All posts tagged GDELT

A Bit More on Country-Month Modeling

My family is riding the flu carousel right now, and my turn came this week. So, in lieu of trying to write from scratch, I wanted to pick up where my last post—on moving from country-year to country-month modeling—left off.

As many of you know, this notion is hardly new. For at least the past decade, many political scientists who use statistical tools to study violent conflict have been advocating and sometimes implementing research designs that shrink their units of observation on various dimensions, including time. The Journal of Conflict Resolution published a special issue on “disaggregating civil war” in 2009. At the time, that publication felt (to me) more like the cresting of a wave of new work than the start of one, and it was motivated, in part, by frustration over all the questions that a preceding wave of country-year civil-war modeling had inevitably left unanswered. Over the past several years, Mike Ward and his WardLab collaborators at Duke have been using ICEWS and other higher-resolution data sets to develop predictive models of various kinds of political instability at the country-month level. Their work has used designs that deal thoughtfully with the many challenges this approach entails, including spatial and temporal interdependence and the rarity of the events of interest. So have others.

Meanwhile, sociologists who study protests and social movements have been pushing in this direction even longer. Scholars trying to use statistical methods to help understand the dynamic interplay between mobilization, activism, repression, and change recognized that those processes can take important turns in weeks, days, or even hours. So, researchers in that field started trying to build event data sets that recorded as exactly as possible when and where various actions occurred, and they often use event history models and other methods that “take time seriously” to analyze the results. (One of them sat on my dissertation committee and had a big influence on my work at the time.)

As far as I can tell, there are two main reasons that all research in these fields hasn’t stampeded in the direction of disaggregation, and one of them is a doozy. The first and lesser one is computing power. It’s no simple thing to estimate models of mutually causal processes occurring across many heterogeneous units observed at high frequency. We still aren’t great at it, but accelerating improvements in computational processing, storage, software—and co-evolving improvements in statistical methods—have made it more tractable than it was even five or 10 years ago.

The second, more important, and more persistent impediment to disaggregated analysis is data, or the lack thereof. Data sets used by statistically minded political scientists come in two basic flavors: global, and case– or region-specific. Almost all of the global data sets of which I’m aware have always used, and continue to use, country-years as their units of observation.

That’s partly a function of the research questions they were built to help answer, but it’s also a function of cost. Data sets were (and mostly still are) encoded by hand by people sifting through or poring over relevant documents. All that labor takes a lot of time and therefore costs a lot of money. One can make (or ask RAs to make) a reasonably reliable summary judgment about something like whether or not a civil war was occurring in a particular country during particular year much quicker than one can do that for each month of that year, or each district in that country, or both. This difficulty hasn’t stopped everyone from trying, but the exceptions have been few and often case-specific. In a better world, we could have patched together those case-specific sets to make a larger whole, but they often use idiosyncratic definitions and face different informational constraints, making cross-case comparison difficult.

That’s why I’ve been so excited about the launch of GDELT and Phoenix and now the public release of the ICEWS event data. These are, I think, the leading edge of efforts to solve those data-collection problems in an efficient and durable way. ICEWS data have been available for several years to researchers working on a few contracts, but they haven’t been accessible to most of us until now.  At first I thought GDELT had rendered that problem moot, but concerns about its reliability have encouraged me to keep looking. I think Phoenix’s open-source-software approach holds more promise for the long run, but, as its makers describe, it’s still in “beta release” and “under active development.” ICEWS is a more mature project that has tried carefully to solve some of the problems, like event duplication and errors in geolocation, that diminish GDELT’s utility. (Many millions of dollars help.) So, naturally, I and many others have been eager to start exploring it. And now we can. Hooray!

To really open up analysis at this level, though, we’re going to need comparable and publicly (or at least cheaply) available data sets on a lot more of things our theories tell us to care about. As I said in the last post, we have a few of those now, but not many. Some of the work I’ve done over the past couple of years—this, especially—was meant to help fill those gaps, and I’m hoping that work will continue. But it’s just a drop in a leaky bucket. Here’s hoping for a hard turn of the spigot.

2 Comments
by Jay Ulfelder on April 8, 2015  •  Permalink
Posted in Measurement, Methods
Tagged , , ,

Posted by Jay Ulfelder on April 8, 2015

https://dartthrowingchimp.wordpress.com/2015/04/08/a-bit-more-on-country-month-modeling/

Down the Country-Month Rabbit Hole

Some big things happened in the world this week. Iran and the P5+1 agreed on a framework for a nuclear deal, and the agreement looks good. In a presidential election in Nigeria—the world’s seventh–most populous country, and one that few observers would have tagged as a democracy before last weekend—incumbent Goodluck Jonathan lost and then promptly and peacefully conceded defeat. The trickle of countries joining China’s new Asian Infrastructure Investment Bank turned into a torrent.

All of those things happened, but you won’t read more about them here, because I have spent the better part of the past week down a different rabbit hole. Last Friday, after years of almosts and any-time-nows, the event data produced for the Integrated Conflict Early Warning System (ICEWS) finally landed in the public domain, and I have been busy trying to figure out how to put them to use.

ICEWS isn’t the first publicly available trove of political event data, but it compares favorably to the field’s first mover, GDELT, and it currently covers a much longer time span than the other recent entrant, Phoenix.

The public release of ICEWS is exciting because it opens the door wider to dynamic modeling of world politics. Right now, nearly all of the data sets employed in statistical studies of politics around the globe use country-years as their units of observation. That’s not bad if you’re primarily interested in the effects or predictive power of structural features, but it’s pretty awful for explaining and anticipating faster-changing phenomena, like social unrest or violent conflict. GDELT broke the lock on that door, but its high noise-to-signal ratio and the opacity of its coding process have deterred me from investing too much time in developing monitoring or forecasting systems that depend on it.

With ICEWS on the Dataverse, that changes. I think we now have a critical mass of data sets in the public domain that: a) reliably cover important topics for the whole world over many years; b) are routinely updated; and, crucially, c) can be parsed to the month or even the week or day to reward investments in more dynamic modeling. Other suspects fitting this description include:

  • The spell-file version of Polity, which measures national patterns of political authority;
  • Lists of coup attempts maintained by Jonathan Powell and Clayton Thyne (here) and the Center for Systemic Peace (here); and
  • The PITF Worldwide Atrocities Event Dataset, which records information about events involving the deliberate killing of five or more noncombatant civilians (more on it here).

