UNDP, USAID and the International Peace Institute recently published a report on uses of technology for the prevention of violence and conflict, exploring global trends and drawing from examples in Kenya, Latin America, Kyrgyzstan, Sudan and South Sudan. The full paper is available here, below is a summary of highlights from my reading. It’s not an attempt to summarize the full contents, only the points that stood out. (Also, full disclosure! I wrote the chapter on Sudan and South Sudan.)
This report explores the ways in which ICTs and the data they generate can assist international actors, governments, and civil society organizations to more effectively prevent violence and conflict.
The report opens with a set of recommendations that emerge from the exploration in its later chapters. One overarching theme is the need to remain practical: examine all tools (rather than get carried away by fads), consider the context (rather than accept pre-made solutions from the outside), and integrate technology into existing initiatives and partnerships (rather than create technology-centred projects). There is also an important emphasis on considering conflict sensitivity and privacy. There aren’t that many tools available to practitioners looking to act on such considerations, but it’s well worth taking a look at this recent paper. The summary also includes a recommendation that special attention should be given to encouraging horizontal information flows. It’s a welcome statement, but I wonder whether in practice implementers are willing to give up that much control.
Introduction (Francesco Mancini)
How can new information and communication technologies (ICTs) aid international actors, governments, and civil society organizations to strengthen their voice and action in order to more effectively prevent violent conflict?
Francesco introduces the question that all case studies are organized around. He points to two over-arching issues that I find particularly useful in thinking about this top. One is the warning-response gap: a reminder that we may collect a lot of data, but it’s the actions with regards to this data that will prevent conflict. The second is the distinction between operational conflict prevention (immediate, affecting direct causes) and structural prevention (longer term, concerned with underlying causes).
As a field of practice in the making, what we term here “Big Data for conflict prevention” is best characterized by its potential rather than by its track record.
This first chapter is a thoughtful exercise in providing a framework to think about big data and conflict prevention, defining terms and setting the scene. The authors explore several taxonomies of big data. One important point they draw out is that big data refers to traces of human actions picked up by digital devices. They distinguish these traces from data that may be more or less exogenous to human actions (prices, climate), but is not a direct digital expression of human actions.
The authors make two concrete suggestions for uses of big data in conflict prevention that, though largely speculative at this stage, are compelling. One is the use of data to understand population movement, mainly through CDRs (if that means nothing to you, check this out). In situations where migration patterns or group movements are known to affect conflict dynamics, such data provided in real-time could be very valuable to operational conflict prevention activities. Second, big data can help understand sentiment in a population by providing a source of perceptions data. UN Global Pulse has piloted a project to analyse perceptions expressed on Twitter in Indonesia. Their implementing partner, CrimsonHexagon, did something similar in Egypt too. There are some concerns about data bias in countries with low Twitter use; they are noted with respect to Egypt here. Perceptions data can also be analysed manually, as was recently done in Kenya in relation to hate speech prior to the elections.
The authors also quote six “provocation” for big data that are interesting:
1. Automating Research Changes the Definition of Knowledge
2. Claims to Objectivity and Accuracy are Misleading
3. Bigger Data are Not Always Better Data
4. Not All Data Are Equivalent
5. Just Because it is Accessible Doesn’t Make itEthical
6. Limited Access to Big Data Creates New Digital Divides
(Source: Boyd, Danah and Crawford, Kate, Six Provocations for Big Data)
Quoting Alex de Waal, the authors also remind us that technology cannot replace politics. Difficult decisions will continue to have to be made; big data can’t give us the answers because the answers are fundamentally political. Linked to this reminder, they hone in on a possible unintended consequence:
There is a real risk that Big Data may undo years of efforts to try and use technology to put affected community at the center of conflict prevention, on both the demand and supply sides.
In other words, as practitioners we should be wary of facilitating a situation where people can say that “the data tells us” anything definitive about conflict. Data may tell us that conflict is likely to happen, but will shed much less light on why it will happen.
Given the sheer scale and demographics of Latin America’s digital natives, it is hardly surprising that they are among the world’s most active users of social media. Indeed, six Latin American countries are included in the top ten most actively spending time in web-based social networks.
This chapter begins with a review of research on the negative uses of technology in Latin America. One example (in Spanish) describes how community organizing to share information on safe areas in Monterey was infiltrated by gangs. This paper uses social media trends and in particular Twitter hashtags to examine the drug war in Mexico.
The examples of ICT use in Government and civil society focus on tools for data collection and sharing. In their framing discussion, they make a clear distinction between these data gathering tools and data analysis ICTs, and pose an important general question (that they unfortunately don’t have the space to answer here):
As more and more data exists online—referred to in some circles as “digital exhaust”—there are emerging questions about what types of ICTs will prevail. Put another way, will horizontal approaches be in a position to analyze data at scale, or will only centralized organizations using vertical approaches and sitting on large regularized datasets be in a position to meaningfully engage with it?
The government initiatives examined focus on data gathering platforms. Most are modeled after NYPD’s COMPSTAT, and are a combination of data centralization through a powerful, well-structured database, with the ability to view data summaries in graphs and on maps. Discussion of the use of technology for dissemination is limited to how social media is used (e.g. in Mexico via @PGJECoahuila).
The section on community initiatives is more interesting, covering uses of technology to report crime and to shift the discourse on crime. The authors give two examples of bounded crowdsourcing to report crime. Unidos Pela Seguranca is a pilot initiative in Brazil that collects information from reporters via an online form, maps it publicly and provides verified information to the police. An initiative of Citivox to track electoral violence sounds like an effective early warning – early response system, but not enough background is provided and there is no further information available online.
Even more interesting are community initiatives to shift the discourse on crime. The simplest ones describe the importance of internet media in providing alternative journalism on crime, such as El blog del narco or Notinfomex. Others go further and actively promote messages of peace, like Nuestra Aparente Rendicion, which even includes a peace map.
Two of the examples provided, however, may be misleading. The paper mentions WikiNarco, but neither WikiNarco.com nor the wikia link provided seem likely to be the site referred to by the authors. The WikiNarco twitter handle has had no activity since September 2011. The authors also report that the Tehuan platform run by Center for Citizen Integration is an example of crowdsourcing violence data. However, the platform appears to be used mainly for citizen reporting on governance and local priorities (most reports on the platform are about traffic and public services, only a handful about public safety).
Early Warning in Kenya (Godfrey M. Musila)
The Kenya example is interesting because technology has been introduced to a well-structured, existing conflict prevention body: the National Steering Committee for Peace Building and Conflict Management (NSC).
The NSC has divided the country into three clusters—urban, rural, and pastoral—reflecting the broad categories of conflict areas and types of conflict. Each of these clusters has different conflict dynamics and indicators. Conflict-early-warning information is collected in each of the clusters by peace monitors and members of District Peace and Development Committees who report directly to the NSC. Each of the forty-seven counties in Kenya is manned by at least one peace monitor.
The author describes two ICT efforts linked to the NSC. The Uwiano Platform was originally set up to monitor violence in hotspots during the run-up to the 2010 referendum on the constitution. The platform received SMS and internet reports from the crowd. Based on analysis of these reports, District Peace and Development Committees could request funds from a rapid-response mechanism to take action at the local level. As a result of its succes, the platform continues to operate after 2010 in the same areas. Kenya’s National Conflict Early Warning and Early Response System also receives crowdsourced information via a web-report (Amani 108 online reporter) and an SMS shortcode (108). The system integrates these reports with field reports from District Peace and Development Committees, and publishes a map of all reports. The author offers no assessment of the success of this initiative.
The assessment of CEWARN’s ICT4Peace initiative is interesting, and points largely to a disappointing attempt at introducing radios (the wrong technology) to an already overly complicated reporting process (via the CEWARN Reporter, a network software program used by CEWARN country coordinators to enter and store the standardized field reports submitted to them by CEWARN field monitors). The author reports that CEWARN currently holds ten years worth of data that has not been analyzed.
ICTs in Kyrgyzstan (Anna Matveeva)
The punch-line at the end of this chapter is well worth the read:
There is no evidence that ICT expedited the response to the June 2010 conflict by the government or the international community. However, mobile phones and interaction on popular websites played a role on the community level in fostering group action toward fight or flight.
To arrive at this conclusion, the author starts by looking at negative uses of technology during conflict events in Kyrgyzstan in 2010:
A common view in Kyrgyzstan is that mobile technology worsened the situation in the June clashes. This is because oral transmission of information through cell phones is susceptible to distortion and can be easily put to negative use.
This common view is only partially correct, since most villages affected were outside the mobile network. Nonetheless, mobiles were used to convey threats, organize violent acts and spread negative propaganda. Online discussion forums and YouTube videos tended to polarize the debate. But not all was negative. Neutral information sources became the early warners: the Ferghana website was trusted by people from both communities, who read its updates and spread messages through mobile phones on crowd movements and on which neighborhoods were under attack.
The role of social media in the run up to the power change in April 2010 is also informative:
The young men who rallied in the streets of Talas and Bishkek did not do so because of Internet postings. Rather, it was witnesses like journalists, students, and NGO representatives posting their impressions in real time on the web who created the effect of a general politicization.
This comment about the difference between offline and online actors is important, and provides a key to understanding the impact of technology that is valid for other contexts (certainly for Sudan). It also leads to this general point:
ICTs need to be considered not in isolation but in how they relate to conventional, face-to-face social interaction: they magnify the messages already in public domain.
The author then discusses an early warning systems using SMS in 2011 run by the Foundation for Tolerance International, suggesting it largely failed due to barriers to accessing SMS. Similarly, UNDP’s electoral violence system (also using SMS reporting) was not very effective, receiving few reports on violations of the Electoral Code of Conduct (most messages were voting-related inquiries). The system worked better during the 2012 local elections, when it was used only by NGO partners to report on specific issues. ACTED also produces a map of potential points of conflict in the Ferghana Valley. (The author couldn’t make it work; I had no problem with it – it’s powered by ArcGIS and not dissimilar to the CRMA initiative in the next case study.)
Keeping in mind her earlier appraisal of how ICTs should be understood in a conflict context, the author makes the following comment about the Foundation for Tolerance, UNDP and ACTED initiatives:
The ICT projects did not pay sufficient attention to the way ICTs and other communication flows were already playing out in the local context during crises.
Based on this assessment, she makes three recommendations that I think are of particular interest. First, we should focus on understanding the fight or flight patterns generated by ICT during an active conflict, and intervening accordingly to mitigate negative effects. Second, it is more effective to integrate conflict prevention initatives in the mainstream websites already used by people, rather than creating new websites. Finally, we should acknowledge that with certain technology will only reach elites – and that in some situations, reaching elites may be a legitimate tactic for conflict prevention.
Sudan and South Sudan (I wrote this one!)
I cover three initiatives that used technology for conflict prevention in Sudan and South Sudan in this chapter. I’ve blogged about two of them before: the Sudan Vote Monitor here and SUDIA’s Community Communications System here and here. My friend Margunn and I both talked about the third project, the Crisis and Recovery Mapping and Analysis, here and here.
There are some obvious shortcomings to these three initiatives. Access to ICTs in Sudan and South Sudan is limited. Government buy-in is difficult to maintain – in fact, since the case study was published, SUDIA has been asked by the authorities to stop this particular activity. Writing this paper reminded me again that the potential of ICT for conflict prevention will be determined mostly by these two issues.
The one take-away point from the analysis of these initiatives that I hadn’t formulated in my mind before is a distinction regarding the use of crowdsourced data. I see little or no evidence that the collection of crowdsourced data for early warning is likely to enable faster and more effective responses to drivers of larger conflicts. However, crowdsourced data can and does help governments and communities better respond to ongoing, low-level, localized disputes, and avoid their escalation. The distinction here is on two aspects: the timing of the response (crowdsourced data helps understand patterns after the initial peak, not prior to it) and the scale of the response (crowdsourced data works best closer to local contexts).