The Global Goals have launched and data is a big part of the conversation. And now, we want to act … create and use data to measure and meet the goals. I’m presenting here a “sketch” of a way to track what data we have, what we can do with it, and what’s missing. It’s a Global Goals Data Census, a bit of working, forked code to iterate and advance, and raise a bunch of practical questions.
The Global Partnership for Sustainable Development Data launched last week, to address the crisis of poor data to address the Goals. Included were U.S. Government commitments to ” Innovating with open geographic data”. In the run up, events contributed to building practical momentum, like Africa Open Data Conference, Con Datos, and especially the SDG Community Data Event here in DC, facilitated by the epic Allen Gunn. And the Solutions Summit gathered a huge number of ideas, many of which touch on data.
- Put all the data in one place
- Create an inventory of indicators: + What exists ; + What goals are they relevant for
- Build a global goals dashboard
The action of taking stock of what data we have, and what we need, looks like a perfect place to start.
OKFN’s Open Data Census is a service/software/methodology for tracking the status of open data locally/regionally/globally. I forked the opendatacensus, got it running on our dev server, made a few presentation tweaks, and configured (all configuration is done via a Google Spreadsheet).
Each row of the Global Goals Data Census is a country, and each column is one of the 17 Goals. Each Goal links to a section of the SDGs Data, built off a machine readable listing of all the goals, targets and indicators.
This is truly a strawman, a quick iteration to get development going. It should work, so give it a quick test to help formulate thoughts for what’s next.
This exercise brought up a bunch of ideas and questions for me. Would love to discuss this with you.
- Does it make sense to track per Indicator, in addition to the overall Goal? There been a lot of work on Indicators, and they will be officially chosen next year.
- There may be multiple available Datasets per Goal of Indicator. The OpenDataCensus assumes only one Dataset per cell.
- For the Global Goals, are there non-open Datasets we should consider, due to legitimate reasons (like privacy).
- Besides tracking Datasets, we want to track the producers, users and associated organizations. The OpenDataCensus assume data is coming from one place (the responsible government entity). Much more complex landscape for the Goals.
- What is the overlap with the Global Open Data Index? Certainly the Goals overlap a little with the Datasets in the Global Census, but not completely. And the Index doesn’t cover every country that has signed up for the Goals. Something to definitely discuss more.
- Undertaking the Census, filling the cells, is the actual hard work. Who is motivated to take part, and how best to leverage related efforts.
- Many relevant data sets are global or regional in scope. How best to incorporate in a nationally focused census? How to fit in datasets like OpenStreetMap which are relevant to many Goals?
- There is an excellent line of discussion on sub-national data. There are also non-national entities which may want to track Goals separately. How to incorporate?
- What other kinds of questions do we want to ask about the Datasets, beyond how Open they are? Should we track things like the kind of data (geo, etc), the quality, the methodology, etc?
- Where could the Global Goals Data Census live? A good use of http://data.org/?
Does this interest you? Let’s find each other and keep going. Comment here, or file issues. One good upcoming opportunity is the Open Government Partnership Summit … will be a great time to focus and iterate on the Global Goals Data Census. There will be an effort to expand adoption of the Open Data Charter, “recognizing that to achieve the Global Goals, open data must be accessible, comparable and timely for anyone to reuse, anywhere and anytime”. I’ll be there with lots of mapping friends and ready to hack.
As you probably know, OpenStreetMap is a global project to create a shared map of the whole world. A user-created map, that is. Of course, in a project where anyone can add their data, there are many forms this can take, but the fundamental idea allows for local ownership of local maps — a power shift from the historical mapping process. Now, if I want to add that new cafe in my neighborhood to the OSM map, I can do it, and quickly. And I don’t have to worry that I’m giving away my data to a corporate behemoth like Google — open and free means no owner.
But lately I’ve noticed a trend away from the radical potential of such hyperlocal data ownership, as OSM gets more widely used and recognized, inspiring everyone from the White House to the Guardian to get involved in remote mapping (using satellite imagery to trace map details onto OSM, like roads and building outlines). But remote activities still require people who know the area being mapped to add the crucial details, like names of roads and types of buildings. So why is the trend worrying?
Since first training 13 young people in the Kibera slum to map their own community using OSM, in 2009, I have focused quite a bit on questions of the local — where things actually happen. And the global aid and development industry, as anyone can see who has worked very closely with people anywhere from urban slums to rural villages, tends to remain an activity that begins in Washington (or at best, say, Nairobi’s glassy office buildings) and ends where people actually live. Mapping, on the other hand, is an activity that is inherently the other way around — best and most accurately done by residents of a place. And, I would argue, mapping can therefore actually play a role in shifting this locus of decision making.