We also have high-quality data sets on national elections (here) and leadership changes (here, described here) that aren’t routinely updated by their sources but would be relatively easy to code by hand for applied forecasting.

With ICEWS, there is, of course, a catch. The public version of the project’s event data set will be updated monthly, but on a one-year delay. For example, when the archive was first posted in March, it ran through February 2014. On April 1, the Lockheed team added March 2014. This delay won’t matter much for scholars doing retrospective analyses, but it’s a critical flaw, if not a fatal one, for applied forecasters who can’t afford to pay—what, probably hundreds of thousands of dollars?—for a real-time subscription.

Fortunately, we might have a workaround. Phil Schrodt has played a huge role in the creation of the field of machine-coded political event data, including GDELT and ICEWS, and he is now part of the crew building Phoenix. In a blog post published the day ICEWS dropped, Phil suggested that Phoenix and ICEWS data will probably look enough alike to allow substituting the former for the latter, perhaps with some careful calibration. As Phil says, we won’t know for sure until we have a wider overlap between the two and can see how well this works in practice, but the possibility is promising enough for me to dig in.

And what does that mean? Well, a week has now passed since ICEWS hit the Dataverse, and so far I have:

  • Written an R function that creates a table of valid country-months for a user-specified time period, to use as scaffolding in the construction and agglomeration of country-month data sets;
  • Written scripts that call that function and some others to ingest and then parse or aggregate the other data sets I mentioned to the country-month level;
  • Worked out a strategy, and written the code, to partition the data into training and test sets for a project on predicting violence against civilians; and
  • Spent a lot of time staring at the screen thinking about, and a little time coding, ways to aggregate, reduce, and otherwise pre-process the ICEWS events and Polity data for that work on violence against civilians and beyond.

What I haven’t done yet—T plus seven days and counting—is any modeling. How’s that for push-button, Big Data magic?

7 Comments
by Jay Ulfelder on April 3, 2015  •  Permalink
Posted in Forecasting, Measurement
Tagged , , , , , ,

Posted by Jay Ulfelder on April 3, 2015

https://dartthrowingchimp.wordpress.com/2015/04/03/down-the-country-month-rabbit-hole/

The State of the Art in the Production of Political Event Data

Peter Nardulli, Scott Althaus, and Matthew Hayes have a piece forthcoming in Sociological Methodology (PDF) that describes what I now see as the cutting edge in the production of political event data: machine-human hybrid systems.

If you have ever participated in the production of political event data, you know that having people find, read, and code data from news stories and other texts takes a tremendous amount of work. Even boutique data sets on narrowly defined topics for short time periods in single cases usually require hundreds or thousands of person-hours to create, and the results still aren’t as pristine as we’d like or often believe.

Contrary to my premature congratulation on GDELT a couple of years ago, however, fully automated systems are not quite ready to take over the task, either. Once a machine-coding system has been built, the data come fast and cheap, but those data are, inevitably, still pretty noisy. (On that point, see here for some of my own experiences with GDELT and here, here, here, here, and here for other relevant discussions.)

I’m now convinced that the best current solution is one that borrows strength from both approaches—in other words, a hybrid. As Nardulli, Althaus, and Hayes argue in their forthcoming article, “Machine coding is no simple substitute for human coding.”

Until fully automated approaches can match the flexibility and contextual richness of human coding, the best option for generating near-term advances in social science research lies in hybrid systems that rely on both machines and humans for extracting information from unstructured texts.

As you might expect, Nardulli & co. have built and are operating such a system—the Social, Political, and Economic Event Database (SPEED)—to code data on a bunch of interesting things, including coups and civil unrest. Their hybrid process goes beyond supervised learning, where an algorithm gets trained on a data set carefully constructed by human coders and then put in the traces to make new data from fresh material. Instead, adopt a “progressive supervised-learning system,” which basically means two things:

  1. They keep humans in the loop for all steps where the error rate from their machine-only process remains intolerably high, making the results as reliable as possible; and
  2. They use those humans’ coding decisions as new training sets to continually check and refine their algorithms, gradually shrinking the load borne by the humans and mitigating the substantial risk of concept drift that attaches to any attempt to automate the extraction of data from a constantly evolving news-media ecosystem.

I think SPEED exemplifies the state of the art in a couple of big ways. The first is the process itself. Machine-learning processes have made tremendous gains in the past several years (see here, h/t Steve Mills), but we still haven’t arrived at the point where we can write algorithms that reliably recognize and extract the information we want from the torrent of news stories coursing through the Internet. As long as that’s the case—and I expect it will be for at least another several years—we’re going to need to keep humans in the loop to get data sets we really trust and understand. (And, of course, even then the results will still suffer from biases that even a perfect coding process can’t avoid; see here for Will Moore’s thoughtful discussion of that point.)

The second way in which SPEED exemplifies the state of the art is what Nardulli, Althaus, and Hayes’ paper explicitly and implicitly tells us about the cost and data-sharing constraints that come with building and running a system of this kind on that scale. Nardulli & co. don’t report exactly how much money has been spent on SPEED so far and how much it costs to keep running it, but they do say this:

The Cline Center began assembling its news archive and developing SPEED’s workflow system in 2006, but lacked an operational cyberinfrastructure until 2009. Seven years and well over a million dollars later, the Cline Center released its first SPEED data set.

Partly because of those high costs and partly because of legal issues attached to data coded from many news stories, the data SPEED produces are not freely available to the public. The project shares some historical data sets on its web site, but the content of those sets is limited, and the near-real-time data coveted by applied researchers like me are not made public. Here’s how the authors describe their situation:

While agreements with commercial vendors and intellectual property rights prohibit the Center from distributing its news archive, efforts are being made to provide non-consumptive public access to the Center’s holdings. This access will allow researchers to evaluate the utility of the Center’s digital archive for their needs and construct a research design to realize those needs. Based on that design, researchers can utilize the Center’s various subcenters of expertise (document classification, training, coding, etc.) to implement it.

I’m not happy about those constraints, but as someone who has managed large and costly social-science research projects, I certainly understand them. I also don’t expect them to go away any time soon, for SPEED or for any similar undertaking.

So that’s the state of the art in the production of political event data: Thanks to the growth of the Internet and advances in computing hardware and software, we can now produce political event data on a scale and at a pace that would have had us drooling a decade ago, but the task still can’t be fully automated without making sacrifices in data quality that most social scientists should be uncomfortable making. The best systems we can build right now blend machine learning and automation with routine human involvement and oversight. Those systems are still expensive to build and run, and partly because of that, we should not expect their output to stream onto our virtual desktops for free, like manna raining down from digital heaven.

2 Comments
by Jay Ulfelder on January 22, 2015  •  Permalink
Posted in Machine Learning, Methods
Tagged , , , , ,

Posted by Jay Ulfelder on January 22, 2015

https://dartthrowingchimp.wordpress.com/2015/01/22/the-state-of-the-art-in-the-production-of-political-event-data/

Another Note on the Limitations of Event Data

Last week, Foreign Policy ran a blog post by Kalev Leetaru that used GDELT to try to identify trends over time in protest activity around the world. That’s a fascinating and important question, but it’s also a really hard one, and I don’t think Kalev’s post succeeds in answering it. I wanted to use this space to explain why, because the issues involved are fundamental to efforts to answer many similar and important questions about patterns in human social behavior over time.

To me, the heart of Kalev’s post is his attempt to compare the intensity of protest activity worldwide over the past 35 years, the entirety of the period covered by GDELT. Ideally, we would do this with some kind of index that accounted for things like the number of protest events that occurred, the number of people who participated in them, and the things those people did.

Unfortunately, the data set that includes all of that information for all relevant events around the world doesn’t exist and never will. Although it might feel like we now live in a Panopticon, we don’t. In reality, we can still only see things that get reported in sources to which we have access; those reports aren’t always “true,” sometimes conflict, and are always incomplete; and, even in 2014, it’s still hard to reliably locate, parse, and encode data from the stories that we do see.

GDELT is the most ambitious effort to date to overcome these problems, and that ambition is helping to pull empirical social science in some new and productive directions. GDELT uses software to scour the web for media stories that contain information about a large but predetermined array of verbal and physical interactions. These interactions range from protests, threats, and attacks to more positive things like requests for aid and expressions of support. When GDELT’s software finds text that describes one of those interactions, it creates a record that includes numeric representations of words or phrases indicating what kind of interaction it was, who was involved, and where and when it took place. Each of those records becomes one tiny layer in an ever-growing stack. GDELT was only created in the 2010s, but its software has been applied to archival material to extend its coverage all the way back to 1979. The current version includes roughly 2.5 million records, and that number now grows by tens of thousands every day.

GDELT grows out of a rich tradition of event data production in social science, and its coding process mimics many of the procedures that scholars have long used to try to catalog various events of interest—or, at least, to capture reasonably representative samples of them. As such, it’s tempting to treat GDELT’s records as markers of discrete events that can be counted and cross-tabulated to identify trends over time and other patterns of interest.

That temptation should be assiduously resisted for two reasons that Leetaru and others involved in GDELT’s original creation have frequently acknowledged. First, GDELT can only create records from stories that it sees, and the volume and nature of media coverage and its digitized renderings have changed radically over the past 30 years. This change continues and may still be accelerating. One result of this change is exponential growth over time in the volume of GDELT records, as shown in the chart below (borrowed from an informative post on the Ward Lab blog). Under these circumstances, it’s unclear what comparisons across years, and especially decades, are getting at. Are we seeing meaningful changes in the phenomenon of interest, or are we really just seeing traces of change in the volume and nature of reporting on them?

Change Over Time in the Volume of GDELT Records, 1979-2011 (Source: Ward Lab)

Second, GDELT has not fully worked out how to de-duplicate its records. When the same event is reported in more than one media source, GDELT can’t always tell that they are the same event, sometimes even when it’s the same story appearing verbatim in more than one outlet. As a result, events that attract more attention are likely to generate more records. Under these circumstances, the whole idea of treating counts of records in certain categories as counts of certain event types becomes deeply problematic.

Kalev knows these things and tries to address them in his recent FP post on trends over time in protest activity. Here is how he describes what he does and the graph that results:

The number of protests each month is divided by the total number of all events recorded in GDELT that month to create a “protest intensity” score that tracks just how prevalent worldwide protest activity has been month-by-month over the last quarter-century (this corrects for the exponential rise in media coverage over the last 30 years and the imperfect nature of computer processing of the news). To make it easier to spot the macro-level patterns, a black 12-month moving average trend line is drawn on top of the graph to help clarify the major temporal shifts.

Intensity of protest activity worldwide 1979-April 2014 (black line is 12-month moving average) (Source: Kalev Leetaru via FP)

Unfortunately, I don’t think Kalev’s normalization strategy addresses either of the aforementioned problems enough to make the kind of inferences he wants to make about trends over time in the intensity of protest activity around the world.

Let’s start at the top. The numerator of Kalev’s index is the monthly count of records in a particular set of categories. This is where the lack of de-duplication can really skew the picture, and the index Kalev uses does nothing to directly address it.

Without better de-duplication, we can’t fix this problem, but we might be less worried about it if we thought that duplication were a reliable marker of event intensity. Unfortunately, it almost certainly isn’t. Certain events catch the media’s eyes for all kinds of reasons. Some are related to the nature of the event itself, but many aren’t. The things that interest us change over time, as do the ways we talk about them and the motivations of the corporations and editors who partially mediate that conversation. Under these circumstances, it would strain credulity to assume that the frequency of reports on a particular event reliably represents the intensity, or even the salience, of that event. There are just too many other possible explanations to make that inferential leap.

And there’s trouble in the bottom, too. Kalev’s decision to use the monthly volume of all records in the denominator is a reasonable one, but it doesn’t fully solve the problem it’s meant to address, either.

What we get from this division is a proportion: protest-related records as a share of all records. The problem with comparing these proportions across time slices is that they can differ for more than one reason, and that’s true even if we (heroically) assume that the lack of de-duplication isn’t a concern. A change from one month to the next might result from a change in the frequency or intensity of protest activity, but it could also result from a change in the frequency or intensity of some other event type also being tallied. Say, for example, that a war breaks out and produces a big spike in GDELT records related to violent conflict. Under these circumstances, the number of protest-related records could stay the same or even increase, and we would still see a drop in the “protest intensity score” Kalev uses.

In the end, what we get from Kalev’s index isn’t a reliable measure of the intensity of protest activity around the world and its change over time. What we get instead is a noisy measure of relative media attention to protest activity over a period of time when the nature of media attention itself has changed a great deal in ways that we still don’t fully understand. That quantity is potentially interesting in its own right. Frustratingly, though, it cannot answer seemingly simple questions like “How much protest activity are we seeing now?” or “How has the frequency or intensity of protest activity changed over the past 30 years?”

