48 hours, people can make a ton of progress.  We’re talking about everything from building visualizations of “stop and frisk” in the New York Police Department, to building algorithms that predict the urgency of human rights violations in Amnesty International. This is happening in a weekend. And so really the benefit there is that kind of opening the door to social organizations. “This is what’s possible. Come, feel, and touch it!  You probably didn’t imagine the age we live in!”

Denver: You’re absolutely right.  Hey, I don’t think there’s been a lot of energy on the demand side, ‘cause I don’t think social organizations even knew this stuff was possible until recently!

Jake: And it’s not just them. I came from for-profit industry.  This is all so new; everyone is getting their head around it.

Denver: Fair enough.

Jake: It’s Frontier Land.

Denver: Well, let’s talk about a couple of examples–and one that I was particularly impressed with–was the work you did with Crisis Text Line. First: tell us, what is Crisis Text Line?  And then, how were you able to help them?

Jake: Totally. Crisis Text Line is an organization that seeks to counsel teenagers that are in crisis.  They’re a fantastic organization… going back to how we pick organizations…because they not only have a very clear model–where they connect teens who are in crisis to counselors via text message–but they also have a data science spirit from the start.

And Nancy Lublin, who was the CEO there, has talked about this.

Denver: I know her, yeah.

Jake: Yeah. So you can vouch for this. One of her first hires was a data analyst/data scientist–because this is gonna be critical, so we love groups like that. And the work they’re doing is really impressive.  Teens don’t use the old crisis hotlines that you might remember like the 9 line where you call if you were in trouble or suicidal. Because teens don’t call on the phone…

Denver: They don’t use a phone. Don’t know how to use a phone!

Jake: Hey, exactly right.

Denver: To talk at least.

Jake: Yeah, right. So, Crisis Text Line provides that text-based counseling and meets teens where they’re at. And from what I understand, they now expanded to hundreds of cities.

So, one of the problems they were facing was that there’s a class of people who use the service called  “repeat texters.”  And the idea is that some people will text the service without an urgent concern, and they’ll just keep texting back. This is like people you hear stories about who go to the ER because they’re lonely, or because it’s a safe place to be.  That’s all well and good, but it really drains the resources that Crisis Text Line could be using to help teens in need.

And so they wondered, could we use our data? All this data…and when I say data, I mean the literal text messages, the time of the text, the words of the text, the behaviors of the people… to understand who is likely to become a “repeat texter.” And so they used to just wait 16 text messages, and if you haven’t said something urgent by then they say: “Okay we’re gonna push you somewhere else; you’re just taking up time.”  So we team them up with a data scientist from Pivotal Labs, and together they crunched the data.  They built the statistical models; they wrote algorithms; they took all this information in.

It allowed them to see with just about 5 texts, with about 90% likelihood, who was going to be a “repeat texter.” And so they could reroute them in that moment way more effectively. And this has led them to believe they can now serve over 10,000 more teens in crisis per year than they could before with the same resources.

Denver: That’s fantastic!

Jake: If you’re a nonprofit, for the listeners on the show, you know how resource constraints can be. If you can save money or time anywhere…

Denver: And better serve your constituency, it doesn’t get any better than that.

Jake: Yeah. And on that note, there was a cherry on top that they even claim they were getting better reviews from the people, even the repeat texters, saying: “Thanks! I didn’t want to keep texting back, I wanted to go somewhere else…Thanks for handling me more quickly.” So it’s a great example of where data and algorithms made their operations that much better.

Denver: Throughout the entire system. You’ve also noted some interesting work here in New York City with the New York City Parks Department. How were you able to help them?

Jake: I love the Parks Department, if for no other reasons than when they talk about New York City’s trees, they refer to them as “ the urban forest,” and I find that beautiful…

Denver: A great way to frame it.

