MobileActive.org

At MobileActive.org, we often write about mobile-based projects that other organizations and practitioners in the field carry out. We don't often highlight our our own mobile project implementations or discuss our own challenges and lessons, as many are sensitive in nature. Here, however, is a project we can talk about. 

As part of a USAID-funded project, MobileActive.org provides new media consulting to NGOs and independent media organizations in developing countries to enhance their communication and coordination efforts. We work in countries as diverse as Zimbabwe, Bosnia, and Peru, Egypt, Guatemala, and Serbia. Recently, we assisted an organization in Benin, West Africa, implement an SMS election observation project. 300+ trained observers took part in monitoring the presidential and legislative elections in March and April 2011.

We worked with the local observation organization FORS Elections to help the organization use new media and mobile technology to report from the polls on the election process. FORS Elections has been monitoring elections for 15 years but wanted to use mobile data collection as part of the observation this year. We used a new mapping platform, Citivox, and FrontlineSMS for the delivery of SMS messages from the polls. 

Training, a Website, and a Monitoring System

The work with FORS Elections consisted of three components:

• Building a website for the project.
• Training staff and data collectors on how to use new media to communicate with the public and other civil society organizations about the election.
• Coaching the organization in setting up and customizing FrontlineSMS and Citivox's mapping and engagement platform for the election-day mobile data collection at polling stations.

The website was done via a customized WordPress theme. We chose WordPress for its strong community presence in Benin, especially among French speaking users, and the documentation and support for those unfamiliar with it. Because of the limited time afforded to the project, using Wordpress allowed us to get a site up quickly.

Training focused on working with FORS Elections staff to develop communication goals, update and run the site, and, more specifically, how to use Twitter and Facebook for engagement with the public both inside and outside of the country. Facebook is popular among people in Benin who are online, while Twitter is not well known and used primarily by expatriates.

We worked with FORS Elections to target the right people and focus communication goals and messages, as well as to evaluate the metrics and analytics to see how people were interacting with FORS Elections online.

SMS Election Observation

We have previously written about systematic election observation in Nigeria, Lebanon, and elsewhere. We have also written about mobile data collection tools for election observation and the importance of not conflating citizen reporting with systematic election observation.

In Benin, the 2011 presidential election had been delayed twice over complaints the voter rolls were not accurate and a large number of voters had not been properly registered, and that preparations were not complete. The vote took place on March 13.

As a result, the political situation was, at times, confusing and chaotic with several postponements of the election day. While this made our work more complicated in some ways, it also allowed us to have more time to work with FORS Elections to set up the systems and train observers. 

In choosing a platform, we initially looked at Ushahidi to map the text messages reporting polling station incidences. We also looked closely at FrontlineSMS as a low-cost SMS messaging platform. We decided against Ushahidi because of its lack of fine-grained permissions that we needed for a variety of security and political reasons. We worked closely with Citivox, the mapping and incidence reporting platform we decided on, to customize its platform, localize it into French, and set up appropriate permissions and security protocols.    In summary, the Benin Matador project utilizes the following tools and platforms:

• We used FrontlineSMS as the SMS hub, installed and tested on 6 computers for incoming SMS and communications with the poll monitors and supervisors. 
• Citivox is a Software As Service (SAS) online mapping application.  
• GSM Modems: We used four USB GSM mobile broadband modems. We opted for HUAWEI E1550 models.
• SIM Cards: We used regular SIM cards used in any phone.  The system relied on one SIM card from each of the following networks in Benin: BBCOM, MTN and Moov.  
• Computers: All six computers met the requirements outlined on the FrontlineSMS website and connected to the Internet.  

For the trained observers, we compiled a list of coded questions. The observers used basic Nokia feature phones to send in reports via SMS using one of four phone numbers we had set up. We also used a public number available for the public at large to crowdsource reports. The other numbers were private for the observers. 

The general SMS observation system in place for FORS Elections worked as follows:

1. Citizens and observers will send an SMS to one of four numbers.
2. The SMS will be received via the SIM cards on the GSM modem connected to one of four computers with FrontlineSMS installed on them.
3. FrontlineSMS will read the SMS from the SIM card, display on the monitor and pass its content to Citivox via an API.
4. All SMS received by Citivox will be initially marked as private.
5. A team of trained volunteers with access to Citivox’s administrative interface will analyze the received incidence messages and make them public.

Issues with FrontlineSMS

FORS Elections chose to use FrontlineSMS to for receiving election observation reports, but we experienced significant problems with the application. As long as the Internet was up and running, Frontline was a very easy platform to use. It is also very user-friendly. But unfortunately, in Benin the Internet connections are spotty at times. As we experienced outages, FrontlineSMS had no way of tracking whicf messages were sent and which were not sent.

What exactly does this mean in terms of our project? During the presidential election, we received close to 3000 incidence and reporting messages but only about 500 messages made it to the Citivox website because we were not able to ascertain which messages had or had not been sent and there were no proper logs.  

We also experienced problems with FrontlineSMS in handling the text message volume. In short, we found that FrontlineSMS is not built to handle massive amounts of data at the same time effectively, Because of these two major issues, we do not recommend it in election reporting projects where a high volume of SMS are sent and mapped via an online platform (especially not in areas where there is intermittent Internet access.) 

Other Challenges: Time, Resources, and Staff

In every election monitoring project, there are limitations involving time, resources, and staff. There is, in essence, never enough time, enough resources, or enough qualified staff. In the case of Benin, we were able to add a week of work because of the potponement of the elections but the time spent was still extremely limited. 

We also needed more staff than we had available. For every response received from the trained observers, there was a necessary human element required in tagging each incoming report.

We didn’t have adequate time to set up Citivox entirely in the way we wanted to. The mapping application can be automated using keywords and tags, but this would have required more training and set-up time. Training staff was focused on the nuances of political communication as FORS Elections needed to stay neutral without supporting either the incumbent or the opposition. This was especially important because the election observation organization is still young, trying to gain exposure, and so every interaction on social media or online counted. 

Successes and Lessons

The biggest success of the project, surely, is the fact that the election, despite serious political differences, pre-election demonstrations, and small incidences of violence, was conducted largely peacefully -- even if the results were not initially accepted by the incumbent party involved. Unlike other election monitoring projects, the FORS Elections monitoring did not report on actual election results from the polls (often referreded to as parallel vote tabulation) but was focused on the conduct and process of the election and any observed voting irregularities. 

As far as lesson is concerned, it is important to put in place systems that are useful but user-friendly enough to be maintained by the local election monitoring organization after the outside trainers leave. We focused on leaving good practices, trained staff, and documentation in place in Benin - in short, we hope for greater capacity for the organization to run other similar projects in the future. 

As far as the technical problems were concerned, we would recommend relying less on sometimes unreliable infrastructure and hosting mission-critical components with mutlitple fail-safe backups in case the local infrastructure is unreliable or when there are network shutdowns. Eliminating single points of failure is critical.

We would recommend working with Citivox again, as the platform is extremely versatile, user-friendly, and localizable.  We would not recommend working with FrontlineSMS again when there is a high volume of inbound or outbound SMS in short time periods, and when there are API transactions involved.  We have seen other organizations have good experiences with using customized versions of RapidSMS (such as one of the largest SMS-reporting observations conducted to date in Nigeria) that held up well to large volume transactions.