More poetically perhaps, I also see maps as representations of who made them rather than a place per se. So if the community maps itself, that’s what they see. And that can shift the planning process to highlight something different — local priorities. After all, a resident of a place knows things faster, and better, than any outsider, and new technology can in theory highlight and legitimize this local knowledge. But just as easily, it seems, it can expedite data extraction.
Even more importantly, as any social scientist will tell you, a resident of a place knows the relevant meaning and context of the “facts” that are collected — the story behind the data. An outsider often has no idea what they’re actually looking at, even if it seems apparent. (An illustrative anecdote: I was recently in Dhaka and we noticed a political poster with a tennis racket symbol. We wondered why use a tennis racket to represent this candidate? That’s not a tennis racket — it’s a racket-shaped electric mosquito zapper, said one Bangladeshi. What you think you see is not the reality that counts!)
So, when the process of mapping gets turned around — with outlines done first by remote tracing and locals left to “fill in the blanks” for those outside pre-determined agendas, that sense of prioritization is flipped. In spite of attempts to include local mappers, needs are often focused on the external (usually large multilateral) agency, leaving just a few skilled country residents to add names to features in places they’ve never set foot in before.
There are degrees of local that we are failing to account for. Thinking that once we’ve engaged anyone “local” (ie, non-foreign) in a mapping project we’ve already leveled the playing field is letting ourselves off the hook too easily. By painting the local with one brush we fail to engage people within their own neighborhood (not to mention linguistic, ethnic group or gender). We underestimate the power of negative stereotypes in every city and country against those in the direst straits, including pervasive ideas of lack of ability, tech or otherwise. This is especially problematic when dealing with the complex dynamics of urban slums, in my experience. But any place where we hope to eventually inspire a cadre of local citizen mappers who care about getting the (data) story right, it’s crucial to diversify.
Having dealt with these challenges nearly from day one of Map Kibera, I’m particularly sensitive to the question “How does a map help the people living in the place represented?” You might say that development agencies and governments have a clear aim for their maps that will help in very important ways. Therefore, a faster and more complete map from a technical standpoint is always better. (Indeed, getting support for open maps in the first place is quite an achievement). But more often than you’d think, the map seems to be made for its own sake or to have a quick showpiece.
Even if our answer is that the map would help if disaster struck, that seems to be missing the point and might not even be true in most cases. Strong local engagement and leadership are said to be “best practice” even in cases of disaster response and preparedness, though they aren’t yet the norm. Since the recent earthquake in Nepal, there has been a very successful mapping response thanks to local OSMers Kathmandu Living Labs. But critically, this group had been working hard locally for some time. And we should see this as a great starting point for widespread engagement, not an end product.
Another problem is that established mapping processes in the field aren’t questioned much, even when it comes to OSM. Since mapping is often something people do who are techies of some kind — GIS people, programmers, urban planners — organizers sometimes seem to forget to simplify to the lowest common denominator needed for a project. Does the project really need to use several types of technological tools and collect every building outline? Does every building address need to be mapped? If not, it just seems like an easy win — why not collect everything? One reason not to is because later when you find you need local buy-in, even OSM may be viewed as an outsider project meant to dominate a neighborhood, a city, especially in sensitive neighborhoods where this has indeed been a primary use of maps. I wonder if people will one day want to create “our map” separately from OSM. A different global map wiki which is geared toward self-determination, perhaps? That would be a major loss for the OSM community.
Perhaps I seem stuck on questions of “whose map” and “by whom, for whom?” Well, that’s what intrigued me about OSM in the first place. We used to talk a lot about the democratizing potential of the internet, about wikis and open source as a model for a new kind of non-hierarchical online organizing. Now, it’s clear that because of low capacity and low access, it’s actually pretty easy to bypass the poor, the offline, the unmapped. And because of higher capacity among wealthier professionals and students in national capitals globally, it’s easier for them to do the job of mapping their country instead. Of course there isn’t anything wrong with people coming together digitally over thousands of miles to create a map. In an idealistic sense it’s a beautiful thing.
But the fact is, we (that is, technologists and aid workers — both foreign and not) still tend to privilege our own knowledge and capacity far and away over that of the people we are seeking to help. We can send messages to the poor through a mobile phone, strategize on what poster to put up where, but the survey to figure out who lives there and what they care about? Still done by outsiders, hiring locals only as data extractors. As knowledge and expertise used to make decisions becomes more data-driven and complex, a class of expert and policy maker is created that is even more out of reach. Access has always been messy. Now (thanks in part to mobile phone proliferation and without much further analysis), we hardly talk about access anymore. The fad has shifted to “big data” and other tech uses at the very top.
To me, OSM was always so much more than just a place where people shared data. It was one small way to solve this problem of invisibility bestowed by poverty.
The possibilities of OSM to empower the least powerful are still tantalizingly close on the horizon. With just a few tweaks to existing thinking, I hope we can tip the scale — we can prioritize the truly local and allow the global to serve it. But to do so we must resist glib and lazy thinking around how those processes actually work. We must pay attention to order of operations (who maps first? whose data shows up as default?), and subtleties of ownership and buy-in, and we must examine who we think of as “local”.
But most of all, and most simply, we have to reorient our thinking in designing mapping projects. It could be quite straightforward, for instance:
- integrating more localized thinking into training courses and OSM materials, suggesting that people on the margins (such as slumdwellers) can and should be learning OSM tools as well;
- training people to think about social context and local priorities not just technology;
- doing more offline outreach and print map distribution; creating new map renderings highlighting levels of local and remote mapping;
- making it the norm that residents of a neighborhood be involved in early project planning;
- engaging with people who have long done offline forms of data gathering and mapping in their communities.
- Considering that first pass on a blank map might matter.
- And considering that leading local technical mappers start to conceptualize their roles more as mentors to communities and to new mappers in those places than as expert consultants to foreign organizations. I’ve seen the light bulb go on when that happens.
Yes, ultimately, we will do best to address the incentive structures of international funding which are keeping us from achieving what we’d all undoubtedly wish to see, which is everyone getting the opportunity to put themselves on the map.
Many will say that it is just too hard, too time consuming, too cumbersome, too expensive, too — something — to really prioritize local in the way I’m talking about. We’ll never hit “scale” for instance. Something needs to be done now. But I would ask you to reconsider. I believe that in fact, sensitive and thoughtful engagement with communities is the only real path to scale and sustainability for many kinds of “crowd-” or “citizen-” based data work that is now happening, and most certainly the only way to reach the real target of any development-oriented data effort — actual improvements in the lives of the world’s poor and marginalized.
This post also appears here on Medium.
I’ll confess, I’ve never been a huge fan of M&E. While I absolutely love the data and statistics and numbers and fascinating insights of a good evaluation, as the founder of a nonprofit in Kenya, Map Kibera, doing a quality job on monitoring and much less evaluation was daunting.
Not only because we were under-resourced, and lacked high staff capacity (our members all coming from the Kibera slum) – but Map Kibera was actually set up in part to counter a problem obvious to any Kibera resident. NGOs and researchers were constantly collecting data, but then were usually never seen or heard from again.
Where Does the Data Go?
Was that data even seen again by the collecting organization after their project reports were turned in? What good was all that time and energy spent – on the part of the organization, but more importantly on the part of the good citizens of Kibera, who were tired of answering questions about their income and toilet habits three times in one week? And didn’t they have a right to access the resulting information as well?
Now many organizations, including Map Kibera, our organization GroundTruth Initiative, and others such as the panelists who joined me at the M&E Tech Feedback Loops Plenary: Labor Link, Global Giving, and even the World Bank have put an emphasis on citizen feedback as the core of a new way of doing development.
It’s possible to imagine a world where some of the main reasons for doing monitoring and evaluation are shifted over to citizens themselves – because they want to hold to account both governmental and non-governmental organizations so that the services not only get to the right people, but those people can drive the agenda for what’s needed where.
While a complicated study on, say, the school system and education needs of Kibera people might provide insight, if it sits on a shelf and doesn’t inspire grassroots pressure to shift priorities and improve education, what is the use of it? Why not instead invest in efforts to collect open data on education jointly with citizens, like our Open Schools Kenya initiative?
The benefit here is that the information is open and collectively tended – meaning kept up to date, shared, made comparable with other data sets (like Kenya’s Open Data releases from the government), and used for more than just one isolated study. It’s a way for the community to assess the status of local education itself. In our own neighborhoods and school systems, we wouldn’t have it any other way.
A Citizen Led Future
As noted by Britt Lake of Global Giving on our panel discussion, taking this concept to its logical conclusion there would be a loss of control by development agencies. Is this the real hurdle to citizen-led data collection? To what extent can aid systems be devolved to the people who are meant to primarily benefit?
A truly forward-thinking organization would embrace this shift, because with better information being collected and used at the grassroots, there will be more aid transparency overall and less opportunity for gaming the system. If you think that no small CBO has ever submitted a bogus progress report which went up the chain at USAID, think again.
Nowhere in this system is there incentive to give an accurate account of failures or document intelligent but unpredicted programmatic adaptations and detours that were made. Yet if there’s one thing I know the Kibera people want, it’s for the many organizations they see around them in the slum to be held accountable for all the funds they receive, and for delivering on their years of promises for improving the slum.
In fact, transparency around aid at this hyperlocal level is something we should even feel an ethical obligation to provide. Here’s hoping that in 20 years, the impact of the open data and feedback movements will mean that public information about projects done in the public good is reflexively open and responsive, and M&E as a separate and often neglected discipline is a thing of the past.
This post was originally featured on ICT Works as a Guest Post.
How can all the information about Kenyan schools, including data released by the Kenyan government, and citizen mapping, have a greater impact on education? We’ve been working for the past few months on a project to make information about schools much more available and useful in Kenya. It’s a joint operation between GroundTruth Initiative, Map Kibera, Development Gateway, Feedback Labs, and the Gates Foundation among others.
Many people collect information about education – and they sometimes make it open and free to use. So, why isn’t it easy to find information about a particular school – for a parent, or for an education researcher? Much of the information that’s out there isn’t connected to the other data – and especially when it comes to informal schools, which provide a great deal of the education services in places like informal settlements.
Citizen data – like mapping schools using OpenStreetMap – should also be easy to combine and compare with official education data. And finally, all this info could be accessible and useful to everyone from parents to policymakers.
So, we’ve started with Kibera as a test location for the Open Schools Kenya project.
Over the past few months, the Map Kibera team has engaged parents, school leaders, and education officials in Kibera to find out how the informal school sector can be more visible, and to assess the demand for information on education. Now, a widespread effort is underway by Map Kibera to make sure that the schools data that the team collected a few years ago is still accurate, and to add new info as well. We’re also collecting photos of each school, no matter how small. Every one will have a page on the website, really bringing the informal school sector to light. Formal schools in Kibera will be there too.
Much of the work so far has been around engaging important leaders in the community, who care about local kids getting the best education. Mikel Maron of GroundTruth was recently in Nairobi working on the project and will be updating in a separate post about this busy trip. Ultimately, the community wants to know more about its schools, and to improve them. So do education supporters throughout Kenya.
But beyond this important mission of organizing and making interoperable many data sets across the vast education sector in Kenya, we’re also working on an ambitious hypothesis: that parents and community leaders in education will want to provide feedback on schools, which in turn will inform policy and improve individual schools. Ultimately, our platform will be a place where people can not only be consumers of information, but will provide their own opinions and suggestions on schools, and, importantly, submit corrections and updates to the data on the site. Given the early positive response to these ideas, we’re optimistic that this will be possible in Kibera and also Kenya-wide.
The project is not just about education, either. It has far-reaching potential in other sectors as well. We hope to demonstrate that citizen data, official data, academic research and more can come together and be part of a conversation with those on the ground who feel the impact most of government policy in every sector – ordinary citizens. And, that this kind of conversation means that people “own” their own information, and we can see the beginnings of a true “feedback loop” or dialogue between citizens and government, through the medium of shared data.
This article was originally posted on the Map Kibera blog, July 3, 2014.
Presenting today at 2014 Land and Poverty Conference
The generation, management, and distribution of land tenure and land rights spatial data is generally restricted in traditional closed information architectures, in part for legitimate legal and security reasons. However, the gap thus far in adoption of more innovative open data practices means a missed opportunity to address critical issues of accountability and access to land rights data for the most marginalized. Open data communities, such as the OpenStreetMap (OSM) project, have proven to be transformational collaboration platforms in domains like disaster response, and now several projects show promise for this open approach to land rights. This paper explores two distinct kinds of contribution from the OSM approach. Firstly examined is the direct use of OSM as a core engagement activity for advocacy, planning, and accountability by communities asserting their rights for representation and security. Lessons from projects in Kibera, Indonesia, East Jerusalem, and La Boquilla, Columbia are detailed. Secondly, OSM is looked at as a starting technical software base & community model for collaborative, open geographic data creation & sharing, forked (in the open source sense) for adaptation to the particular access rules and data structures required for land tenure registration.
OpenStreetMap is a free and open map database of the entire world, build from the active collaboration of tens of thousands of volunteers globally. It is sometimes called the “Wikipedia of Maps”, since contributing to OSM is open to absolutely everyone, and the data and code ecosystem are all in the commons, licensed for re-use and re-distribution. Such “crowd-sourced” initiatives have been held up to intense scrutiny, and have been found to meet or exceed the data quality of traditional sources, and are certainly more accessible. Thus, OSM has been adopted by companies, governments, universities, international organizations, and software developers. The domain of disaster response, particularly through the work of the Humanitarian OpenStreetMap Team (HOT), has been transformed by the practice of OSM, after the devastating 2010 Haiti earthquake, where OSM became the base map for the response. More recently, Typhoon Haiyan saw over 1000 contributors map millions of features and damage point, for use by organizations such as the American Red Cross, UNOCHA, MSF, the World Bank, and the New York Times.
In the first part, community mapping in OSM has proven a valuable supplement to official tenure process, or to help communities advance their rights agenda. A core principal of OSM is that the most accurate map will be made, with the right basic training, by people residing in a location, not necessarily survey and cartographic experts. This especially holds true in communities at risk of dispossession by authorities, where official maps may portray a geographic mis-reality more conducive to their plans, or the community itself may not be physically accessible and suspicious of outsiders. Take the Kibera slum of Nairobi, the object of many controversial upgrading and resettlement plans, was a blank spot on official maps, until the public infrastructure (including roads and paths, sanitation infrastructure, health services, etc) was mapped by Map Kibera in OSM. This project (and now organization) taught young residents the technology to make the map themselves, and represent their community. Being a completely open system, the goal was not to map private residences, but to make visible the community as a whole, and of particular interest here, track the events and reactions to upgrading programs. Spatial data is used as the context for citizen media, in the form of bloggers and videographers, who geolocate stories on the Voice of Kibera, an instance of the Ushahidi platform. Stories have covered demolitions in Soweto East, resettlement to the “decanting site”, and most recently efforts for the Nubian community (the original settlers of the slum) to gain title and recognition. The community, evidence based perspective, linked to specific locations, shared openly, both online and offline, has provided a much needed monitoring and accountability function in these contentious processes.
Another example is found in Indonesia, where HOT undertook a wide ranging effort to collect data in OSM for disaster preparedness risk models. This project involved training and connections with a large number of actors, from disaster preparedness agencies of the Indonesian government, to university students, to city officials, planting the seed of an open data community in Indonesia. As is often the case with open data, unexpected uses and connection arose, specifically with the ACCESS civil society strengthening program. Part of ACCESS activities involved “poverty mapping”, participatory mapping of community infrastructure and socio-economic data, a step to create community development plans. These poverty maps were created on paper in a facilitated process, and while quite detailed, beautiful and useful, the opportunity for re-use, analysis, and access to these maps were limited. HOT adapted the OSM workflow to these participatory processes, which allowed for better and easier data collection, and connection into a host of open source tools to visualize and analyse that data. A major change to OSM tools was made to handle private and secure data. While communities were comfortable sharing openly geographic data on community infrastructure, they did not want to share detailed socio-economic information. To accommodate that, the OSM editing tools were adapted to work with multiple databases, both the open, public database, as well as private database. This provides a means to selectively share community data, without invoking another tool.
In the second part, OSM is examined as both a technical and community starting point to open creation and sharing of tenure data. It’s entirely fair to say that OSM has been the most successful approach to collaborative geographic data ever. That’s due to some key decisions of the architecture, such as flexible tagging rather than rigid ontology, and transparent history of individual contributors. The software ecosystem is quite robust, including what might be the most user friendly map editing client developed, iD. The community approach is decentralized, distributed responsibility, yet quite unified through clear purpose. However, not all kinds of data are appropriate for OSM, as OSM focuses on visible infrastructure, land use, and some administrative boundaries, but by community convention, does not contain cadastre (among many other things). As well, OSM is completely open for anyone to edit, and that is simply not possible for land tenure data; there must be some process for validation and protection of submissions. There are other projects that have started with OSM architecture and principals, and adapted to a different database. For instance the USGS prototyped its use for the National Map; the Moabi project is deploying OSM architecture for collaboration on natural resource extraction data; and the before mentioned poverty mapping project in Indonesia. In order to support approaches like the Social Tenure Domain Model, there are very particular adaptation requirements to OSM. Access control to both data editing and data structures is required, and experiments show that OSM architecture is amenable to such changes. Just as crucial, OSM makes individual contributors visible, and relies on their participation, communication and collaboration. Despite the necessary institutional element of land tenure, a successful program must retain such individual empowerment.