Jake: It’s a gorgeous image. I’m already sold. And one of their data challenges was they said: “We have a lot of programs we run; we want to keep people safe in the city, and the parks well-kept. So, one of the things we do is we prune tree limbs.  We’ll see kind of a suspicious tree limb hanging out over the street, and we’ll go out, cut it down preemptively to make sure it doesn’t fall on somebody, or fall on a car and cause damage. And the thing is we do this kind of by gut… We have evidence; we know it works,  but can we test that?”  And the nice thing is that the Parks Department has data about every single tree in that urban…

Denver: Every single tree?

Jake: Well, I should do a fact check…

Denver: Wow, that’s impressive!

Jake: Many, many trees. But they know a lot of really detailed information about: when it was planted, what type it is, what treatment has been done to it, if anyone pruned it or anything like that. So, they’ve got this rich database just sitting there.  We team them up with the data scientist–this guy Brian D’Alessandro– who works at an ad company. Spends his whole day figuring out if he shows you an ad…

Denver: Yeah. Am I gonna click or not, right?

Jake: Exactly! Gonna click or not, right?

Denver: Best minds in the world trying to figure whether we’re gonna click an ad or not.

Jake: That’s right!  That’s the great Jeff Hammerbacher quote. And it’s great though because that same premise, “if I show you an ad, will you click or not?” applies exactly to these problems. If I prune a tree limb, are there fewer emergencies on this block, or not? And so here you can port the same algorithm being used at an ad company– big data– right into the social impact challenge. Lo and behold!  Brian was able to find that there actually are 22% fewer tree emergencies on those blocks where the New York City Parks Department is pruning these limbs.

Denver: Keep on cutting.

Jake: Keep on cutting, right? At least you’ve got something to compare against. And you now see other urban forestry programs writing to New York and saying:  “Whoa, how did you get that number? Can we do this?” And we love seeing that, because we know one of these innovations for one organization often has legs elsewhere across the sector.

Word clouds to me are like the fast food of information communication. They are like deep-fried spreadsheets.

Denver: You’ve also worked with the American Red Cross, correct?

Jake: Yeah, absolutely! That was a great project that one of our chapters did in DC.  We have a volunteer chapter network that does this work, and they teamed up with them to help understand how they could reduce injury and death from fires in homes. There are over 25,000 people every year who die in fires that they think could be prevented simply by having a working smoke detector. Lots of people don’t have the smoke detectors, and so the Red Cross pledged: “We’re going to get a million more smoke detectors and free fire alarm education out into America to prevent this problem.” But where do you start? How do you begin to do it? Just go knock on every door in America? There’s no way!

Denver: That’s what I would do.

Jake: Yeah, right!  So, could we take a data-driven approach? And they teamed up with volunteers in DC who said: “Let’s combine data about the communities that the government makes available. Let’s get data about fire-related injuries and deaths that have happened.” And from that, they were able to build statistical models that could predict where the areas are that are least covered… just through our communities that are least likely to have smoke alarms. Then they went further to say, “Where are the households where we predict in the future…based on past data…we predict are going to have challenges?   That was the second one. And then third…and this is kind of a nuance on the second… Where is it most likely, that if we bring this education program to those places we’ve predicted are gonna have fires, is it going to make the biggest difference?

And that’s everyone’s holy grail..  Where am I gonna have the most impact?

Denver: Absolutely!

Jake: And so they built a great fire risk map.  It’s up and live; you can see it on And it’s going to be used by the Red Cross to figure how to inform their services. And worth noting, it started out at one of those weekend DataDives…

Denver: There you go.

Jake: …where you said: “Hey, what can happen from that… while the team in the Red Cross were so excited by those results in just 48 hours, they kept working together over the next year in their spare time. Lo and behold, this is the result!  It’s going to save American lives all across country.

Denver: All great stuff. I’m gonna go to a bit of  a lightning round with you, Jake, if I can?

Jake: Sure, I’m on it.

Denver: Five quick questions. You hate Word Clouds

Jake: Oh… do I ever!

Denver: Why?

