Micro-dynamics of Reciprocity in an Asymmetric Conflict (Updated)

Thomas Zeitzoff is a PhD candidate at New York University. I came across his research thanks to the Ushahidi network, and am really glad I did. He wrote a really neat paper earlier this year on “The Micro-dynamics of Reciprocity in an Asymmetric Conflict: Hamas, Israel, and the 2008-2009 Gaza Conflict,” which he is revising and submitting to peer-review.

Updated: Thomas kindly sent me the most recent version of his paper which you can download here (PDF).

I’ve done some work on conflict event-data and reciprocity analysis (see this study of Afghanistan), but Thomas is really breaking new ground here with the hourly temporal resolution of the conflict analysis, which was made possible by Al-Jazeera’s War on Gaza project powered by Ushahidi.

Abstract

The Gaza Conflict (2008-2009) between Hamas and Israel was defined the participants’ strategic use of force. Critics of Israel point to the large number of Palestinian casualties compared to Israelis killed as evidence of a disproportionate Israeli response. I investigate Israeli and Hamas response patterns by constructing a unique data set of hourly conflict intensity scores from new social media and news source over the nearly 600 hours of the conflict. Using vector autoregression techniques (VAR), I find that Israel responds about twice as intensely to a Hamas escalation as Hamas responds to an Israeli escalation. Furthermore, I find that both Hamas’ and Israel’s response patterns change once the ground invasion begins and after the UN Security Council votes.

As Thomas notes, “Ushahidi worked with Al-Jazeera to track events on the ground in Gaza via SMS messages, email, or the web. Events were then sent in by reporters and civilians through the platform and put into a Twitter feed entitled AJGaza, which gave the event a time stamp. By cross-checking with other sources such as Reuters, the UN, and the Israeli newspaper Haaretz, I was able see that the time stamp was usually within a few minutes of event occurrence.”

Key Highlights from the study:

• Hamas’ cumulative response intensity to an Israeli escalation decreases (by about 17 percent) after the ground invasion begins. Conversely, Israel’s cumulative response intensity after the invasion increases by about three fold.

• Both Hamas and Israel’s cumulative response drop after the UN Security Council vote on January 8th, 2009 for an immediate cease-fire, but Israel’s drops more than Hamas (about 30 percent to 20 percent decrease).

• For the period covering the whole conflict, Hamas would react (on average) to a “surprise” 1 event (15 minute interval) of Israeli misinformation/psy-ops with the equivalent of 1 extra incident of mortar re/endangering civilians.

• Before the invasion, Hamas would respond to a 1 hour shock of targeted air strikes with 3 incidents of endangering civilians. Comparatively, after the invasion, Hamas would only respond to that same Israeli shock with 3 incidents of psychological warfare.

• The results conrm my hypotheses that Israel’s reactions were more dependent upon Hamas and that these responses were contextually dependent.

• Wikipedia’s Timeline of the 2008-2009 Gaza Conflict was particularly helpful in sourcing and targeting events that might have diverging reports (i.e. controversial).

Perhaps the main question I would have for Thomas is how he thinks his analysis and findings could be used for conflict early warning and rapid response to violent conflict.

Patrick Philippe Meier

How Ushahidi Can Become a Real Early Response Platform

As is well known in the field of disaster management, preparedness and contingency planning is core to the success of people-centered early warning networks. See my previous post on “Insights from Disaster Early Warning.” While Ushahidi users can now subscribe to alerts by SMS, there are currently no response protocols linked to these individual alerts. The latter is not necessarily Ushahidi’s responsibility, but the team can play an important role by integrating a form for response protocols within the Ushahidi platform.

Ushahidi is closing the feedback loop between crowdsourcing and crowdfeeding by introducing an alert subscription feature. However, the notion of crowdsourcing response requires further develop- ment so that operational protocols can be implemented.

In other words, although users can subscribe to alerts, this does not mean they will be prepared to react or know what the best response is when they receive said alerts.

I witnessed this first-hand when setting up a community-based conflict early warning and response network in Timor-Leste. To be sure, the notion of “preparing for conflict” is not one that always comes naturally.

Imagine if the disaster early warning community only focused on forecasting and payed no attention to preparedness and contingency planning (PCP). Millions more would die every year without training, shelters and regular drills. PCP can also be applied to conflict early warning and rapid response.

An organization implementing Ushahidi would simply need to do the following: for every indicator category the organization identifies, a response protocol for that alert would be added to an Ushahidi “Reponse Protocol Form.”

So for Indicator “A”, a series of response protocols for that indicator would be listed in the Response Form. These protocols could be customized based on where and when the incident took place. The protocols could include information on nearby shelters, hospitals, police stations, food stocks, etc.

This is primarily logistical but PCP can also be applied to identify and formulate tactics for conflict management.

In Timor-Leste, I ran a short and informal PCP excercise by asking members of a local community to think through what they would do if a land dispute occured in their village. For example, who would they call or notify when they hear about the incident. How would they try and intervene to prevent the situation from getting worse? These questions generated a rich set of action-oriented protocols that drew on their own traditional conflict resolution mechanisms.

Ushahidi can plan an integral role in encouraging organizations and communities to think preventively by adding a simple technical functionality to the platform. “Response Protocol Forms” could be used to follow up SMS alerts with SMS Response Protocols. Clearly, the organization deploying Ushahidi would be responsible for ensuring the protocols are correct and up-to-date.

How Ushahidi Can Become a Real Early Warning Platform

Ushahidi currently uses incident reporting (or more technically event-logging) as the methodology to document violent events that have taken place. While Ushahidi’s use of FrontlineSMS accelerates the crowdsourcing of crisis information, the violence reported on Ushahidi has by definition already occurred and thus cannot be prevented.

This blog post is the first of a two-part series on taking Ushahidi to the next level. The second part in this series  addresses “How Ushahidi Can Become a Real Early Response Platform.”

Current Setup

Ushahidi’s use of SMS to crowdsource crisis information means that violent incidents can be reported in quasi real-time, a significant advantage over traditional conflict early warning systems such as the Horn of Africa’s Conflict Early Warning and Response Network (CEWARN) and the now defunct FAST Early Warning System by Swisspeace.

To be sure, the use of SMS and geo-tagging means that escalating violence can be documented as it happens and where it happens. This can alert organizations of the need to contain or prevent further bloodshed. However, as has been argued by scholars and practitioners, as more blood is spilled the probability of reversing the violence without military intervention grows slim.

Situation Reports

The more robust applied social science methodologies I have come across for field-based conflict early warning combine both incident and situation reporting. Think of them as two types surveys. While the list of indicators in incident reports (IncRep) comprises violent events that seek to be prevented, the list of indicators in situation reports (SitReps) comprises events that are thought to render IncRep indicators more likely.

In other words, if event E is listed on an IncRep, the corresponding SitRep would include events E – t, i.e., those events that usually precede the violent event E in time. These SitRep events can be political, social, economic, ecological, historical etc., and should be grouped into such categories.

However, events E – t that tend to mitigate or prevent events E should also be included in SitReps. We want to know what is going right in order to identify existing entry points for conflict management. Violent conflict is never total in the Clausewitzian sense of total war. There are always pockets of cooperation and intervention. These, however indirect, need to be identified and understood.

SitReps are completed periodically, e.g., on a weekly basis, unlike IncReps, which are completed episodically, i.e., as incidents of violence take place. In other words, regular and consistent situation reporting should be encouraged. SitRep events should be formulated as indicators framed as questions.

For example, “Are university students having their freedom of speech curtailed?” would imply that a regime has acted (an event) to restrict student behavior. This act could elicit student protests and thence “a violent crackdown by government forces,” which would be an IncRep indicator.

SitRep surveys should also use a Likert scale, i.e., answers to SitRep indicator questions should range from “strongly disagree” to “strongly agree.” This means the answers can be weighted. See the CEWARN screenshot above for an operational example of a situation report.

Rationale

Instead of only monitoring incident reports, i.e., violence that has already come to pass, Ushahidi users would be able to monitor and analyze the causal factors themselves to determine whether they are increasing in number and intensity. This could signal potential early warning signs before the causal factors translate into violent events.

As more IncReps and SitReps are completed, users can also empirically assess which SitRep indicators act as the real triggers (and “preventers”) of the violent events documented in IncReps. In addition, after weekly SitReps are completed over several months, they can be aggregated to form a baseline for what the “average” week looks like.

Baseline Analysis

This means that users could then compare each new situation report with the baseline as depicted above. Inflection points, the doted circles, are points of interest since they denote change in trends. The dotted lines represent the threshold beyond which an organization will intervene. Alerts 1 and 2 signal that the threshold has almost been reached.

The important point to note is that baseline analysis of SitRep indicators can identify when the causes of violent conflict are increasing in intensity and frequency. Furthermore, patterns might be identifiable in the causes of episodic conflict and these could inform structural prevention strategies as well as conflict sensitive programming.

Next Steps

It would be great for Ushahidi to at least provide users with the option of setting up their own Situation Report form. I would also recommend that Ushahidi automatically flag when thresholds are crossed. In addition, Ushahidi should integrate some basic statistical techniques so that SitRep indicators (causes and preventers of conflict) could be automatically flagged when they are statistically correlated with IncRep indicators, i.e., violent events.

The purpose of including SitReps in Ushahidi is to encourage evidence-based programming and make early warning less of a hypothetical possibility.

OECD: The Future of Conflict Early Warning and Response

The OECD’s publication on “Preventing Violence, War and State Collapse: The Future of Conflict Early Warning and Response,” (PDF) has finally been published. I was solicited by the OECD to be the main peer reviewer for the publication, which was authored by my colleague David Nyheim.

I had a lot to add so the peer review process turned into a consulting assignment back in September 2008. My main contribution to the publication included paragraphs on:

• Evaluating CEWARN, ECOWARN and CEWS
• Fourth Generation Early Warning Systems
• Current Trends in Warning and Response
• People-Centered Early Warning
• Forecasting Armed Conflict
• Advances in Technology

I added references to the Harvard Humanitarian Initiative’s (HHI) Program on Crisis Mapping and Early Warning (CM&EW) as well as to Ushahidi’s approach to crowdsourcing crisis information.

Patrick Philippe Meier

Conflict Early Warning Blog: One Year On

I started this blog one year ago and it’s been great fun! I owe the Fletcher A/V Club sincere thanks for encouraging me to blog. Little did I know that blogging would so stimulating or that I’d be blogging from the Sudan. I have authored a total of 48 blog posts on conflict early warning and early response.

The Top 10 posts:

1. Crimson Hexagon: Early Warning 2.0
2. CSIS PCR: Review of Early Warning Systems
3. Conflict Prevention: Theory, Police and Practice
4. New OECD Report on Early Warning
5. Crowdsourcing and Data Validation
6. Sri Lanka: Citizen-based Early Warning/Response
7. Online Searches as Early Warning Indicators
8. Conflict Early Warning: Any Successes?
9. Ushahidi and Conflict Early Response
10. Detecting Rumors with Web-based Text Mining System

I also started the iRevolution blog at the same time and have authored a total of 212 blog posts there. That makes 260 posts in 12 months. Now I know where all the time went!

The Top 10 posts:

1. Crisis Mapping Kenya’s Election Violence
2. The Past and Future of Crisis Mapping
3. Mobile Banking for the Bottom Billion
4. Impact of ICTs on Repressive Regimes
5. Towards an Emergency News Agency
6. Intellipedia for Humanitarian Warning/Response
7. Crisis Mapping Africa’s Cross-border Conflicts
8. 3D Crisis Mapping for Disaster Simulation
9. Digital Resistance: Digital Activism and Civil Resistance
10. Neogeography and Crisis Mapping Analytics

I look forward to a second year of blogging! Thanks to everyone for reading and commenting, I really appreciate it!

Patrick Philippe Meier

Ushahidi and Conflict Early Response

Ushahidi’s approach to conflict early warning/response is refreshingly different from mainstream conventional approaches.

Conflict early warning systems like CEWARN in the Horn of Africa, ECOWARN in West Africa and the African Union’s CEWS are all top-down, centralized and hierarchical. Some argue that these systems actually take both a top-down and bottom-up approach since field monitors (at the bottom) document early signs of conflict escalation for policy makers (at the top). True, but as we know all too well, policy makers rarely close the feedback loop by responding early  and effectively to conflict warnings.

To paraphrase Erik Hersman at Ushahidi, this lack of response is perhaps like the “pothole theory”: you generally don’t care about the pothole on a street, unless it’s yours. This helps to explain why we don’t respond to problems further down the street. Ushahidi therefore takes a different approach; one that I like to call the “bottom-bottom” approach.

Just how different is Ushahidi’s approach to that of other NGOs? Take for example Swisspeace’s  conflict early warning system, FAST, one of the early pioneers in the field of conflict early warning. The architects of FAST understood that early warning information needs to be actionable and customized to meet the demands of the end users. They described this using the analogy of planning food for a dinner party.

“We not only need to know how many people are coming but who is coming, the time of the day, and the season. Without such knowledge, we may prepare the perfect dinner for the wrong set of people” (Krummenacher and Schmeidl 2001, PDF).

The rhetoric of conventional early warning systems labels local at-risk communities as the intended beneficiaries; but they are rarely included as end users of early warning activities.  To be sure, FAST never invited at-risk communities to the “dinner party” since the organizational challenges and financial costs of preparing the “perfect dinner” for the bottom billion are too bewildering. As a result, the invitee list quickly gets reduced to VIPs.

FAST’s use of the dinner party as an analogy clearly reveals the command-and-control mindset of conventional conflict early warning systems, also known as first-generation systems. Organizing the dinner is described as a centralized, almost egocentric activity: “we need to know,” so “we may prepare”. How about making it a pot-luck and use evite?

This, in essence, is the philosophy behind Ushahidi.

Ushahidi seeks to develop a more decentralized yet customized approach to throwing a dinner party. In conventional conflict early warning systems, blanket alerts are disseminated to an unknown guest list. Because the latter are never invited to dinner, there’s no way of telling whether the alerts were useful let alone received. (The field of advertising faces similar challenges, incidentally).

So why not let the end users decide for themselves what types of alerts to subscribe to? As Erik Hersman recently wrote in relation to early warning alerts,

“I don’t want to just get updates from random strangers in my locale. I want to only receive the ones that are “important” to me. I want to be notified when there is an emergency, major traffic jam or something else pertinent to me.”

Ushahidi is therefore developing a customized SMS/e-mail service alert option. Users will be able to specify what types of alerts they are interested in receiving and/or the particular location they want to receive alerts about.

This mindset is what makes Ushahidi different and why I call theirs a “bottom-bottom” approach. Needless to say, I’d rather attend Ushahidi’s dinner party over FAST’s.

Citizen Journalists Provide Better Early Warning

Are citizen journalists playing an increasingly important role in documenting violent conflict and human rights violations—particularly in terms of early warning? I chose Kenya as a case study to assess the role of citizen journalists in crisis environments as compared to the mainstream media. My colleagues Kate Brodock, Briana Kramer and I used event-data analysis to code reports of violent and peaceful events as documented by about a dozen citizen journalist bloggers between December 27, 2007 and January 27, 2008.

We did the same for mainstream media, ranging from print media (national newspapers) to radio and television programs. I also included the Ushahidi data because I wanted to carry out a three way comparison between mainstream news media, citizen journalism and a dedicated crowdsourcing platform.

We then created a Google Earth layer to visualize the data over time and space. Below is a YouTube video I created of the animation (for slower Internet connections). Here is the Google Earth layer (KMZ). The data can also be visualized on Google Maps here.

Yellow icon = mainstream news reports; Blue icon = citizen journalism blogs; Green icon = Ushahidi reports.

Our preliminary findings:

• Mainstream media reported actual death count before citizen journalists; however, on many accounts, mainstream media did not report on incidents leading to actual deaths, i.e., early warning signs;
• Citizen journalist reports and Ushahidi reports did not overlap geographically with mainstream media reports;
• Citizen journalists tended to report as soon as violence started, well before mainstream media;
• The number of comments on citizen journalist blogs increased during the 30-day period, or during particular periods of violence;
• The comment section was also used as a medium for real-time updating;
• Many citizen journalist bloggers used real-time updates sent to them via SMS, primarily from rural areas;
• Citizen journalism reports declined after the launch of Ushahidi;
• Ushahidi reports document an important number of violent events not reported by the mainstream media and citizen journalists;
• Contrary to news media and citizen journalist reports, Ushahidi data always had specific location information;
• Ushahidi reports also covered a wider geographical area than both mainstream news and citizen journalist bloggers.

For further information on our project’s methodology and sources, please see this short powerpoint presentation (PDF) which we have also uploaded on Slideshare. For more on crisis mapping, please see this page. For additional information on the role of digital technology during Kenya’s post election violence, see this narrative-based analysis (PDF) by my two colleagues Josh Goldstein and Juliana Rotich.

We look forward to receiving as much feedback as possible so we can improve our methodology and analysis in future case studies. We’re especially keen to have others interpret the dynamics displayed in the animation above. In the meantime, please contact me if you’d like to join the team and contribute to our next case study, which will be of Georgia.

Crowdsourcing Warning AND Response

I’ve spent much of the past two weeks hanging out with the Ushahidi team in Kenya and South Africa. When the team invited me to join their Board of Advisers last month, I was honored and gladly accepted because Ushahidi’s crowdsourcing crisis information approach is both innovative and promising. The project was figured in Kenya’s leading national newspaper, the Daily Nation, just yesterday.

During this week’s MobileActive conference in Jo’burg, Ushahidi’s Program Director, Juliana Rotich, conveyed to me the team’s strong interest in prioritizing early response after they release Ushahidi 2.0 next month. Juliana described the difficulty they had in convincing NGOs in Kenya to make use of Ushahidi during the post-election violence in order to map human rights abuses and share information. “We’ve got a major coordination problem when it comes to NGOs, not only for information collection but also response.”

I emphasized that the novelty of Ushahidi’s approach vis-a-vis humanitarian early warning is crowdsourcing; meaning I would not place emphasis on NGOs per se. One of the persistent problems with the field of conflict early warning and response is that those most in need of early warning, local at risk communities, seldom have the peer-to-peer, networked communication tools they need to warn each other.

I thus recommended that Ushahidi retain their decentralized approach and apply crowdsourcing to early response. Yes, crowdsource warning AND response. Of course, local decentralized response is not always effective, so warnings must include concrete recommendations for response. These recommendations can be based on already existing preparedness and contingency plans. Indeed, Kenya already had these plans in place to respond to expected violence during the elections, but the plans were not implemented by officials, let alone communicated to local at risk communities.