I’ll wrap this up by saying that I am still really, really excited about the new possibilities for social scientific research opening up as a result of projects like GDELT and, now, the Open Event Data Alliance it helped to spawn. At the same time, I think we social scientists have to be very cautious in our use of these shiny new things. As excited as we may be, we’re also the ones with the professional obligation to check the impulse to push them harder than they’re ready to go.

1 Comment
by Jay Ulfelder on June 6, 2014  •  Permalink
Posted in Measurement, Protests and Popular Uprisings
Tagged , , ,

Posted by Jay Ulfelder on June 6, 2014

https://dartthrowingchimp.wordpress.com/2014/06/06/another-note-on-the-limitations-of-event-data/

A Coda to “Using GDELT to Monitor Atrocities, Take 2”

I love doing research in the Internet Age. As I’d hoped it would, my post yesterday on the latest iteration of our atrocities-monitoring system in the works has already sparked a lot of really helpful responses. Some of those responses are captured in comments on the post, but not all of them are. So, partly as a public good and partly for my own record-keeping, I thought I’d write a coda to that post enumerating the leads it generated and some of my reactions to them.

Give the Machines Another Shot at It

As a way to reduce or even eliminate the burden placed on our human(s) in the loop, several people suggested something we’ve been considering for a while: use machine-learning techniques to develop classifiers that can be used to further reduce the data left after our first round of filtering. These classifiers could consider all of the features in GDELT, not just the event and actor types we’re using in our R script now. If we’re feeling really ambitious, we could go all the way back to the source stories and use natural-language processing to look for additional discriminatory power there. This second round might not eliminate the need for human review, but it certainly could lighten the load.

The comment threads on this topic (here and here) nicely capture what I see as the promise and likely limitations of this strategy, so I won’t belabor it here. For now, I’ll just note that how well this would work is an empirical question, and it’s one we hope to get a chance to answer once we’ve accumulated enough screened data to give those classifiers a fighting chance.

Leverage GDELT’s Global Knowledge Graph

Related to the first idea, GDELT co-creator Kalev Leetaru has suggested on a couple of occasions that we think about ways to bring the recently-created GDELT Global Knowledge Graph (GKG) to bear on our filtering task. As Kalev describes in a post on the GDELT blog, GKG consists of two data streams, one that records mentions of various counts and another that captures connections  in each day’s news between “persons, organizations, locations, emotions, themes, counts, events, and sources.” That second stream in particular includes a bunch of data points that we can connect to specific event records and thus use as additional features in the kind of classifiers described under the previous header. In response to my post, Kalev sent this email to me and a few colleagues:

I ran some very very quick numbers on the human coding results Jay sent me where a human coded 922 articles covering 9 days of GDELT events and coded 26 of them as atrocities. Of course, 26 records isn’t enough to get any kind of statistical latch onto to build a training model, but the spectral response of the various GKG themes is quite informative. For events tagged as being an atrocity, themes such as ETHNICITY, RELIGION, HUMAN_RIGHTS, and a variety of functional actors like Villagers, Doctors, Prophets, Activists, show up in the top themes, whereas in the non-atrocities the roles are primarily political leaders, military personnel, authorities, etc. As just a simple example, the HUMAN_RIGHTS theme appeared in just 6% of non-atrocities, but 30% of atrocities, while Activists show up in 33% of atrocities compared with just 4% of non-atrocities, and the list goes on.

Again, 26 articles isn’t enough to build a model on, but just glancing over the breakdown of the GKG themes for the two there is a really strong and clear breakage between the two across the entire set of themes, and the breakdown fits precisely what baysean classifiers like (they are the most accurate for this kind of separation task and outperform SVM and random forest).

So, Jay, the bottom line is that if you can start recording each day the list of articles that you guys review and the ones you flag as an atrocity and give me a nice dataset over time, should be pretty easy to dramatically filter these down for you at the very least.

As I’ve said throughout this process, its not that event data can’t do what is needed, its that often you have to bring additional signals into the mix to accomplish your goals when the thing you’re after requires signals beyond what the event records are capturing.

What Kalev suggests at the end there—keep a record of all the events we review and the decisions we make on them—is what we’re doing now, and I hope we can expand on his experiment in the next several months.

Crowdsource It

Jim Walsh left a thoughtful comment suggesting that we crowdsource the human coding:

Seems to me like a lot of people might be willing to volunteer their time for this important issue–human rights activists and NGO types, area experts, professors and their students (who might even get some credit and learn about coding). If you had a large enough cadre of volunteers, could assign many (10 or more?) to each day’s data and generate some sort of average or modal response. Would need someone to organize the volunteers, and I’m not sure how this would be implemented online, but might be do-able.

As I said in my reply to him, this is an approach we’ve considered but rejected for now. We’re eager to take advantage of the wisdom of interested crowds and are already doing so in big ways on other parts of our early-warning system, but I have two major concerns about how well it would work for this particular task.

The first is the recruiting problem, and here I see a Catch-22: people are less inclined to do this if they don’t believe the system works, but it’s hard to convince them that the system works if we don’t already have a crowd involved to make it go. This recruiting problem becomes especially acute in a system with time-sensitive deliverables. If we promise daily updates, we need to produce daily updates, and it’s hard to do that reliably if we depend on self-organized labor.

My second concern is the principal-agent problem. Our goal is to make reliable and valid data in a timely way, but there are surely people out there who would bring goals to the process that might not align with ours. Imagine, for example, that Absurdistan appears in the filtered-but-not-yet-coded data to be committing atrocities, but citizens (or even paid agents) of Absurdistan don’t like that idea and so organize to vote those events out of the data set. It’s possible that our project would be too far under the radar for anyone to bother, but our ambitions are larger than that, so we don’t want to assume that will be true. If we succeed at attracting the kind of attention we hope to attract, the deeply political and often controversial nature of our subject matter would make crowdsourcing this task more vulnerable to this kind of failure.

Use Mechanical Turk

Both of the concerns I have about the downsides of crowdsourcing the human-coding stage could be addressed by Ryan Briggs’ suggestion via Twitter to have Amazon Mechanical Turk do it. A hired crowd is there when you need it and (usually) doesn’t bring political agendas to the task. It’s also relatively cheap, and you only pay for work performed.