Jake: I don’t think we can do this in a lightning round.  But I will say this: data is confusing. You want to communicate data to people in a way that gives them the “So what?” Word clouds to me are like the fast food of information communication. They are like deep-fried spreadsheets.  You look at a word cloud, if you don’t know what I’m talking about, they’re those big words that you say, Obama speech said: “America” –biggest in the middle. And you feel good because it looks cool, and you can kind of understand it. But beyond knowing what the biggest word is, what do you get out of it?

Denver: Tells you nothing.

Jake: It tells you nothing. It doesn’t tell you anything about the sentences; you can’t tell what the content of something is just by the most number of words…

Denver: No, I look at the colors.

Jake: …colors. Yeah! And the colors mean nothing. It’s just a disgrace to actually conveying information…

Denver: Number 2: If you were starting a career as a data scientist and had only one program to work with, what would be your tool of choice, and why?

Jake: Oh, for the statisticians out there, I cut my teeth on R.  And it’s going to offend everyone that has ever trained me, or known me, to say that I think I would say–Python!

Denver: Wow.

Jake: Yeah. It’s a general purpose language; you can program most things, as well as do statistical computing. Start there, and then work your way up to R.

Denver: At DataKind you had a “no jargon” rule. How did that get started, and how do you enforce it?

Jake: Oh, great question! Well, we do use a patented NJR system that is the No Jargon Rule system. We basically make sure we get rid of acronyms across the board. The rule came into practice because our world is bringing together data scientists and social organizations. They’ve got their own terms that neither one knows. And very few people, as Henry Timms from the 92nd Street Y pointed out, who know what both an API and SDGs are. And if you’re scratching your head, this is because you’re probably on one side or the other. So we say…

Denver: I think many don’t know both.

Jake: Yeah! Fair enough, exactly. So, no one wants to be the dumb one.  So we tried to do that for them.

Denver: No, I’ve always put my phone under the table and looked it up. So nobody knew I didn’t know what it was.

Jake: Right on.

Denver: What is the coolest, or one of the coolest data maps you have ever laid your eyes on?

Jake: I’m so dry about this; there are so many flashy, cool data maps out there.  But I always go back to the practical. And I actually think of the old John Snow– cholera– Back then, if people aren’t familiar, it’s a very, very old map during the cholera outbreak. People were like: “How do we stop this?”

Denver: He started data science with that.

Jake: Exactly! He said: “Well, you know, let’s just plot on a map of London where cholera is happening.” And he found real density right around this one water pump. They went, pulled off the handle, and it’s still standing there as a testament to data saving lives.

Denver: And fifth question, final question. Something significant that you’ve changed your mind about in the last five years.

Jake: That’s a great one. I would have to say, I have really come around on–believe it or not–this big data thing. Funny, when I was coming into it, I was trained very much that you build models of as little data as possible.  And it was almost a badge of pride: if you can build a computer system that needs just a little bit of data. And, Lo and Behold!  Having way more data is way more helpful,  and you can learn much more about the world!  So, I switched over. That’s really a kind of wonky thing for the tech nerds out there, but…

Denver: It’s an important one.

Jake: It was a big one for me, for sure.

Denver: Getting back to DataKind, you’re a nonprofit organization; you’ve been growing fast.  You’re spreading your wings across the world. What’s your funding model? How do you generate your revenues?

Jake: Yeah. Thanks for asking. We are a mix of foundation funding and corporate funding. What we are actually funding by the way is– we have about 12,000 volunteers across our globe. We have  6 volunteer chapter networks, and what we seek to do, of course, is in each of those worlds, bring together data scientists and social organizations to co-create the solutions that make them better, and ideally bubble those solutions up from just one-off projects– to things that might help the sector. And so we really lean on foundation funders for that unrestricted funding. Of course, who doesn’t?

Denver: Right.

Jake: I mean there’s not a lot of, if any, data science funding for nonprofits.  So, we really fill that need. If you’re going to be able to innovate with this expensive resourcing, we’re  kind of one of the few games in town!  But we also think of our foundations as true partners. And we have this kind of ethos of reciprocity at DataKind. It’s not just: “Hey, just pay us for this project, and we’ll tell you how it goes.” We’re all figuring out a lot together, so we work with foundations and say: “Hey, please fund us to do this work.  But also, we want to know: how do your grantees help your mission be improved by this? What are you seeing out of this?”   

Denver: It’s a smart way to do it.

Jake: Well, I think it’s the only way. We know data and tech. I’m not an expert in female black leadership in communities. I’m not an expert in homelessness in various cities. It’s not just a funding relationship.  It is a true partnership. And we feel the same way about corporations who, on the other side, have data and technology resources that could be giving back to social good. And so, in the same way, they all fund projects… or fund DataKind’s work, and we say: “Hey, come aboard! What resources can you bring to the fight?”

Denver: You just mentioned a second ago, you have chapters around the world… six of them. Where are those chapters, and how do they further your mission? Are you thinking of additional chapters?

Jake: Oh, fantastic! The chapters are really, in a lot of ways, the backbone of our growth. The chapters are in DC, San Francisco, Bangalore, Singapore, the UK and Dublin. Kind of  a good spread there.

Time zones are a pain in the butt, but well worth it. And they’re really volunteer-led chapters. So, they do what we do at the headquarters here in New York– connecting data scientists and NGOs in their own communities, but with their own special flare. The NGO scene in Bangalore is completely different than what you see in DC. Skill sets of data scientists vary across the world. It’s this great, vast learning network; and the work coming out of them is just incredible . You see so many projects coming out where they’re bringing together these incredible people we would never have access to otherwise, and organizations that are making big strides in building these projects together. Of course, we want to expand our chapter network; we are trying to grow well. The applications are not open now, but we are always scouting for future DataKind chapter leaders.

Denver: I have seen you speak a number of times, and you are a great data storyteller…

Jake: Oh, thank you.

I think the key to good data storytelling, aside from the stuff that you’ll read where you want to find the right medium…is that data is best telling a story when you are looking at it as a process.

Denver: …which is a bit of an oxymoron, I think, to a lot of people.  Because for most people in the social sector, Jake, they’ll tell you that to get your point across, you have to have great stories about your work.  And then you have to back up those great stories with data. Never would they think that data and storytelling could go together, but you have proven that they can. What is the secret of being a great data storyteller?

Jake: Well, I am humbled and certainly consider myself much lower on the ranks of data storytelling than folks like maybe Jer Thorp, who’s a great data artist in this space.

Denver: “Data artist.”  That’s cool.

Jake: It’s the coolest title. I’m so jealous of it.

Denver: Give up “Founder and Executive Director.” Go with “Data Artist.”

Jake: Seriously, right? Oh man, I don’t know how he came up with that one, but it’s great! So, I think that you’re right. People think of data like what you can see in a spreadsheet, right? I think the key to good data storytelling, aside from the stuff that you’ll read where of course you wanna find the right medium for your data story telling, we could talk about the design specifics–I think more than that. It is that data is best telling a story when you are looking at it as actually a process. The storytelling… all along the way, not just: “Hey, I got this number. 55% people improved in my program, and now I’m gonna put it in bold font and show a graph.” That’s the kind of data storytelling I think people think of… where it comes at the end.

I think the best data storytelling is when you are showing what’s happening along the way. I want to see how students are progressing along that path, I want to know:  What are you learning from that?  That’s really when data is almost in the background, and what you’re doing is carrying people along the path of your mission. You are bringing them to some new enlightened understanding, and I think data plays a deep role in that.  But it really gets to shine… not just when it’s packaged at the end… but when it allows people to see what goes on in your work, and what the real impact is.

Denver: They feel like they are a participant as opposed to reading a bunch of recommendations at the end… which is more of a passive activity. Allows them to be  a little bit more actively engaged.

Jake: Absolutely! And it’s easier said than done. It is the hardest part of this…we could do a whole different podcast on that!

And while he could only hope for it, we’re now in that macroscope age. That data– about 9 billion tweets of people around the world expressing their interests, interacting with each other– it’s there. It’s credit card transactions, migration patterns… We have instrumented the world so rapidly and so suddenly that we have given ourselves the macroscope. And data science–these algorithms and statistical models–can help make some sense of that.

Denver: Let me close with this: We’ve been talking about the data revolution that’s going on right now.  But to provide some kind of context that will allow people to understand its significance, why don’t you share with our listeners how this is just the  latest development in a line of tools that  helps us understand our world better?

Jake: Thank you for that. I think that’s needed. And I’ll just emphasize again, like we said at the top:  when people hear “data,”  they’re thinking they’ve got to report to their funders. “I’ve got to to collect data to show them I’m making a difference.” It’s this real kind of survival feeling. “I’ve got to have data, but I also have this love-hate relationship with data; it’s against me; it’s numbers; I hated math…”

Denver: It’s like doing your taxes.

Jake: Yeah! Exactly, right? That’s a shame because we are, without exaggeration, at this new era of data and understanding about the world. I often say we’re almost potentially at a new Age of Reason.  The way  I’ve summed this up I actually borrowed  from a guy I saw…Moritz Stefaner… who uses this analogy I love. There’s a book called …The Macroscope…I think it was in 1979…

Joël de Rosnay came out and said: Throughout time immemorial, humans have built technologies to see the unknown. So you think back to the far reaches of human history… we first built the telescope. That allowed us to see the stars and the galaxies, and we’ve seen the infinitely great. And that allowed us to learn more about our place in the universe than we ever knew before. It’s a huge advancement in human knowledge.

And then you go ahead… we invented the microscope.  That allows us to see the infinitely small– microbes and bacteria– and we make huge advances in human health and medicine. Again, humanity overall learns… it’s a new phase shift. And so he was opining: We’re in 1979 now; what we’re missing though, what we really need is a “macroscope…”

Something that helps us to see the infinitely complex, the patterns of society and of nature that are not observable by the naked eye. And while he could only hope for it, we’re now in that macroscope age!  That data– about 9 billion tweets of people around the world expressing their interests, interacting with each other, it’s there! It’s credit card transactions, migration patterns… We have instrumented the world so rapidly and so suddenly that we have given ourselves the macroscope. And data science–these algorithms and statistical models–  can help make some sense of that.

Denver:  What does it all mean?

Jake: Yeah. It is to me this new moment that allows us to see things we’ve never seen before about the way we work. And so that’s the promise of this age!  Forget the spreadsheets; forget the matrix and binary.  It’s the chance to see ourselves in ways we’ve never seen before.

Denver: That’s a great way of looking at it. So, how do people listening get involved? What do nonprofits do to apply for some assistance? How do data scientists who want to become pro bono volunteers become part of the DataKind team?  And where do people go if they want to financially support this work?

Jake: Fantastic questions. So the easiest way is to go to and go to “Get Involved.”  If you’re a social organization that thinks you could use data science.. even just wants to know if… come aboard, sign up! And just fill out a few questions.  You do not need to know much, honestly!  That’s our job. We know, like we said, this stuff is all new. We’ll talk you through it.   Data scientists? If you want to give back in your spare time, come aboard to the same site. And if funders want to get involved, of course, we always ask foundations to reach out…you can just go to the donate page…

Denver: Can individuals give?

Jake: Yes. Individuals can give… thank you, on the “Donate” page. Thank you for that.  Or foundations or corporations, please get in touch with us in the “Contact” page. We have an ambitious mission. We want to bring data science services to hundreds of organizations over the next few years to really move these social issues forward. So, please, come join us in this fight. It’s going to require everyone.

Denver: Sounds great. Well, Jake Porway, the Founder and Executive Director of DataKind. Thanks so much for appearing on the show. You have a wonderful gift to actually make all this stuff sound like fun.

Jake: It is fun.

Denver: All while helping to change the world for the better. It was a great pleasure to have you on the program.

Jake: It was totally my pleasure. Thank you so much

The Business of Giving can be heard every Sunday evening between 6 and 7 PM Eastern on AM 970 The Answer in New York and on I Heart Radio. You can follow us at bizofgive on twitter and at

Share This: