Calmer minds supposedly emerged from a one-day cooling period, and Cambodians took to the polls on July 28, 2013. 123 seats of the national assembly were contested in 24 provinces (called ‘constituencies’ in Cambodia). Voting day started at 7am, with lines already forming in many places before polls opened. In 19,009 polling stations throughout the country, voting concluded at 3:00pm, followed by the counting process and polling station results posted outside each station from as early as 6 pm in some placesAfter the opposition leader hastily withdrew an early letter in which he claimed the win for CNRP, Hun Sen’s CPP called their own preliminary results around 8pm. According to these CPP results, the ruling party won 68 seats, leaving 55 to the opposition. The mainstream media followed the CPP's lead, and later that night COMFREL concluded their Parallel Vote Tabulation with similar results (67 to 56 seats). Official tallies were at that time still being consolidated at the commune and later at the constituency (provincial) level.
I would like to leave it to more qualified observers to discuss the political implication of this huge opposition gain and the different scenarios that had already been speculated about before. While the opposition has already rejected the results, the coming days will show how things fall into place. Rather, I will focus this blog post on the diverse monitoring efforts set up before the election in a multilateral effort towards accurately depicting the conduct of the election and preventing election-day fraud.
The Situation Room
Cambodian civil society mobilized an impressive number of volunteer election monitors. The DHRAC had deployed 2,000 observers, the Khmer Youth Association (KYA) fielded 700 people, TIC/CISA provided 900 observers, joined by NICFEC,’s contingent of 3,000. The largest group was deployed by COMFREL with 10,000 short-term observers (STOs) and 250 long-term observers (LTOs). These combined ca. 15,000 observers could not cover all of the 19,009 polling stations, but there was still a considerable chance that any malpractice would be noticed.
In order to coordinate the diverse election monitoring efforts, COMFREL established a situation room on July 26th - a best practice they had adopted after a personnel exchange with several African observer NGOs. The aim of such a set-up is to serve as an information-sharing platform for diverse electoral stakeholders, facilitate real-time evidence-based analysis, and enable rapid response to problems of electoral process. In an ideal scenario, the latter goal is achieved by cooperating closely with election management bodies (EMBs) and security forces and by inviting them to join the situation room. While monitoring incoming reports from STOs or from citizens via phone and social media, and disseminating them to the public in regular press conferences, the information should also be shared on a continuous basis with decision-makers of state agencies with an appropriate mandate to take action. This model cuts short communication times and is meant to clear up what one could call “the fog of election day” (freely adopted after Clausewitz’s phrase "Fog of War", describing the difficulty of decision-making during conflict).
Source: author
Pioneered by PAFFREL and others, situation rooms have worked reasonably well in Sierra Leone or Nigeria, and the turnout for COMFREL’s effort was huge. A coalition of 30 different NGOs attended and shared their observations with the media and international organizations such as ANFREL, the EU, UNHCR, OSCE/ODIHR, La Francophonie and others. Yet, the concept did not come to full fruition in Cambodia due to several factors:
The motto “together in the situation room” was unfortunately misleading, because neither the NEC nor any other state agencies deemed it necessary to attend. The NEC instead set up its own press tent and briefed the mainstream media and an observation team from the International Conference of Asian Political Parties – which, by the way, glowingly endorsed the election after being invited by the CPP.
While this came as no surprise, given the generally dismissive and disinterested stance of the Cambodian EMB, nevertheless this crucial ingredient highlights the situation room's shortcomings. For instance, when it was revealed that the indelible ink employed for the election could be easily and completely washed off with household detergents, no NEC representatives were present to respond to the problem or even present a counter view. In fact, the NEC had invited political party representatives and observers for an official testing of the ink – only to be absent when the results were discussed. The situation room committee took great lengths not to antagonize the NEC too much, instead stressing that the information was meant to help the election administration address this crucial problem, and to increase the transparency of the whole process for voters.
Similarly, when bloated voter lists, over-production of ballots (about 27% more ballots than needed) and over-issuance of temporary Identification Certificates for Election (ICEs) to Vietnamese voters were alleged in the situation room, there was no reaction from the NEC.What can any observation effort realistically achieve under such adverse circumstances? It certainly highlights the fact that models like the “citizen co-production” or “co-governance” of elections as a public service is only possible when there is a ‘willing’ EMB, ready and able to participate. While the above factors put serious constraints on the functioning of the situation room as a direct link between observers (foreign and domestic) and the state it nevertheless did generate a platform of information-sharing among the watchdog organizations. Parallel joint debriefings of all involved observer groups also served this aim.
Video evidence on the (not so) indelible ink is presented (Source: author)
Although turnout was huge, and even increased significantly on election-day, Cambodian national TV stations were notably absent. Thus, the effort of disseminating election-relevant information to the public was more or less stopped in its tracks. Thai TV was present in force, as were some Khmer radio and print media. But there was a strange sense of preaching to the choir, since the gathered civil society members were overwhelmingly critical of the ruling party, and some media outlets could certainly be considered partisan. With no dissenting voice in the room to present a reaction from either the government, the NEC, or the armed services, the gathering became a rather one-sided affair. This became apparent when on July 26th an audience question challenged the CNRP’s anti-Vietnamese campaigning as hate speech and wanted it considered an election irregularity. Responses from the plenary were overwhelmingly dismissive, going so far as to agree with Hun Sen’s characterization as a “Vietnamese puppet”. This rhetoric had become so strong in the CNRP’s campaigning that the Vietnamese embassy felt compelled to comment about it.Vietnamese citizens being bussed in to cast their vote for the ruling party was a recurring theme, and it was difficult to discern xenophobia from genuine reports. In some instances, observers reported hundreds of Vietnamese being put up in a hotel in Phnom Penh or crossing the border in tourist buses. The derogatory term “yuon” was used widely in the observer circles as well as opposition media. It is certainly a plausible scenario that the voter list tampering together with the mass-issuance of ICEs would facilitate the addition of ‘foreign’ voters. Especially, since these last-minute documents are issued in lieu of other forms of identification by the commune chief - 98% of whom are CPP party members. But one would have to wait for a conclusive report to make a judgment call.
The absence of mainstream Cambodian media in itself was certainly a strong example of the strategic manipulations of vote choice so prevalent in this election.
In the absence of a ‘willing’ EMB, theory tells us that - at the least - monitoring should increase the likelihood of exposure and lead to increased international costs of cheating, because international actors will withdraw “democracy-contingent benefits” from regimes who engage in election fraud. Domestic costs should also be higher, because civil society organizations will use information about irregularities to increase leverage for social accountability, theoretically imposing reputational costs on cheating parties. Of course, for these mechanisms to work, monitoring groups need to generate accurate, timely, and reliable information. In other words, they need to clear up the fog surrounding election day.While the alliance of NGOs excelled at getting a plethora of anecdotal and personal evidence – sometimes through witnesses presenting the evidence themselves – there was a structural problem of how to systematically receive, prioritize, classify, and take action on incoming reports.
To the attending press and foreign observers it was not entirely clear, which of the reports constituted massive and widespread irregularities, and which reports were of more isolated incidents. This kind of analysis is of course a huge logistical task, since it requires a division of labor and many organizational units feeding into each other’s work. It should have nevertheless been the priority of the situation room to make sense of a huge stream of information and concisely funnel it to the public as a big-picture view. In a way, the situation room turned into a mishmash of disconnected evidence, perhaps adding more clouds to the fog of election day instead of clearing them. On the other hand, perhaps this is understandable in such a heated and time-sensitive environment as an election.
To be fair, an intervention team did evaluate available evidence according to whether it violated either election or general law, and it subsequently requested action from the local or national election commission, or the Ministry of Interior. In some instances, this cooperation with state agencies worked. For instance, there were several cases in Phnom Penh where companies tried to prevent their employees from taking Sundays off to go home to provinces to vote. They threatened to fire employees who left or stationed guards in front of employees’ housing. This occurred in the garment and electronics sectors. In all reported cases, the COMFREL intervention team notified the NEC and other state agencies. With the help of the Ministry of Labour and Vocational Training several cases were resolved, and employees were finally allowed to go upcountry to cast their ballot.
Crowdsourcing or crowdinforming?
COMFREL also invited the participation of the general public in the election watch effort. In an open call to the 'crowd', they opened channels to report all irregularities via the Cambodian Voter Voice (CVV) platform. Anybody could report observations via phone call, email, tweet, Facebook, or by filling out an online form. These reports would then be verified by COMFREL’s volunteer team to the extent possible and then displayed in an online map.
Call for participation via the CVV platform (Source: author; © Taylor Lowe)
The expected benefits of crowdsourced election monitoring include widening the information base for assessments of the quality of elections by exponentially increasing the number of eyes placed on the electoral process. By doing so, crowdsourced monitoring should increase the leverage of civil society and international actors, should they want to sanction the cheating party by withdrawing aid or mobilizing protest for example. Thus, the crucial component of this methodology is the engagement of organized civil society and the citizenry at large.
Hardly any empirical evaluations of these expected impacts exist at this time, though Michael Callen, James Long, and colleagues are doing some great work in this regard, as are Catie Bailard and Steven Livingston. Whatever the impacts might be, one would expect them to be strongly contingent on the level of buy-in from state actors, linkages, and alliances of civil society organizations, ICT penetration rates, provisions for data validation, and the level of ‘crowd’ participation in the platform.
One serious obstacle to the crowdsourced initiative was the government ban on photos or videos within 100 meters of polling stations. This seriously restricted the collection of vivid evidence, which Archon Fung cites as a major advantage of 'popular election monitoring'. Still, it was possible to take pictures of the final vote tally. Likewise, the presence of security personnel, campaigning during the cooling period, or other irregularities before election day could be documented, as in fact they were by the dozen via YouTube and Facebook.
However, the CVV failed to make use of the ‘crowd’ to increase its information base for a more basic reason. There are two general approaches to crowdsourced election monitoring: The open approach is a truly open call to anybody and everybody to report whatever they observe via different technical avenues. Here, data validation issues are the main concern. The way to address this issue are threefold: a) Large volunteer teams for manual validation and georeferencing, as in the case of the Uchaguzi platform in the 2013 Kenyan election; b) sophisticated automated systems based on natural language processing; or c) approaches that “crowdsource the filter” and employ user ratings, commenting, redundancy, and the included evidence to calculate 'veracity' scores. In addition to the data validation, outreach becomes a prime concern for such open systems.
CVV took neither of these routes, as it was primarily designed to be a crowdinforming platform. It used a bounded or closed approach, in which primarily COMFREL’s own long-term observers (LTOs) provided information. On the one hand, these reports didn’t have to be verified by such stringent measures, as they originated from already trained and vetted observers. In other words, CVV relied on a 'trust network' (as Lederach calls it) to ensure accurate reporting, and used the mapping function mainly as a tool to visualize these findings in the situation room and to the general public. This intended function was well-served by the platform, as frequent questions in the situation room showed, It also became apparent how important a core of verified and reliable reports is when informing the media about the extent of irregularities. An open system would inevitably be criticized of being not trustworthy and easy to manipulate.
On the other hand, the CVV platform did contain the call to the public from the very beginning. It simply was not promoted by any concise outreach strategy. When a cartoon which informed citizens about reporting options (see figure above) was published on July 25th, it received several hundred shares on Facebook within an hour, more than the 'delible ink' video in the same period of time. The desire to put scrutiny on the election and get involved in the observation effort was obviously present, at least among Facebook users. Another indication was the countless YouTube videos uploaded within seconds of irregularities. This was the case in a small-scale riot in the Stung Mean Chey area near the capital, where voters angrily demanded to observe the counting protest. When a local official hit a monk who was making a claim on behalf of the protesters, the crowd erupted into an angry mob, throwing stones at the polling station and burning a police car. Videos of the incident circulated on social media within seconds, almost as fast as the news reached the situation room via phone.
Thus, it might be necessary to distinguish between different goals for crowdsourced election monitoring: One is to create an accurate depiction of the conduct of the election. The second is greater civic participation or engagement with the election process that goes beyond putting one’s ballot in a box. A bounded approach is more suited for the first purpose, while an open approach might be better for the second one. Trade-offs are inevitable.
Source: Phnom Penh Post
In summary, my observations above about the election's observers might sound overly critical. But to be sure, given the difficult relationship with state agencies and antagonistic media and the systematic manipulations of the electoral process, COMFREL and others exhibited great professionalism. They employed a range of standard instruments of election observation, such as voter registry audits or parallel vote tabulation. If anything, the following lessons need to put into the context of these great achievements:
a) Electoral authoritarian regimes do not cooperate much with civil society monitoring efforts. A model of 'citizen co-production', i.e. the joint fixing of problems within the electoral process, is unlikely to work in such circumstances
b) Domestic monitors need to ensure the trustworthiness and timeliness of their data. They also need to develop systems by which to effectively aggregate, prioritize and effectively disseminate huge amounts of incoming data.
c) Crowdinforming systems are a good tool to funnel such data to the public. Crowdsourcing systems are a tool for increasing the information base but at the same time exacerbate the problems of trustworthiness and call for a huge outreach effort.
Max Grömping, 29 July 2013, Phnom Penh
