January 31, 2014

The New Digital Divide: Transforming the Global South into Reliable Data

Transforming the world's most hard-to-access and uncertain landscapes into digital data. Sutika Sipus 2014.
Everyday urban professionals, data scientists, economists, and geographers sit in front of a computer screen and create extraordinary visualizations and statistical methods to unravel the world.  Geographic information systems such as QGIS, statistical programs like "R", spreadsheet softwares like Excel and lines of python code have empowered us with the ability to understand economies at scale, measure and predict public health, monitor pollution and deter violence.  Data is good.

Yet what about cities, states, and nations that do not or cannot generate reliable data?  In his recent book, Poor Numbers, author Morten Jerven reveals the faulty statistics collected and published by government agencies throughout Africa.  Over the last three years that I was in Afghanistan, I witnessed nearly every single aid agency or government research contractor rely upon "perception based" data which means researchers confronted too much danger in the field to collect actual information, but could only ask locals their opinion on matters ranging from conflict to education and corruption.  This method is safe but provides zero validity.  It might as well be make-believe.

The result is the global data gap.  Governments and institutions that can transform intangible social dynamics into quantifiable data can conduct sophisticated analysis and move forward at a faster pace. This sensibility was the foundation of my initiative in Mogadishu, to create a comprehensive map of the city that fused business and residential management with geography.  As my operation was too small to go beyond the proof of concept, the vision was eventually passed via the local government and integrated into a longstanding UN initiative to develop a city planning department which is advancing with some success.  Yet while Mogadishu may be on the cusp of a digital governance revolution, problems persist.  Data dies.  Situations change.  More dramatically, very little of the world is generating the data sets commonly enjoyed throughout the west.

The global data gap is economically inhibitive. Imagine if your company sought a new market opportunity because the markets your normally serve are saturated with your product and your competitors.  Most companies would never imagine distribution in an African nation, partly because of misled beliefs on stability of those markets, but that those misconceptions are ultimately founded on a lack of reliable data.  With no local data, there is no global opportunity.

This is also a failure for companies that already working in data-deficient nations.  A few months ago I had a meeting with Afghanistan's largest tele-communications provider, Roshan, and when I asked about coverage, they could only give vague feedback.  When I asked for data on every household using Roshan to access the internet in Kabul, they could not give this information because Kabul doesn't have a postal address system, so all installations are tied to a person's name and neighborhood, but not a specific address.  In this instance I created alternative solution, where after about three weeks of combing selected neighborhoods, I was able to generate a GPS location for every Wi-Fi network and mobile tower in each area which could then be joined with the existing data.  We could filter Roshan networks vs competitor networks and now had sufficient data to improve marketing and coverage strategies.

Location and evaluation of strength of Wi-Fi access in Kabul, Afghanistan. Sutika Sipus 2014.
Having worked throughout Africa and Asia as a researcher since 2007,  I have developed an array of techniques to get past this problem, focusing on the creation and testing of indirect indicators.  In Zimbabwe economic wealth could be measured by counting the number of water jugs in front of each house.  In the Philipines, one could count denim jeans swinging on the clothesline of an apartment.  In a variety of Somali refugee camps I found that metal roofing materials separated the less-poor from the more-poor.  In Afghanistan I have steadily been testing and re-testing the presence of graffiti as a predictor of social protest and conflict with success.  The advantage of these Rapid Rural Appraisal techniques is that they are safe, fast, efficient, and quantifiable.  To determine an RRA indicator requires extensive time on the ground, but once established, we can effectively measure anything, anywhere.  There are of course other methods, standard survey techniques, but my efforts generate GPS location, culturally relative valuation, and easily shared outcomes.  RRA is not new, but my method of fusing RRA with traditional research methods, GIS tools, and mobile technologies does create a new outcome. I produce valid, quantifiable and mappable data that is customized to the problem and the location, but can accommodate different scales.

Digital Data Collection and Mapping.
Cambodia. Sutika Sipus 2014.
To me, the global data gap is a new frontier of untapped opportunity.  Maybe more people will realize this sooner than later and I'll encounter some digital cowboys, wandering deserts with laptops and satellite phones, their backpacks sagging beneath the weight of external hard drives.  I won't be the only one canvasing the worlds most remote locations.

Maybe soon more companies will ask "what about Nigeria?  what about Ghana or Bangladesh?" and they will need answers.  They will look online and see some global statistics that are 5 years old and impossible to trust.  They will need a fresh perspective they can trust and they can see.  Something they can drop into their software and understand.  Good thing I'm easy to find.

January 20, 2014

The GIS Interface I Always Wanted


I have a love-hate relationship with geographic information systems.  I use these systems frequently and have been thrilled to witness a recent explosion of interest in mapping across various disciplines.  But overall, I have always hated the software.

Part of this is the fault of GIS software UI/UX design.  Current GIS software is designed for a user to work with a database or data tables, and by generating a final set of data, the software will produce a visual product - the map.  Many GIS packages do providing CAD-like drawing or design tools, however application of these tools does not modify existing data.  They may allow one to create a new and empty shape file layer, but it is necessary to then attach data to that shape file.  If I want to make changes to a cartographic layer that already contains data, then I have limited options from the graphics side, and more options from the data side.

The outcome is that seemingly simple goals in GIS contain many steps to achieve and are not intuitive to a new user.   I recall when I was a new student to GIS that the strategy I created to learn the software was "right click at all times,"  otherwise it was impossible to know how to proceed.  Of course anyone who is a frequent GIS user finds all these smaller steps as intuitive because they have used it for a long time.  Consequently many GIS professionals have adapted to the logic of the system and have difficulty to see its flaws.

The other problem with current GIS platforms is the way they are taught.  When learning GIS, students are  introduced to the vocabulary of geography and cartography as the theory component, much of which focusses on the the map, not the data.   In the meanwhile they must undertake assignments and exercises focussed on the software.  The theory and the application remain separate most of the time.  A better GIS class would start with explaining and exploring databases without maps.  Once the student is comfortable with a database, geographic coordinates would be introduced (all using a common projection etc), maps would be generated, and finally the class would begin to explore geographic concepts.  In this manner the teaching is attuned to the system and moves forward in a simple and linear fashion.  

Over the last few years GIS has exploded with options.  When I was first introduced to these tools in 2006, the only options were ESRI products and GRASS as the open source alternative. Today we now have tools such as QGIS, TileMill, Open Street Map, and maps are frequently created tiles for a base map.  Yet I believe there is still plenty of room to reinvent the GIS interface.  I believe we can create a better GIS interface and user experience design to expand versatility.  

An ideal software package would provide each user with 2 options for user interaction.  One option is the same as the current approach, in which a technician works with the data to generate a visual outcome.  The alternative is software that allows one to create a visual modification which will also shift the underlying data accordingly.  This is a step beyond existing products, such as TileMill, which provide the means for designing beautiful maps, but do not allow one to conduct data analysis.

In the above image I rendered an ideal GIS interface, based strongly from Adobe photoshop, on account that photoshop is highly intuitive many users.  This interface features layers in the right corner like photo layers, but the most significant feature is the ability to work across Tables - Drawing/Selection Tools - CARTO in a fluid system.  For example, a user could click on physical elements of a map to select them in a table, then change the look of those items in the CARTO window. 

The array of GIS options today are phenomenal.  Yet I continually am working across different systems because there is no ideal platform for working with data, styles, and design in a fluid manner.  Perhaps we don't need a single platform for all that, but if anyone ever develops a method for underlying data to shift in response to graphic decisions, I suspect we will encounter an entirely new era of cartography.