Thanks to our collaboration with Dartmouth’s Dickey Center, the marginal cost of the human coding isn’t huge, so it’s not clear that Mechanical Turk would offer much advantage on that front. Where it could really help is in routinizing the daily updates. As I mentioned in the initial post, when you depend on human action and have just one or a few people involved, it’s hard to establish a set of routines that covers weekends and college breaks and sick days and is robust to periodic changes in personnel. Primarily for this reason, I hope we’ll be able to run an experiment with Mechanical Turk where we can compare its cost and output to what we’re paying and getting now and see if this strategy might make sense for us.

Don’t Forget About Errors of Omission

Last but not least, a longtime colleague had this to say in an email reacting to the post (hyperlinks added):

You are effectively describing a method for reducing errors of commission, events coded by GDELT as atrocities that, upon closer inspection, should not be. It seems like you also need to examine errors of omission. This is obviously harder. Two possible opportunities would be to compare to either [the PITF Worldwide Atrocities Event Data Set] or to ACLED.  There are two questions. Is GDELT “seeing” the same source info (and my guess is that it is and more, though ACLED covers more than just English sources and I’m not sure where GDELT stands on other languages). Then if so (and there are errors of omission) why aren’t they showing up (coded as different types of events or failed to trigger any coding at all)[?]

It’s true that our efforts so far have focused almost exclusively on avoiding errors of commission, with the important caveat that it’s really our automated filtering process, not GDELT, that commits most of these errors. The basic problem for us is that GDELT, or really the CAMEO scheme on which it’s based, wasn’t designed to spot atrocities per se. As a result, most of what we filter out in our human-coding second stage aren’t things that were miscoded by GDELT. Instead, they’re things that were properly coded by GDELT as various forms of violent action but upon closer inspection don’t appear to involve the additional features of atrocities as we define them.

Of course, that still leaves us with this colleague’s central concern about errors of omission, and on that he’s absolutely right. I have experimented with different actor and event-type criteria to make sure we’re not missing a lot of events of interest in GDELT, but I haven’t yet compared what we’re finding in GDELT to what related databases that use different sources are seeing. Once we accumulate a few month’s worth of data, I think this is something we’re really going to need to do.

Stay tuned for Take 3…

8 Comments
by Jay Ulfelder on January 15, 2014  •  Permalink
Posted in Mass Atrocities, Measurement, Methods, Violent Conflict
Tagged , , , , ,

Posted by Jay Ulfelder on January 15, 2014

https://dartthrowingchimp.wordpress.com/2014/01/15/a-coda-to-using-gdelt-to-monitor-atrocities-take-2/

Using GDELT to Monitor Atrocities, Take 2

Last May, I wrote a post about my preliminary efforts to use a new data set called GDELT to monitor reporting on atrocities around the world in near-real time. Those efforts represent one part of the work I’m doing on a public early-warning system for the U.S. Holocaust Memorial Museum’s Center for the Prevention of Genocide, and they have continued in fits and starts over the ensuing eight months. With help from Dartmouth’s Dickey Center, Palantir, and the GDELT crew, we’ve made a lot of progress. I thought I’d post an update now because I’m excited about the headway we’ve made; I think others might benefit from seeing what we’re doing; and I hope this transparency can help us figure out how to do this task even better.

So, let’s cut to the chase: Here is a screenshot of an interactive map locating the nine events captured in GDELT in the first week of January 2014 that looked like atrocities to us and occurred in a place that the Google Maps API recognized when queried. (One event was left off the map because Google Maps didn’t recognize its reported location.) The size of the bubbles corresponds to the number of civilian deaths, which in this map range from one to 31. To really get a feel for what we’re trying to do, though, head over to the original visualization on CartoDB (here), where you can zoom in and out and click on the bubbles to see a hyperlink to the story from which each event was identified.

atrocities.monitoring.screenshot.20140113

Looks simple, right? Well, it turns out it isn’t, not by a long shot.

As this blog’s regular readers know, GDELT uses software to scour the web for new stories about political interactions all around the world and parses those stories to identify and record information about who did or said what to whom, when, and where. It currently covers the period 1979-present and is now updated every day, and each of those daily updates contains some 100,000-140,000 new records. Miraculously and crucial to a non-profit pilot project like ours, GDELT is also available for free. 

The nine events plotted in the map above were sifted from the tens of thousands of records GDELT dumped on us in the first week of 2014. Unfortunately, that data-reduction process is only partially automated.

The first step in that process is the quickest. As originally envisioned back in May, we are using an R script (here) to download GDELT’s daily update file and sift it for events that look, from the event type and actors involved, like they might involve what we consider to be an atrocity—that is, deliberate, deadly violence against one or more noncombatant civilians in the context of a wider political conflict.

Unfortunately, the stack of records that filtering script returns—something like 100-200 records per day—still includes a lot of stuff that doesn’t interest us. Some records are properly coded but involve actions that don’t meet our definition of an atrocity (e.g., clashes between rioters and police or rebels and troops); some involve atrocities but are duplicates of events we’ve already captured; and some are just miscoded (e.g., a mention of the film industry “shooting” movies that gets coded as soldiers shooting civilians).

After we saw how noisy our data set would be if we stopped screening there, we experimented with a monitoring system that would acknowledge GDELT’s imperfections and try to work with them. As Phil Schrodt recommended at the recent GDELT DC Hackathon, we looked to “embrace the suck.” Instead of trying to use GDELT to generate a reliable chronicle of atrocities around the world, we would watch for interesting and potentially relevant perturbations in the information stream, noise and all, and those perturbations would produce alerts that users of our system could choose to investigate further. Working with Palantir, we built a system that would estimate country-specific prior moving averages of daily event counts returned by our filtering script and would generate an alert whenever a country’s new daily count landed more than two standard deviations above or below that average.

That system sounded great to most of the data pros in our figurative room, but it turned out to be a non-starter with some other constituencies of importance to us. The issue was credibility. Some of the events causing those perturbations in the GDELT stream were exactly what we were looking for, but others—a pod of beached whales in Brazil, or Congress killing a bill on healthcare reform—were laughably far from the mark. If our supposedly high-tech system confused beached whales and Congressional procedures for mass atrocities, we would risk undercutting the reputation for reliability and technical acumen that we are striving to achieve.

