Sources & Methods #40: Data Science with Eric Schles


Show Notes:

6:50 - [On working at the Manhattan District Attorney’s Office in New York] There is some emotional burnout when the work that you do involves slave traders. That can be challenging for a lot of folks. But overall, the ability to know that the work you’ve done has made someone’s life better - there’s nothing like that in the world. It’s better than any other possible thing you could do.

10:07 - The General Problem is that we don’t want people enslaved. Fairly simple. People are not property. People are people. At a high level, that’s the goal. To have no one in the world treated as property.

This [slavery] problem is fundamentally an economics problem.

[On parents enslaving their kids or family members to beg] If you lift people out of poverty, then that doesn’t happen anymore. That’s a path forward to removing that form of slavery. Because that’s not motivated by malice or cruel intent. It’s motivated by simply not having means to do anything else. And this extreme desperation.

13:30 - There are some things we can do within traditional economic stimulus-style things for developing countries that will alleviate a great deal of the problem space. So I originally tried to work on that. The problem there is that economists aren’t particular interested in this. I don’t think they’re disinterested, I want to be clear. I think that their research dollars come from very specific sources and it’s hard to find research dollars for this.

14:29 - In international development, there’s plenty of money and everyone’s interested in this. But you can’t do this only from an aid perspective. You have to stimulate economies. There have to be network effects. Aid can help and lift specific people out of poverty. But we’ve seen that aid can be detrimental. There’s well documented evidence on this. If you give people aid, they become dependent on aid. And if the aid dries up… and the aid goes away, the cycle of dependence basically leaves these people worse off than they were originally.

15:23 - I was lucky enough while I was in the federal government to see how truly well intentioned so many federal employees really are. And I had the specific pleasure of interacting with someone from USAID. They were working on a project in West Africa. What they did was - it’s mind boggling how simply and brilliant and effective this was - was give folks really old cell phones and really old laptops with Excel. They were able to use this to coordinate crop outputs. And look at the weather.

They leveraged basic technology and they were able to get crop outputs up like crazy. And they brought these subsistence farmers up substantially.

These people didn’t need our help the next year.

If you give people standard tools and you are well intentioned, you can actually make a lot more change. Especially for third world or disadvantaged groups.

22:34 - I really recommend checking out the Polaris Project on Global Modern Slavery.

1:04:04 - Research Question #1: We need to figure out the frequency of trafficking

Research Question #2: I’m very interested in the discovering the probability that someone is being trafficked. How likely is a specific neighborhood to be an area for trafficking?

1:09:30 - Specialization is how I determine what problems to solve. I will usually work on issues that have technical solutions.

1:13:07 - I think understanding how automating works is a great skill, not least because it makes you highly employable. People need to make a living.

Follow Eric’s Work on Github:


Eric Schles on LinkedIn

Sources and Methods #33: Gabe Weatherhead

Gabe Weatherhead 101:

Macdrifter (Gabe's personal site)

Gabe (not really any more) on Twitter

Nerds on Draft


Technical Difficulties (Gabe's old podcast)

Show notes:

6:50 - - Resource to make your website (or others) searchable.

8:50 - Podcasts are great for ruminating on a topic, where you’re just riffing and talking real time about it, but it’s not great for thinking through things. People that I’ve seen who become more podcasters than writers, I think their logic suffers for that. They’re not ever really thinking things through at a deep level other than to have a kneejerk conversation.

11:02 - I think there are a couple different flavors of app developers. There are those who are in it for the money. Then there’s the app makers who have their own personal problem they’re trying to solve. Then of course there’s the career developer. We were more in the bucket of we had a problem we wanted to solve, and we wanted customers to pay to help us solve that problem. We learned a lot. 1) iOS development is not cheap. 2) There’s not much of an equation or plan that you can follow. It’s like viral marketing is luck of the draw and what sticks, no one really knows. We were naive, thinking ‘if you make a good thing, people will come to it.’ That’s not really how it works.

14:52 - We learned that we were really terrible at business because we were in love with our own principles too much. And we learned that you have to give up some of your principles to be really good at business.

19:42 - I have left Twitter without announcing it. I don’t think Twitter and Facebook are good for people, in general. I think that it’s not healthy the way people are interacting with each other, and focusing in on things and they’re too easily deceived by what they think the world looks like, and I’m a victim of that, looking at the world through the kaleidoscope of Twitter, thinking I understand how everything works because I read! The conscious choice to limit my view of the world almost predestined everything. Coming out of this election, I would be shocked because I viewed everything differently whereas maybe the old way was I watched the news, I read different newspapers, I talked with different people, I consumed all of this different content and pulled it in and came to different conclusions.  

24:45 - I left Facebook because of their attitude towards privacy, and the way they’ve repeatedly tried to trick their users into revealing more information about themselves. I’m a big privacy proponent in my own life, I don’t really like algorithms following me. I feel like a bit of a nut talking about this.

There was a conversation where someone told someone to delete their Facebook app. I said, to a large number of people, that’s like telling them to delete their mailbox. That’s their connection with family and friends, that’s how they know what’s going on with their actual associates. And to tell them oh you should get rid of that is naive at best. I don’t need facebook because I have other ways to connect with friends and family, and they mostly aren’t on facebook. (Laughs) I’m a hermit.

I don’t think a non-commercial product will replace facebook. Money is what drives it. So collecting user data will always be the business. And selling ads will be the business because that makes the most money. In order for their to be an alternative that didn’t track people for people who cared about privacy, most people don’t want to pay for that service and they especially don’t want to pay what the company would make on ads instead. So I don’t think that will be replaced. I do think facebook will eventually die away because younger people care about different things. Younger people are using different services.

29:11 - Right now, I don’t think there’s enough money in privacy for people to make much off of it.

Quick App Notes:

I like Slack a lot.

I like Sublime Text. There’s a commercial version I use a lot.

I use Keyboard Maestro a lot.

I have Hazel on Mac. Copied on Mac and iOS.

1Password of course.

I’m really hopeful that the Devonthink Team adds document provider service to Devonthink on iOS.

40:00 I would own an iPhone at this point just to use Devonthink. I have so many notes because that’s how I’ve always been. I’ve always been the guy that created 3x5 cards with bits of information and kept them in a shoebox and created a little tagging system with little colored flags. That was just how I worked. On the computer, it made sense – I’m not going to run out of space.

A great example of this is Christmas shopping. I have a little note that’s just information about my wife’s shoe size, ring size, which is great for gifts. I also have a note with the gifts I’ve already given people, so I don’t get them the same gift five years later. Little bits of information.

43:17 – Apple Notes is great.

49:00 – Task Paper is also great.

1:05:18 – I love mind mapping, that’s a big thing that I do. And my favorite is iThoughts.

1:07:56 – I use Arq which encrypts things before it goes to backup servers. Backblaze is great too. And Little Snitch is useful for monitoring what apps are connecting to the internet.

1:10:33 – I recommend people get this: Napkin. Great little app for Mac.

Gabe’s Picks:


1)    The Hayao Miazaki Blu-Ray Collection


1)    QC35 Headphones by Bose. They are fantastic. Get the iMic flip USB cable.


1)    I use Audible. All the time.

2)    The Subtle Art of Not Giving a F*** by Mark Manson

Sources and Methods #10: Jason Lyall


Jason Lyall 101:

Personal Website (you can find links to all his research here)

Yale Professor Page


Jackson Institute at Yale

Institution Social and Policy Studies

Show Notes:

3:58 - I’m interested in the effectiveness of violence - when it works, whether it works - and what it does to the victims. And I’m interested in insurgents and how they react to violence, and how does it affect them? Do they ramp up their attacks, or scale down their attacks?

4:40 - Counter insurgency is not a new practice. But it’s often been approached in a fairly crude fashion. What we’re trying to do is bring some new literature to the table from spatial economics literature or survey literature and better understand it.

9:00 - Study looking at artillery fire that the Russians were lobbing more or less randomly into populated settlements. They were doing this as a way of trying to separate the people from the insurgency, showing people what the consequences of continued support for the insurgency would be. Most of our theoretical literature would say that indiscriminate violence is incredibly counter productive and would not be be effective - would generate new insurgents and new grievances - so you should see this uptick in violence. What’s interesting in the Russian case is that you don’t see that uptick. It actually seems to have decreased insurgent violence in the periods after the shelling. So it’s an outlier case... And I would say it is what has led to the dismantling of the insurgency in Chechnya.

12:12 - I think the way I would read that paper - we should not assume that violence has any one kind of effect. We should be looking for the different effects, and looking for the conditions of those effects.

14:30 - The metrics used in this research is really good for short term effects, but entirely different methods may be necessary to take the long term view and understand those effects. The field lab I run is trying to merge those two perspectives.

17:58 - We’re not good at trends, we’re not good at regional patterns, we’re not even good at generalizability inside the same case we’re studying - and these are problems.

22:00 - It is possible to get the data in difficult areas, but you have to be really really smart and careful. You can do it, but it takes a tremendous amount of time. And the risk of failure is incredibly high. In Afghanistan, I’ll typically have two or three projects going at the same time with the understanding that one of them will likely fail.

23:04 - If you adopt the right methodology, you can get more data than we ever thought possible.

26:22 - I have a love/hate relationship with the Asia Foundation. I know what they’re trying to do, but I have questions about how they’re doing it.

27:32 - The latest Asia Foundation Report here.

27:49 - Surveys for me are about mapping patterns down, rather than getting really deep insights into what a particular concept means.

29:00 - For local level questions or regional questions - I think these are a nice middle ground, using a mid-range methodology, where I can do lots of things but not everything - is the way to go.

29:35 - Smart polling should be a key part of our toolkit, but the thing that we usually lack is baseline data. By the time we start caring about a place, it’s too late to run baselines. So we’re forever behind, because we never know what was there before. So I’m a big proponent of using smart, tailored surveys. Specific questions. Specific areas.

30:54 - With that, these surveys just measure attitudes, not behavior. Important to remember that.

33:01 - In Afghanistan, I think we vastly underuse SMS as a tool.

36:40 - Explaining Support for Combatants during Wartime: A Survey Experiment in Afghanistan (co-authored by Lyall and includes the Endorsement polling technique). A primer on  the list technique in polling.

45:55 - Bombing to Lose? Airpower and the Dynamics of Violence in Counterinsurgency Wars, published in August 2014 by Lyall, found:

Evidence consistently indicates that airstrikes markedly increase insurgent attacks relative to non-bombed locations for at least 90 days after a strike.
We’d be hard pressed to find an example where an insurgency could be destroyed from an air - where air power alone has actually done this.

55:56 - I build project-specific infrastructure when I get to work. Sometimes I’m working with satellite imagery. Each project has it’s own methodology, research. A problem I’m having right now is that there’s too many platforms in play - we’re working to get everyone on the same

1:00:40 - Morning Routine:

  • I knock off around 2 or 3AM. The last thing I do is write a note to myself about what I should be focusing on the next day. If I don’t get centered quickly, I find that I lose the day, getting distracted by email, etc.

  • So that note shows me - which I keep with me - what I need to do, what I have to get through that day, it’s like my touchstone.

  • I set aside an hour or two every Friday about what I want to accomplish the following week, and these notes each morning help preserve my space to think.

  • It’s often a physical sticky - it tells me what I’m doing, and where I am in the master plan. Otherwise I write these things in Evernote.

1:04:22 - Also use:

  • RunKeeper, to remind myself to keep working out.

  • I’m a big fan of turning off my social media, letting it go dark. I use a software blocker often for up to a week at a time to keep me off of it.

Jason’s Book: The Peripheral by William Gibson. And The Book of Strange New Things by Michel Faber.  

Jason’s Film: Guardians of The Galaxy

Jason’s Song: Alt-J’s Taro