So, back to the drawing board we went. To separate the signal from the static and arrive at something more like that valid chronicle we’d originally envisioned, we decided that we needed to add a second, more laborious step to our data-reduction process. After our R script had done its work, we would review each of the remaining records by hand to decide if it belonged in our data set or not and, when necessary, to correct any fields that appeared to have been miscoded. While we were at it, we would also record the number of deaths each event produced. We wrote a set of rules to guide those decisions; had two people (a Dartmouth undergraduate research assistant and I) apply those rules to the same sets of daily files; and compared notes and made fixes. After a few iterations of that process over a few months, we arrived at the codebook we’re using now (here).

This process radically reduces the amount of data involved. Each of those two steps drops us down multiple orders of magnitude: from 100,000-140,000 records in the daily updates, to about 150 in our auto-filtered set, to just one or two in our hand-filtered set. The figure below illustrates the extent of that reduction. In effect, we’re treating GDELT as a very powerful but error-prone search and coding tool, a source of raw ore that needs refining to become the thing we’re after. This isn’t the only way to use GDELT, of course, but for our monitoring task as presently conceived, it’s the one that we think will work best.

monitoring.data.reduction.graphic

Once that second data-reduction step is done, we still have a few tasks left to enable the kind of mapping and analysis we aim to do. We want to trim the data set to keep only the atrocities we’ve identified, and we need to consolidate the original and corrected fields in those remaining records and geolocate them. All of that work gets done with a second R script (here), which is applied to the spreadsheet the coder saves after completing her work. The much smaller file that script produces is then ready to upload to a repository where it can be combined with other days’ outputs to produce the global chronicle our monitoring project aims to produce.

From start to finish, each daily update now takes about 45 minutes, give or take 15. We’d like to shrink that further if we can but don’t see any real opportunities to do so at the moment. Perhaps more important, we still have to figure out the bureaucratic procedures that will allow us to squeeze daily updates from a “human in the loop” process in a world where there are weekends and holidays and people get sick and take vacations and sometimes even quit. Finally, we also have not yet built the dashboard that will display and summarize and provide access to these data on our program’s web site, which we expect to launch some time this spring.

We know that the data set this process produces will be incomplete. I am 100-percent certain that during the first week of January 2014, more than 10 events occurred around the world that met our definition of an atrocity. Unfortunately, we can only find things where GDELT looks, and even a scan of every news story produced every day everywhere in the world would fail to see the many atrocities that never make the news.

On the whole, though, I’m excited about the progress we’ve made. As soon as we can launch it, this monitoring process should help advocates and analysts more efficiently track atrocities globally in close to real time. As our data set grows, we also hope it will serve as the foundation for new research on forecasting, explaining, and preventing this kind of violence. Even with its evident shortcomings, we believe this data set will prove to be useful, and as GDELT’s reach continues to expand, so will ours.

PS For a coda discussing the great ideas people had in response to this post, go here.

[Erratum: The original version of this post said there were about 10,000 records in each daily update from GDELT. The actual figure is 100,000-140,000. The error has been corrected and the illustration of data reduction updated accordingly.]

22 Comments
by Jay Ulfelder on January 14, 2014  •  Permalink
Posted in Mass Atrocities, Methods, Violent Conflict
Tagged , , , , ,

Posted by Jay Ulfelder on January 14, 2014

https://dartthrowingchimp.wordpress.com/2014/01/14/finding-atrocious-needles-in-the-gdelt-haystack/

A Research Note on Updating Coup Forecasts

A new year is about to start, and that means it’s time for me to update my coup forecasts (see here and here for the 2013 and 2012 editions, respectively). The forecasts themselves aren’t quite ready yet—I need to wait until mid-January for updates from Freedom House to arrive—but I am making some changes to my forecasting process that I thought I would go ahead and describe now, because the thinking behind them illustrates some important dilemmas and new opportunities for predictions of many kinds of political events.

When it comes time to build a predictive statistical model of some rare political event, it’s usually not the model specification that gives me headaches. For many events of interest, I think we now have a pretty good understanding of which methods and variables are likely to produce more accurate forecasts.

Instead, it’s the data, or really the lack thereof, that sets me to pulling my hair out. As I discussed in a recent post, things we’d like to include in our models fall into a few general classes in this regard:

  • No data exist (fuggeddaboudit)
  • Data exist for some historical period, but they aren’t updated (“HA-ha!”)
  • Data exist and are updated, but they are patchy and not missing at random (so long, some countries)
  • Data exist and are updated, but not until many months or even years later (Spinning Pinwheel of Death)

In the past, I’ve set aside measures that fall into the first three of those sets but gone ahead and used some from the fourth, if I thought the feature was important enough. To generate forecasts before the original sources updated, I either a) pulled forward the last observed value for each case (if the measure was slow-changing, like a country’s infant mortality rate) or b) hand-coded my own updates (if the measure was liable to change from year to year, like a country’s political regime type).

Now, though, I’ve decided to get out of the “artisanal updating” business, too, for all but the most obvious and uncontroversial things, like which countries recently joined the WTO or held national elections. I’m quitting this business, in part, because it takes a lot of time and the results may be pretty noisy. More important, though, I’m also quitting because it’s not so necessary any more, thanks to  timelier updates from some data providers and the arrival of some valuable new data sets.

This commitment to more efficient updating has led me to adopt the following rules of thumb for my 2014 forecasting work:

  • For structural features that don’t change much from year to year (e.g., population size or infant mortality), include the feature and use the last observed value.
  • For variables that can change from year to year in hard-to-predict ways, only include them if the data source is updated in near-real time or, if it’s updated annually, if those updates are delivered within the first few weeks of the new year.
  • In all cases, only use data that are publicly available, to facilitate replication and to encourage more data sharing.

And here are some of the results of applying those rules of thumb to the list of features I’d like to include in my coup forecasting models for 2014.

  • Use Powell and Thyne’s list of coup events instead of Monty Marshall’s. Powell and Thyne’s list is updated throughout the year as events occur, whereas the publicly available version of Marshall’s list is only updated annually, several months after the start of the year. That wouldn’t matter so much if coups were only the dependent variable, but recent coup activity is also an important predictor, so I need the last year’s updates ASAP.
  • Use Freedom House’s Freedom in the World (FIW) data instead of Polity IV to measure countries’ political regime type. Polity IV offers more granular measures of political regime type than Freedom in the World, but Polity updates aren’t posted until spring or summer of the following year, usually more than a third of the way into my annual forecasting window.
  •  Use IMF data on economic growth instead of the World Bank’s. The Bank now updates its World Development Indicators a couple of times a year, and there’s a great R package that makes it easy to download the bits you need. That’s wonderful for slow-changing structural features, but it still doesn’t get me data on economic performance as fast as I’d like it. I work around that problem by using the IMF’s World Economic Outlook Database, which include projections for years for which observed data aren’t yet available and forecasts for several years into the future.
  • Last but not least, use GDELT instead of UCDP/PRIO or Major Episodes of Political Violence (MEPV) to measure civil conflict. Knowing which countries have had civil unrest or violence in the recent past can help us predict coup attempts, but the major publicly available measures of these things are only updated well into the year. GDELT now represents a nice alternative. It covers the whole world, measures lots of different forms of political cooperation and conflict, and is updated daily, so country-year updates are available on January 2. GDELT’s period of observation starts in 1979, so it’s still a stretch to use it models of super-rare events like mass-killing onsets, where the number of available examples since 1979 on which to train is still relatively small. For less-rare events like coup attempts, though, starting the analysis around 1980 is no problem. (Just don’t forget to normalize them!) With some help from John Beieler, I’m already experimenting with adding annual GDELT summaries to my coup forecasting process, and I’m finding that they do improve the model’s out-of-sample predictive power.

In all of the forecasting work I do, my long-term goals are 1) to make the forecasts more dynamic by updating them more frequently (e.g., monthly, weekly, or even daily instead of yearly) and 2) to automate that updating process as much as possible. The changes I’m making to my coup forecasting process for 2014 don’t directly accomplish either of these things, but they do take me a few steps in both directions. For example, once GDELT is in the mix, it’s possible to start thinking about how to switch to monthly or even daily updates that rely on a sliding window of recent GDELT tallies. And once I’ve got a coup data set that updates in near-real time, I can imagine pinging that source each day to update the counts of coup attempts in the past several years. I’m still not where I’d like to be, but I think I’m finally stepping onto a path that can carry me there.

3 Comments
by Jay Ulfelder on December 29, 2013  •  Permalink
Posted in Coups, Forecasting, Methods, Statistics
Tagged , , , , , ,

Posted by Jay Ulfelder on December 29, 2013

https://dartthrowingchimp.wordpress.com/2013/12/29/a-research-note-on-updating-coup-forecasts/

Playing Telephone with Data Science

You know the telephone game, where a bunch of people sit in a circle or around a table and pass a whispered sentence from person to person until it comes back to the one who started it and they say the version they heard out loud and you all crack up at how garbled it got?

Well, I wonder if John Beieler is cracking up or crying right now, because the same thing is happening with a visualization he created using data from the recently released Global Dataset on Events, Language, and Tone, a.k.a. GDELT.

Back at the end of July, John posted a terrific animated set of maps of protest activity worldwide since 1979. In a previous post on a single slice of the data used in that animation, John was careful to attach a number of caveats to the work: the maps only include events covered in the sources GDELT scours, GDELT sometimes codes events that didn’t happen, GDELT sometimes struggles to put events in their proper geographic location, event labels in the CAMEO event classification scheme GDELT uses doesn’t always mean what you think they mean, counts of events don’t tell you anything about the size or duration of the events being counted, etc., etc.  In the blogged cover letter for the animated series, John added one more very important caveat about the apparent increase in the incidence of protest activity over time:

When dealing with the time-series of data, however, one additional, and very important, point also applies. The number of events recorded in GDELT grows exponentially over time, as noted in the paper introducing the dataset. This means that over time there appears to be a steady increase in events, but this should not be mistaken as a rise in the actual amount of behavior X (protest behavior in this case). Instead, due to changes in reporting and the digital recording of news stories, it is simply the case that there are more events of every type over time. In some preliminary work that is not yet publicly released, protest behavior seems to remain relatively constant over time as a percentage of the total number of events. This means that while there was an explosion of protest activity in the Middle East, and elsewhere, during the past few years, identifying visible patterns is a tricky endeavor due to the nature of the underlying data.

John’s post deservedly caught the eye of J. Dana Stuster, an assistant editor at Foreign Policy, who wrote a bit about it last week. Stuster’s piece was careful to repeat many of John’s caveats, but the headline—“Mapped: Every Protest on the Planet since 1979”—got sloppy, essentially shedding several of the most important qualifiers. As John had taken pains to note, what we see in the maps is not all that there is, and some of what’s shown in the maps didn’t really happen.

Well, you can probably where this is going.  Not long after that Foreign Policy piece appeared, I saw this tweet from Chelsea Clinton:

In fewer than 140 characters, Clinton impressively managed to put back the caveat Foreign Policy had dropped in its headline about press coverage vs. reality, but the message had already been garbled, and now it was going viral. Fast forward to this past weekend, when the phrase “Watch a Jaw-dropping Visualization of Every Protest since 1979” made repeated appearances in my Twitter timeline. This next iteration came from Ultraculture blogger Jason Louv, and it included this bit:

Also fruitful: Comparing this data with media coverage and treatment of protest. Why is it easy to think of the 1960s and 70s as a time of dissent and our time as a more ordered, controlled and conformist period when the data so clearly shows that there is no comparison in how much protest there is now compared to then? Media distortion much?

So now we get a version that ignores both the caveat about GDELT’s coverage not being exhaustive or perfect and the related one about the apparent increase in protest volume over time being at least in part an artifact of “changes in reporting and the digital recording of news stories.” What started out as a simple proof-of-concept exercise—“The areas that are ‘bright’ are those that would generally be expected to be so,” John wrote in his initial post—had been twisted into a definitive visual record of protest activity around the world in the past 35 years.

As someone who thinks that GDELT is an analytical gusher and believes that it’s useful and important to make work like this accessible to broader audiences, I don’t know what to learn from this example. John was as careful as could be, but the work still mutated as it spread. How do you prevent this from happening, or at least mitigate the damage when it does?

If anyone’s got some ideas, I’d love to hear them.

12 Comments
by Jay Ulfelder on August 26, 2013  •  Permalink
Posted in Methods
Tagged , , , , , , ,

Posted by Jay Ulfelder on August 26, 2013

https://dartthrowingchimp.wordpress.com/2013/08/26/playing-telephone-with-data-science/

Forecasting Round-Up No. 4

Another in an occasional series of posts calling out interesting work on forecasting. See here, here, and here for earlier ones.

1. A gaggle of researchers at Penn State, including Phil Schrodt, have posted a new conference paper (PDF) showing how they are using computer-generated data on political interactions around the world (the oft-mentioned GDELT) to forecast various forms of political crisis with respectable accuracy.

One important finding from their research so far: models that mix dynamic data on political interactions with slow-changing data on relevant structural conditions (political, social, economic) produce more accurate forecasts than models that use only one or the other. That’s not surprising, but it is a useful confirmation nonetheless. Thanks to GDELT’s public release, I predict that we’ll see a lot more social-science modelers doing that kind of mixing in the near future.

2. Kaiser Fung reviews Predictive Analytics, a book by Eric Siegel. I haven’t read it, but Kaiser’s review makes me think it would be a good addition to my short list of recommended readings for forecasters.

3. Finally, the 2013 edition of the Failed States Index (FSI) is now up on Foreign Policy‘s web site (here). I call it out here to air a few grievances.

First, it makes me a little crazy that it’s hard to pin down exactly what this index is supposed to do. Is FSI meant to summarize recent conditions or to help forecast new troubles down the road? In their explication of the methodology behind it, the makers of the FSI acknowledge that it’s the largely former but also slide into describing it as an early-warning tool. And what exactly is “state failure,” anyway? They never quite say, which makes it hard to use the index as either a snapshot or a forecast.

Second, as I’ve said before on this blog, I’m also not a big fan of indices that roll up so many different things into a single value on the basis of assumptions alone. Statistical models also combine a lot of information, but they do so with weights that are derived from a systematic exploration of empirical evidence. FSI simply assumes all of its 12 components are equally relevant when there’s ample opportunity to check that assumption against the historical record. Maybe some of the index’s components are more informative than others, so why not use models to try to find out?

Last but not least, on the way FSI is presented, I think the angry reactions it elicits (see comments on previous editions or my Twitter feed whenever FSI is released) are a useful reminder of the risks of presenting rank-ordered lists based on minor variations in imprecise numbers. People spend a lot of time venting about relatively small differences between states (e.g., “Why is Ethiopia two notches higher than Syria?”) when those aren’t very informative, and aren’t really meant to be. I’ve run into the same problem when I’ve posted statistical forecasts of things like coup attempts and nonviolent uprisings, and I’m increasingly convinced that those rank-ordered lists are a distraction. To use the results without fetishizing the numbers, we might do better to focus on the counter-intuitive results (surprises) and on cases whose scores change a lot across iterations.

5 Comments
by Jay Ulfelder on June 24, 2013  •  Permalink
Posted in Forecasting
Tagged , , , ,

Posted by Jay Ulfelder on June 24, 2013

https://dartthrowingchimp.wordpress.com/2013/06/24/forecasting-round-up-no-4/

Challenges in Measuring Violent Conflict, Syria Edition

As part of a larger (but, unfortunately, gated) story on how the terrific new Global Data on Events, Language, and Tone (GDELT) might help social scientists forecast violent conflicts, the New Scientist recently posted some graphics using GDELT to chart the ongoing civil war in Syria. Among those graphics was this time-series plot of violent events per day in Syria since the start of 2011:

Syrian Conflict New Scientist

Based on that chart, the author of the story (not the producers of GDELT, mind you) wrote:

As Western leaders ponder intervention, the resulting view suggests that the violence has subsided in recent months, from a peak in the third quarter of 2012.

That inference is almost certainly wrong, and why it’s wrong underscores one of the fundamental challenges in using event data—whether it’s collected and coded by software or humans or some combination thereof—to observe the dynamics of violent conflict.

I say that inference is almost certainly wrong because concurrent data on deaths and refugees suggest that violence in Syria has only intensified in past year. One of the most reputable sources on deaths from the war is the Syria Tracker. A screenshot of their chart of monthly counts of documented killings is shown below. Like GDELT, their data also identify a sharp increase in violence in late 2012. Unlike GDELT, their data indicate that the intensity of the violence has remained very high since then, and that’s true even though the process of documenting killings inevitably lags behind the actual violence.

Syria Tracker monthly death counts

We see a similar pattern in data from the U.N. High Commissioner on Refugees (UNHCR) on people fleeing the fighting in Syria. If anything, the flow of refugees has only increased in 2013, suggesting that the violence in Syria is hardly abating.

UNHCR syria refugee plot

The reason GDELT’s count of violent events has diverged from other measures of the intensity of the violence in Syria in recent months is probably something called “media fatigue.” Data sets of political events generally depend on news sources to spot events of interest, and it turns out that news coverage of large-scale political violence follows a predictable arc. As Deborah Gerner and Phil Schrodt describe in a paper from the late 1990s, press coverage of a sustained and intense conflicts is often high when hostilities first break out but then declines steadily thereafter. That decline can happen because editors and readers get bored, burned out, or distracted. It can also happen because the conflict gets so intense that it becomes, in a sense, too dangerous to cover. In the case of Syria, I suspect all of these things are at work.

My point here isn’t to knock GDELT, which is still recording scores or hundreds of events in Syria every day, automatically, using open-source code, and then distributing those data to the public for free. Instead, I’m just trying to remind would-be users of any data set of political events to infer with caution. Event counts are one useful way to track variation over time in political processes we care about, but they’re only one part of the proverbial elephant, and they are inevitably constrained by the limitations of the sources from which they draw. To get a fuller sense of the beast, we need as often as possible to cross-reference those event data with other sources of information. Each of the sources I’ve cited here has its own blind spots and selection biases, but a comparison of trends from all three—and, importantly, an awareness of the likely sources of those biases—is enough to give me confidence that the civil war in Syria is only continuing to intensify. That says something important about Syria, of course, but it also says something important about the risks of drawing conclusions from event counts alone.

PS. For a great discussion of other sources of bias in the study of political violence, see Stathis Kalyvas’ 2004 essay on “The Urban Bias in Research on Civil Wars” (PDF).

17 Comments
by Jay Ulfelder on May 16, 2013  •  Permalink
Posted in Mass Atrocities, Methods, Uncategorized, Violent Conflict
Tagged , , , , ,

Posted by Jay Ulfelder on May 16, 2013

https://dartthrowingchimp.wordpress.com/2013/05/16/challenges-in-measuring-violent-conflict-syria-edition/

« Older Posts

  • Author

  • Follow me on Twitter

  • Follow Dart-Throwing Chimp on WordPress.com

  • Enter your email address to follow this blog and receive notifications of new posts by email.

    Join 13,609 other subscribers

  • Archives

%d bloggers like this: