COLOMBIA'S NATIONAL CIVIL REGISTRY LAUNCHES AN ANTIDISINFORMATION INITIATIVE, 2018−2019 SYNOPSIS When a wave of online misinformation jeopardized the integrity of primary elections in Colombia, Juan Carlos Galindo, who headed the country's National Civil Registry, decided it was time to address this emerging threat to democracy. The registry, which worked with the National Electoral Council, would soon conduct the first local elections since the country's 2016 peace agreements, and Galindo wanted to ensure that voters had correct information about the process, including the locations and open hours of polling stations. He asked his team to find appropriate ways to respond to misinformation, mindful of low public trust, frequent strategic use of disinformation by political parties, and limited resources to target voters at the local level. Building on the experience of the registry's Mexican counterpart, head of international partnerships Arianna Espinosa led the design and implementation of a plan to deal with the problem. The team struck deals with social media platforms, independent fact checkers, and political parties to take part in the fight against false information and used an artificial-intelligence-powered platform to detect and respond to false news about the election process during the campaign. By election day, the team had refuted a total of 21 misleading claims and published 59 verified news items and videos on social media, but the limited reach of the publications and minimal engagement with some of the key stakeholders prevented the registry from having the impact it aimed for. After the election, the new head of the registry refocused on building more-transparent processes and providing accessible information for citizens about elections while curtailing some of the initiatives Espinosa had introduced. This case is part of a series on combatting false information, including both misinformation (unintentional), disinformation (intentional), and fake news, one form of disinformation. Alexis Bernigaud drafted this case study based on interviews conducted with officials, journalists, and civic leaders in Colombia and Spain from January through May 2023. Case published July 2023. INTRODUCTION On March 11, 2018, as millions of Colombians were voting in primary elections to choose candidates for the scheduled May presidential contest, several polling stations in Bogotá, Medellín, and some other cities ran out of ballots. At one polling station, citizens shouted "Tarjetones, tarjetones!" (ballot papers!) as they were temporarily denied their right to vote. Juan Carlos Galindo, head of the country's Registraduría Nacional del Estado Civil1 (National Civil Registry), which was responsible for organizing primary elections in conjunction with the Consejo Nacional Electoral (National Electoral Council), activated a contingency plan to dispatch additional ballot papers to the polling stations (text box 1). However, rumors had already started to spread. During the primary election, in which voters would indicate whom they wanted to represent their party in the general election, citizens could choose candidates for one of two coalitions: one on the left, called Consulta Inclusión Social para la Paz (Consultation Social Inclusion for Peace), and one on the right, called Gran Consulta por Colombia (Great Consultation for Colombia). Each party had a separate ballot. But ballots had run out only for voters choosing the candidate for the right-wing opposition Democratic Center Party launched five years earlier by former president Álvaro Uribe Vélez, anchor for the conservative coalition Gran Consulta por Colombia. Some of the party's supporters claimed the registry was trying to prevent them from voting. For example, then senator and Democratic Center Party supporter Jaime Amín called it a "fraud."2 Outrage quickly spread on social media.3 Text Box 1: The National Civil Registry The Registraduría Nacional del Estado Civil (National Civil Registry) had three main missions: to record and store data related to vital events in the lives of citizens and residents, such as records of births, naturalizations, and deaths; to manage the distribution of identity documents; and to organize the country’s elections together with the Consejo Nacional Electoral (National Electoral Council), made up of nine magistrates elected by the Colombian Congress. The registry acted as the operational arm of the National Electoral Council. The council was in charge of registering parties, allocating public funding for campaigns, ruling on complaints about irregularities, and declaring official results. The registry played a more technical role and was in charge of overseeing organization of the electoral process, updating voter rolls, and overseeing vote counting and the dissemination of results. The registry had offices all around the country both at the departmental (district) level (there were 33 departmental registry offices) and at the local level (1,057 municipal, auxiliary, and special registry offices). The registrar, who was selected every four years through a competitive process, supervised the work of all registry offices from the institution’s Bogotá headquarters. 1 See https://www.registraduria.gov.co/IMG/pdf/20220712_informe-pacto-global_2021.pdf. Description drawn from International Foundation for Electoral Systems, Election FAQs: Colombia, May 29, 2022, accessed at https://www.ifes.org/tools-resources/faqs/elections-colombia-2022-presidential-elections and from https://www.registraduria.gov.co/IMG/pdf/20220712_informe-pacto-global_2021.pdf. Galindo, a former magistrate and law professor who was unaffiliated with any political party and who had served as registrar since late 2015, was personally targeted by attacks on social media. Memes and fake images suggested he was using his position for partisan purposes. After local registry staff at polling stations noticed the problem, they alerted the national headquarters, and Galindo and his team launched a counteroffensive to set the record straight via a press release, social media posts, and interviews in national newspapers. "It was very difficult to make the media and citizens understand that it was not fraud and that [the contingency plan] had been agreed upon in advance with all political parties and movements," Galindo recalled. Traditional media shared the registry's explanation widely during the days after the vote, but the false news and misinformation spread faster and reached more people than the verified information did. The registry did not try to contact social media platforms to delete or limit the spread of false statements. The spate of misinformation and disinformation surrounding the March 2018 ballot-paper incident was not an isolated event in Colombian politics. Misinformation (false or misleading content shared without harmful intent) as well as disinformation (falsehoods created in order to distort public understanding) proliferated during the 2016 referendum on the peace agreement between the government and the Fuerzas Armadas Revolucionarias de Colombia (FARC, or Revolutionary Armed Forces of Colombia) and then during the subsequent 2018 presidential elections. During the 2018 presidential campaign, in which the Democratic Center Party's candidate Ivan Duque prevailed over left-wing candidate Gustavo Petro and over a center coalition led by Sergio Fajardo, the left- and right-wing contenders and their supporters dispensed fake news as an element of political strategy.4 The Misión de Observación Electoral5 (Electoral Observation Mission), a civil society organization promoting civil and political rights in Colombia, found that purported informational videos were giving false guidance about election procedures, including claiming that citizens who had already voted in the first round of the election did not have to vote during the second round.6 Nongovernmental organizations had first mobilized to try to correct the problem in 2016. After the peace referendum that year, two civil society organizations launched their own fact-checking initiatives. Consejo de Redacción (Editorial Board), an association promoting investigative journalism, launched ColombiaCheck, a news website dedicated to fact checking, and La Silla Vacía (the Empty Chair), an online news site, created a fact-checking section called Detector de Mentiras (Lie Detector). Both websites checked information related to the electoral process. But the government itself had no coordinated strategy for the discouraging of false information about the electoral process. The registry had started using its social media accounts to share accurate information during the May 2018 presidential campaign. To respond to specific false claims in an ever-shifting context, however, the institution needed a more dynamic approach that engaged public institutions, media outlets, universities, political parties, social media platforms, and citizens themselves, all of whom had crucial roles to play in promoting information integrity. THE CHALLENGE With the approach of the October 2019 regional elections to select governors, department assembly members, mayors, municipal councils, and local administrative boards, the civil registry had to act fast to deliver a more comprehensive-and effective-approach. But Galindo's team faced several hurdles, including an ambiguous mandate, weak public trust in government, limited capacity at local registry offices, the ubiquitous use of disinformation by political parties for political advantage, and the need to craft new relationships with major social media companies. At the time, several different parts of the Colombian government were beginning to take responsibility for addressing false information within their specific domains. Galindo's team reached out to the president's communications staff and the Ministry of Technology and Communications to ask for their views about how to respond to election-related false information but there was no reply. Not everyone thought the registry had a role to play in addressing the spread of false statements that might affect voting integrity. During elections, the civil registry's responsibilities included guaranteeing "the organization and transparency of the electoral process" and "the reliability of vote counts and electoral results."7 The legal mandate did not identify a specific role in combating false news. Although this kind of role could be construed as essential to accomplishment of the registry's election obligations, Galindo said he had to consider the impact of a registry-led antidisinformation campaign on public opinion in a country in which, according to the 2019 Gallup World Poll, only 27% of citizens said they trusted government.8 A wary public could easily perceive any initiative that might appear to affect free speech as an attempt to influence elections-especially if the law did not explicitly grant the institution such authority. Citizen distrust of government included distrust of the registry despite the fact that the registry benefited from administrative, contractual, and budgetary autonomy and that Galindo, the registrar, was chosen not by a political leader but by the presidents of the Constitutional Court, the Supreme Court, and the Council of State.9 In 2019, only 30.7% of citizens told canvassers for the national department of statistics they believed the registry behaved impartially, even though that level of trust made it the second-best-rated institution in the country-just below the military.10 Luis Felipe Murcia, a team leader in the registry's department of international relations in 2019, acknowledged the issue. "In people's minds, elections are rigged, and the registry is not doing things right," he said. According to the Karisma Foundation, a civil society organization working on human rights and technology, the distrust was fueled largely by lack of transparency in the voter count software the registry used.11 Those perceptions weren't the only potential hurdles. Any effective strategy for identifying and countering disinformation would require the registry to develop new administrative capacity, including staff, skills, and resources that could identify and respond to problems at the local level. If the institution itself lacked the resources to counter misinformation and disinformation, it would have to find partners that could do that job. The effectiveness of fact-checking media such as La Silla Vacía and ColombiaCheck in curbing misinformation was limited by the audience the organizations reached. In a 2018 article, Oscar Palma, professor of political science at Bogotá's Rosario University, described these media initiatives as "getting through only to a special, educated segment of the Colombian population." He said the organizations' impact was limited because rural populations did not read their articles.12 In order to make a difference, the registry would have to ensure that its fact-checking work reached a broad segment of the population and that readers would view the work as both accurate and reliable. In particular, communicating with people in remote parts of the country presented steep challenges. Some locations were media deserts, and many people relied on information conveyed through WhatsApp groups, which the registry could neither monitor nor use to present correct instructions about how to cast a ballot. And even though the registry had local offices in cities and each of the country's 32 departments, these offices did not have an established role in monitoring disinformation or in responding to false news. Because many political candidates had used disinformation as a strategy, getting them to join the fight presented an additional problem. In 2018, the presidential candidates had signed a pact against violence and disinformation, but political parties still used disinformation in their presidential campaigns.13 If one did not trust the others' commitment to refrain from disinformation practices, the pact would not hold, and indeed, it appeared already to have failed as the 2019 elections approached. The electoral council, working with the registry, would have to broker a new deal. Finally, it was crucial to recruit the help of social media companies such as Facebook and Twitter because election-related falsehoods and misinformation propagated most often on those platforms. FRAMING A RESPONSE At the end of 2018, Galindo launched a conversation within the registry to determine whether fighting disinformation was part of its mandate. During this preliminary phase, Galindo relied on Arianna Espinosa, a former humanitarian response professional he had hired in January 2017 as an adviser on international partnerships. After only six months on the job, Espinosa had a record of innovation and achievement that had prompted Galindo to name her head of cooperation and international partnerships and to turn to her for help in assessing how best to address misinformation and disinformation. Because the registry's communications department was often on the front line of false information on social media, Galindo also appointed Silvia Hoyos, the registry's communications director, to colead the effort. Espinosa first had to assess the scope of the registry's powers for responding to misinformation. "[On paper], that was not part of our mandate," said Espinosa. "But society evolves faster than the law, and disinformation represented a growing security risk for the electoral process." The law gave the registry the responsibility to "safeguard the transparency of the electoral process [and the] trustworthiness of electoral results" and to "protect the exercise of the right to vote."14 To protect the vote against disinformation, the team considered whether the registry should share verified information and contradict false information related to the electoral process. Establishing the basic parameters for an antidisinformation program was essential. Espinosa and communications director Hoyos started looking for examples of other electoral bodies that had launched antidisinformation initiatives. Although they studied the experiences of such countries as France and the United States, they focused most of their inquiry on nearby countries in Latin America, where previous contacts made collaboration easier; and they identified Mexico as a prototype. At the beginning of 2019, they reached out to Mexico's Instituto Nacional Electoral (National Electoral Institute), an autonomous public agency responsible for organizing federal elections. They next organized a trip to learn more about the initiative the institute had launched for Mexico's 2018 general elections. Officials at Mexico's National Electoral Institute introduced the Colombian registry team to a project called Certeza, which enabled the institute to fight online misinformation on election day. That 2018 initiative had several elements. * Detection. The institute used social media listening tools-software that could analyze online conversations about a topic or event-to monitor Facebook and Twitter and extract posts that mentioned certain topics or words, thereby expediting identification of potentially false information. The institute's technological processes department also had a platform that created stacks of items for staff members to review-a step that helped organize the work process. * Verification. Next, a team analyzed and checked the information the social media listening tools generated. The group sometimes asked local staff members to investigate in person to ensure it was correctly understanding context and content. The team then focused on the subset of statements that contained misinformation or disinformation about the electoral process, flagging the statements for attention. * Response. The team defined the type of response each item merited: (1) direct answers to authors of tweets, (2) images showing why the information was false, or (3) informational videos or graphics, used most often when social media users frequently misunderstood a topic. The institute also created partnerships with media outlets, which shared the verified information the institute produced. Mexico's institute further entered into several agreements with social media platforms. For example, Facebook pledged to launch civic engagement tools such as an Informed Reader button, which appeared the day before the vote and enabled Facebook users to access a page on which the institute had put information related to the voting process.15 The institute and Facebook worked together to create training materials to help journalists understand the electoral process and learn to detect election-related fake news. On election day, the institute provided a physical space within its offices for Facebook employees to work and share real-time information through live streams. For its part, Twitter (1) trained institute staff to use its live video Periscope tool (retired in 2021), (2) broadcast the presidential debates, and (3) shared the institute's information on how to vote through its accounts TwitterLatAm and TwitterGov. Google broadcast the presidential debates on YouTube and promoted accurate information about the electoral process on the search engine's main page. Espinosa said she drew four main lessons learned from Mexico's experience. * Avoid the appearance of partisanship, refrain from responding to disinformation attacking specific candidates, and focus only on the electoral process. * Address electoral process disinformation across a broad front, not only disinformation that targeted the electoral management body. * Forge alliances with media outlets that can produce verified information, and help share the registry's messages with a broader public. * Craft memorandums of understanding with social media platforms to ensure the social media companies commit to fighting disinformation; Notably, the registry's efforts would remain distinct from other Colombian government efforts to combat disinformation. The Colombian police in 2001 had created a cybercrime unit that became a Police Cyber Center within the Criminal Investigation Directorate. The unit dealt with instances of non-election-related disinformation. (It would later use open-source intelligence to help fight fake news related to the 2020 COVID-19 pandemic, but it had no role in the 2019 election.) Needing a capable technology partner to help formulate and implement a strategy against disinformation, the registry turned to Indra-Minsait, a Spanish company that provided digital services to manage the Colombian electoral process. Indra-Minsait had been in charge of consolidating results-registering, updating, and checking vote totals from polling places-for the 2011 and 2015 regional elections as well as the legislative and presidential elections in 2014 and 2018. Indra-Minsait had prior experience in Europe, where it had developed measures to protect IT systems from cyberattacks and fought disinformation. Building on discussions with Galindo and his team, the director of electoral solutions at the company oversaw the design of a plan for detecting and combating disinformation in Colombia and presented it in March 2019. The proposed plan consisted of: * Launching a survey to better understand patterns and trends in disinformation * Developing an artificial intelligence (AI) platform to identify false information that could affect the electoral process, thereby reducing demand for staff time and enabling fact checkers to prioritize high-risk items * Collaborating with independent fact checkers who could share fact-checked articles they had published, which Indra-Minsait could then use to refine its detection algorithm and identify other probable disinformation for verification * Securing agreements with social media companies so as to take affirmative steps in the fight against disinformation * Evaluating the initiative's performance and sharing with citizens the lessons learned In addition, Galindo suggested the idea of a pact against disinformation-a pact that all political parties would commit to. Unlike Sweden, which protected its 2018 general election against foreign disinformation by providing local-level election authorities with tools to understand the threat and methods for responding,16 Galindo and his team did not try to harness the registry's local offices to develop a decentralized strategy against misinformation. "The strategy had to be implemented by the central office so as to avoid inconsistencies between branches and provide standardized responses. Employees from the registry at the local level were not allowed to confirm or contradict news," Galindo said. During the early months of 2019, the registry adopted the plan, blended the plan's components with other initiatives already in progress, and pushed hard to implement all of the elements in time to protect the integrity of the 2019 regional elections. GETTING DOWN TO WORK With only a few months remaining before the start of the campaign period, Galindo and his team set to work. They cocreated a platform to provide information for citizens about the electoral process; they tried to learn more about how disinformation circulated among citizens and about the public's level of concern; they partnered with social media platforms, independent fact checkers, and universities to identify and respond to false information; they engaged politicians in a renewed resolve to forgo the use of disinformation as a campaign strategy; they worked with a consulting firm to develop a platform capable of detecting potentially false information; and they produced and disseminated verified information to the public throughout the campaign and on voting day. They also sought to build-and protect-a reputation for independence and neutrality. "We didn't want people to associate the registry with political leaders," said Murcia, one of the registry's staff members. Creating a comprehensive information system Even before Indra-Minsait's plan against disinformation was ready, Galindo's team had started to create an online platform that would centralize control of information about the electoral process. "To fight misinformation, before anything else, you have to provide information," Galindo said. "Back then, information on preparation and implementation of the electoral process was spread across the websites of a variety of institutions such as the Office of the Inspector General, the police, the Office of the Comptroller General, and the Ministry of the Interior, so I thought it would be easier to have a single system whereby the same message could get through to all users." After initial talks with the heads of the institutions, the registry held a series of meetings, beginning in February 2019 to create an election-focused portal on the registry's website and select the content to post online. "It was pretty easy to sell the idea to other institutions because of the benefits of a unified source of information, so all of them assigned a few staff members to the team, and we managed to get the new system started in a short time," said Galindo. Staff from the National Electoral Council, the Ministry of the Interior, the police, the National Disability Council, and other government institutions formed 14 working groups that created materials to provide the general public with information on the vote and developed guidelines for people who played important roles in the electoral process, such as security guards and police, poll workers, party poll watchers (called electoral witnesses),17 candidates, and authorized observers.18 The registry launched the portal, named Sistema Integral de Capacitación Electoral (SICE, or Comprehensive Electoral Training System), in April 2019. Exploring the public's view of disinformation Following the example of the European Commission, Indra-Minsait's first step was to better understand disinformation in the country-especially what fueled its use and its spread. Both the registry and Indra-Minsait worked with Spanish consulting firm GAD3 to design a survey, which aimed to understand citizens' level of exposure to fake news, the extent to which the public perceived that its own behavior contributed to the problem, and the public's level of concern. The survey adapted questions drawn from the Eurobarometer, a biannual European Union-wide survey implemented by the European Commission since 1973, which showed in 2019 that a majority of Europeans felt they "often came across news or information that they believed misrepresented reality" and that the existence of such news was "a problem for democracy in general."19 GAD3 interviewed 1,216 Colombian voters, contacted on the voters' mobile phones, devices used by most residents. The number of respondents in each province was proportional to the province's population and representative-at the national level-of age and gender, though the firm used quota samples in each area instead of drawing a random sample of respondents. The company implemented the survey at the end of July, and the results highlighted the prevalence of disinformation in the country: for example, 39% of respondents declared they encountered fake news on a daily basis, and 88% identified fake news as a threat to democracy.20 The survey showed that the peace agreement and the activities of government attracted more fake news than other topics did. The survey also helped the registry identify three elements that made the Colombian population vulnerable to fake news. * The lack of awareness on the parts of voters who did not understand disinformation and its electoral consequences * The lack of commitment by citizens who knew about fake news but did not believe they had a role to play in fighting it * The overconfidence of certain voters who understood how social media worked but sometimes believed disinformation because it aligned with their own political biases Soliciting help from online platforms Drawing on the experience of Mexico as well as the example of the European Commission's 2018 code of practice on disinformation,21 the registry sought commitments from online platforms Facebook, Google, and Twitter to help combat fake news. "We wanted to sign memorandums of understanding through which the firms would commit to fight fake news by creating bots that would detect potential disinformation and by putting banners under potentially false content," said Espinosa, who added that, "blocking content would have infringed free speech." Twitter, Facebook, and Google signed such agreements with the registry, thereby expressing their intention to help mitigate disinformation and the spread of fake news. But the memorandums were very limited in scope and varied by corporation. Facebook committed to identify people who paid for political ads on its platform; Twitter opted to promote what it termed relevant information from all the actors during elections; and Google said it would collaborate with authorities to combat the dissemination of false news.22 Unlike Mexico's electoral institute, which convinced Facebook to prioritize its own requests to flag content as false on election day, the registry was unable to secure commitments to specific actions to protect the election against misinformation. "The arrangement was not as strong as we originally wanted because they did not want to sign anything more, but it was a first step toward raising awareness via social media platforms in Colombia," said Galindo. Building capacity In early 2019, the registry formed its own fact-checking team, which focused on verifying information related to the election process and disseminating fact-checked content. Because the registry team remained small, with no additional staff members, it had to create a work process that would enable it to address the constant influx of suspected disinformation. "At first, we didn't really have a process; we had meetings from time to time," said Espinosa. The registry sought outside help in order to expand the scale and scope of this work, but in an atmosphere of low trust, it had to settle for a roundabout, indirect collaboration with nongovernmental organizations. Galindo and his team were inspired by the Mexican electoral institute, which worked with a coalition of media outlets that agreed to share verified information the institute provided on election day, but they pursued a different approach they considered more appropriate for Colombia. For example, because Colombian media outlets tended to have partisan affiliations, partnering with them risked politicizing the initiative. A coalition on the Mexican model was inadvisable under those circumstances, they suggested. Instead, Galindo and his team approached the country's two main fact-checking organizations-ColombiaCheck and La Silla Vacía-both of which were perceived as nonpartisan and which already routinely posted fact-checked articles online. "We organized meetings to discuss our strategy with them. We had to build trust, so we told them they would be completely independent and could decide whether and how they would like to contribute," Espinosa recalled. The registry succeeded in getting help from both organizations. The outlets agreed to share with Indra-Minsait lists of the fact-checked articles they had published online. Indra-Minsait could then use these posts to identify hot topics and specific words associated with fake news in order to develop an algorithm to detect other probable disinformation. Using the algorithm, Indra-Minsait could then flag likely false stories to the two organizations and to registry staff for investigation. Indra-Minsait's previous work with Spanish members of the International Fact-Checking Network,23 a global initiative whose members included ColombiaCheck and La Silla Vacía, created a basis for trust and made collaboration feasible. The registry also tried but failed to convince ColombiaCheck and La Silla Vacía to participate directly in its campaign against disinformation. La Silla Vacía, which had very few staff members at the time, told the registry that, given its limited resources, it was willing to participate only if the level of commitment was low. For ColombiaCheck, collaborating with a public entity was an issue. "At first, the registry told us about its idea of having us as 'official fact-checkers' for the campaign, but we said our policy was not to enter into contracts with government institutions," said Pablo Medina, director of ColombiaCheck in 2019. Although neither of the two news websites wanted to be the project's official fact checkers, they were open to assisting through a third party such as Indra-Minsait and agreed to fact-check some of the potential disinformation flagged by registry staff while reserving the right to choose which cases to investigate.24 Finally, the registry offered to make available to ColombiaCheck and La Silla Vacía the AI-powered detection platform Indra-Minsait was building. "[The idea was that] having a better-informed public debate during the campaign would help foster trust in the electoral process. So even if our staff focused only on news related to the registry and the voting process, we wanted to help them fact-check the campaign in general in order to create an atmosphere of trust," said Espinosa. Both of the fact-checking organizations signed agreements that enabled them to use the platform. However, to underscore their independence, they also refused to commit to use the system in their work or to check news shared by the registry. "We wanted to remain independent and free to check the publications or not," said Juan Esteban Lewin, director of La Silla Vacía in 2019. ColombiaCheck agreed to organize training sessions to help the registry develop its fact-checking methodology and improve skills. The registry was inspired by ColombiaCheck's categorization method and created its own system to categorize information as accurate (green), disputable (yellow), or false (red). "They shared their knowledge on how to decide whether a piece of information was false or just debatable. They really helped us strengthen that part of the strategy," said Espinosa. During the campaign, the registry would use those labels to alert citizens and provide verified information to counter false claims. Universities, too, were possible sources of assistance, but building scholarly collaboration proved challenging. For instance, Pontificia Universidad Javeriana, one of the country's oldest and most prestigious higher-education institutions, had faculty members who were potentially interested in helping to combat fake news. In July, Espinosa met with representatives of the university's political science and communication departments to discuss a partnership. Espinosa and her team hoped that professors and students would participate in the fact-checking process on a voluntary basis, thereby expanding the scale of the work the registry and the two national fact-checking organizations had undertaken. But university representatives refused to sign any agreement involving the use of unpaid student labor that did not contribute to training. Instead, the universities opted to work with civil society fact-checking alliance RedCheq, which paid university professors to train students and coordinate their work. (See ISS companion study on RedCheq.) The registry lacked the budget to invest in similar, additional arrangements. Although Javeriana did not take part in the production and dissemination of fact-checked articles, researchers from the university did help the registry analyze the results of its survey on misinformation. The institution also agreed to host a launch event for the registry's antidisinformation strategy, planned for August. Trying to get politicians on board Galindo very much wanted to persuade all political parties in Colombia to disavow-formally and in public-the use of disinformation during the campaign, counting on the symbolic power of such an event. From May to August, Jaime Hernando Suárez, delegate registrar on electoral matters, held a series of meetings to explain the registry's strategy to representatives from all political parties and to persuade them to sign a pact the registry drafted. On August 21, 2019, Galindo and Suárez managed to bring together representatives from 17 political parties as well as then president Ivan Duque at Javeriana University in Bogotá, where they publicly signed an accord pledging not to share false news during the campaign. In particular, parties promised to promote the diffusion of information on the electoral process as part of their candidates' campaigns so as to avoid producing or promoting false information targeting their political opponents and to encourage their candidates to publicly acknowledge any instances of disinformation and cooperate in remedying the problems. That agreement, which the registry called the Ethical Pact for Information, contained no enforcement mechanisms, and Galindo called it "a symbol, more than anything, so that, if we witnessed disinformation during the campaign, we could say, 'Look at what is happening despite the agreement.'" The team hoped the parties' public commitment would disincentivize the use of false information during the campaign. "We had no way to know whether or not they would [keep their promise], but we put it on the agenda of the leaders of every political party," added Espinosa. Launching the campaign and the detection platform Galindo officially launched the antidisinformation initiative under the hashtag #VerdadElecciones2019 in mid-August, a little more than two months before the regional and municipal elections. The social media hashtag enabled the registry to send messages to raise awareness of disinformation and misinformation. The registry created infographics and uploaded videos on Facebook, Instagram, and Twitter to explain the electoral process in simple terms and to address common misconceptions. In September, upon completion of the tendering process, Indra-Minsait was officially awarded a contract to provide IT services aimed at preventing and monitoring events that could put the security of the electoral process at risk. The deliverables included the creation of an AI-powered platform to detect false news susceptible to become viral and influence the vote or affect the reputation of the registry.25 Once it was selected, the company used the fact checkers' data to develop and refine algorithms that could identify potential disinformation. It retrained its system with news the fact checkers had checked and categorized as true, false, or disputable. Employing deep-learning techniques, it categorized posts and news articles, computed a risk index, and flagged content that should be given priority by fact checkers. The AI-based detection platform was geared to measure two risk factors. The first was the likelihood of a social media post being false based on its similarity with other content previously identified as false or misleading. The second factor was virality, the potential number of users that could be affected by a social media item.26 Users of the platform, including registry staff and journalists from La Silla Vacía and ColombiaCheck, could access the results of this detection process on a webpage that listed social media content ranked by the system's computed risk index. Several dashboards displayed information on the accounts, issues, and regions most affected by disinformation. The website also enabled users to manage workflow by assigning content to specific members of staff. Fact-checking the campaign In the lead-up to the October 27, 2019, election, registry staff focused daily on identifying potentially false content, selecting claims that needed to be fact-checked, verifying the accuracy of claims, and publishing verified information through a variety of platforms. "We assigned people from the communications team and the international relations team to carry out these functions during the three months leading up to the election. We were also in touch with someone from the electoral team, who helped us draft content on technical aspects of our work," said Espinosa. "As elections approached, we had to organize more-frequent meetings," Espinosa added. "We started working during office hours, but in the last days before the vote, we also had to work after office hours." In the detection phase, the group primarily used the AI platform, which prioritized posts that could affect the election. In addition, staff members sometimes received chain messages on their personal WhatsApp accounts that contained false information, and they included the messages in the list of items to fact-check. In rare cases, local registry workers reached out to the registry to inform the registry that fake news was spreading at the local level. "We didn't put a formal strategy in place for that to happen, but because the registry's antimisinformation strategy was publicized, they sometimes took that initiative," said Espinosa. Citizens could use the registry's hashtag to ask the registry to check specific items they suspected to be problematic.27 In a second phase, the team discussed whether or not to react to the news. If the news threatened the election process and was shared by a high number of users, Espinosa and her team tried to identify sources to verify the claims. In the verification phase, the team looked for evidence to support or contradict the claims. Team members sometimes reached out to other departments within the registry or to employees in local registry offices who could collect data on the ground. "On election day, we sometimes called people working at polling stations. We needed the information fast because we knew that contradicting false information too late would be irrelevant," Espinosa recalled. And even though that process enabled the team to successfully gather information in some cases, communication with local staff proved to be difficult in some instances. "It was not their job to work on fake news; they were trying to make sure the voting process went smoothly in their area of competence, so they didn't have a lot of time to help us," Espinosa said. Building on the information gathered, Espinosa and her team categorized the information as "accurate," "false," or "disputable"-three categories inspired by the ones used by ColombiaCheck. Finally, the registry produced verified information and shared it with the public. In most instances, the registry created visuals showing screenshots of misleading posts along with short paragraphs explaining why the information was false or disputable. But the team also tried to develop other ways to set the record straight. "We had to decide how we could best refute fake news-for example, by using memes or videos on YouTube for certain tweets; and we also had to decide whether or not the registry should get involved," said Espinosa. "We had to think of formats that would actually reach people and interest them. On the communications team, we had someone who would take pictures; we had a designer; and we had someone who could make videos." Throughout the campaign, the team produced short, fast-paced videos in which Espinosa explained specific elements of the voting process. In one video, for example, Espinosa demonstrated the use of ballpoint pens-available at polling stations-in order to show that, contrary to some claims, the ink could not be erased to falsify results. The team incorporated sound effects, references to memes, and humor into the videos to make them more engaging. The registry used SICE-the electoral information portal launched in April-to post the antidisinformation content its staff developed. The team also shared verified information by means of its Facebook, Instagram, Twitter, and YouTube accounts. In addition, on an ad hoc basis, the registry contacted local newspapers when news was relevant for specific areas of the country so that the newspaper outlets could share the verified information. "When the news was very local, they sometimes paid attention to it," Espinosa recalled, though she said local media did not often respond. Fact-checking organizations La Silla Vacía and ColombiaCheck sometimes used registry information to help debunk misinformation and disinformation. For example, in October 2019, as election day approached, ColombiaCheck used a fact-checking publication from the registry as a source for its article countering a false quote from a former leader of the Liberal Party about the registry's alleged role in rigging elections,28 as well as a post in which the registry contradicted a senator who claimed that the location of polling stations in the city of Buenaventura had changed.29 A busy election day To respond quickly to problems that arose on election day, Galindo had introduced months earlier a new mobile application that enabled registry representatives at polling stations, departmental registrars, municipal registrars, and employees from headquarters to communicate when incidents occurred. The mobile application sought to avoid the delayed responses to polling center problems that had marred the March 2018 election. Galindo's goal was not to engage registry staff on the ground to flag disinformation at a local level but, rather, to flag issues related to the voting process so that the registry could intervene before issues could generate controversy on social media. The registry's IT department had led the development of a smartphone-based mobile application that enabled registry representatives at polling stations to confirm, for instance, that the polling station had opened on time and that electoral witnesses had arrived or to flag problems related to the voting process. On election day itself, the registry team swung into high gear. "We were constantly working on contradicting false information," said Murcia. "For instance, we would read a post saying that people were being denied their right to vote at a polling station in Arauca, so we would look for information on what was happening there and then explain that only people with a valid ID could vote. We spent the whole day trying to clarify these types of issues." Examples of fake news the registry addressed on election day included (1) a claim by then senator and 2018 presidential candidate Gustavo Petro that 900 electoral witnesses from his coalition Colombia Humana had been deleted from the list by the registry,30 (2) a fake official document announcing that an anticorruption consultation was taking place on the same day as the regional elections,31 (3) a tweet by a mayoral candidate in Bogotá claiming that a power outage was affecting the transmission of election data in the city,32 and (4) sharing of incorrect information on the voting process to elect members of Santa Marta's district council.33 Galindo's effort to streamline communication with local offices enjoyed only mixed success. Some polling station workers struggled to use the app despite the advance circulation of training materials, and some stations chose not to provide information through the app during voting. However, local registry staff succeeded in using the app to alert the head office to instances in which police refused to let people into a polling place with their children or cases in which electoral witnesses failed to follow rules. Galindo said the app had a positive impact on the smooth conduct of the election and helped minimize risks, including misinformation. OVERCOMING OBSTACLES Although the registry team had hoped that La Silla Vacía and ColombiaCheck would take active roles in the fact-checking process and would use the resources the registry put at their disposal, the two fact-checking organizations rarely used the AI-powered detection platform during the campaign. "We realized that the platform was not that helpful to us. It would show us articles that were very general and not checkable, often from big newspapers," said Medina, past director of ColombiaCheck. "We ended up checking only one or two articles that came from the platform." In parallel, both of the fact-checking organizations reviewed fake news identified through their standard operating procedures and published articles with corrections on their web pages, but much of their work, which was subject to staffing limitations, continued focusing on issues unrelated to the voting process. According to Lewin of La Silla Vacía, the AI-powered platform did not enhance capacity for several reasons: (1) It flagged primarily articles that his team had already identified using other sources of information. "With the limited staff that we had-and given the never-ending avalanche of content on social media, it did not really help us," he said. (2) Because both ColombiaCheck and La Silla Vacía worked with Facebook's Third-Party Fact-Checking Program,34 they had access to CrowdTangle, a social monitoring tool they thought worked better for them than Indra-Minsait's platform for analysis of trends on Facebook. (3) For Lewin, the amount of time needed to familiarize himself with a new platform was also a major barrier. "From my experience with CrowdTangle, it takes about six or eight months to understand how it works and the logic behind it. After a while, you understand the types of content that the algorithm is more likely to select," he said. The heavy workload he had and the imminence of the election therefore made it impossible for his team to benefit from the tool. "If we had had more capacity, we would have probably used the platform more," he said. Registry staff regularly shared with ColombiaCheck and La Silla Vacía the cases of misinformation it had detected, but the two websites drafted articles in only a few instances. "They would send us information once or twice a week via WhatsApp. We checked some of them-mainly the ones that had nationwide impact-but some of the others were too local, so we didn't have enough resources to spend time on them," said ColombiaCheck director Medina. The national focus of the two fact-checking organizations was an obstacle to successful cooperation when disinformation was local. ASSESSING RESULTS #VerdadElecciones2019 was Colombia's first attempt at creating a comprehensive strategy against election disinformation, but it remained a small-scale initiative. Although it enjoyed some success, those involved in the program acknowledged shortcomings. During the campaign, which ran from July 27 to October 26, 2019, the team checked 31 news items and refuted 21 of them. In total, it published 59 verified news items and videos on social networks and on the electoral information portal called SICE. * SICE attracted 800,000 individual visits. Although no comprehensive report evaluated the total reach of the initiative on social media, available indicators pointed to a very limited audience in a country with 40 million potential voters.35 * On Twitter, videos posted by the registry reached 1,000 to 3,000 views and were shared by only a few dozen individual users. The registry's posts were almost never shared by high-profile accounts, which could have increased their reach. * Awareness of the registry's program itself was limited even among Colombian academic researchers working on related subjects. For example, Cristina Vélez, who cofounded nonprofit organization Linterna Verde, which analyzed the dynamics of public speech on social media and their implications for social and political spheres, said in 2023 that she had not heard of the initiative. On election day, October 27, the registry's AI-powered platform analyzed 90,000 pieces of content published on social media and categorized them according to the pieces' ability to affect the electoral process. Registry staff reviewed a total of 18 social media posts out of 90,000. Although Indra-Minsait highlighted the fact that hashtag #VerdadElecciones2019 was a trending topic on Twitter in Colombia at the beginning of election day, the verified information published by the registry was substantially less likely to be shared and liked by users on Twitter than the original, misleading publications were.36 Alejandro Vergara, director of technological processes at Mexico's National Electoral Institute, said a limited reach did not mean that publishing responses to fake news was useless. "Even if it doesn't reach many people, the fact that a post setting the record straight was published and that it can be found on the Internet has value," he said. "In the near future, a lot of information will come from AI-powered tools, so it is important that these can be fed with correct information." Despite the registry's low number of publications and limited reach, members of the team said they considered it a successful step forward. "I think it has been a success, and it was groundbreaking," said Espinosa. "I think the strategy was effective; we felt [disinformation] did not affect the electoral process or the institution in 2019," Galindo added. Galindo hypothesized that the pact signed by political parties and coalitions also helped reduce the amount of disinformation related to the electoral process. No political parties claimed that the election results were manipulated. REFLECTIONS The people most closely engaged in #VerdadElecciones2019 drew on the experience to consider what they might do differently to improve information integrity in future elections. Addressing the problem of election misinformation and disinformation at scale presented a big challenge for the future. "We realized it was a much bigger issue than we thought. Bigger than the solution that we were offering," said Luis Felipe Murcia, a team leader in the National Civil Registry's department of international relations. Future iterations of the initiative could try to strengthen information integrity in election-related public discourse by 1) reaching more people with accurate information, 2) focusing its effort on specific demographics at high risk of receiving and believing misinformation, 3) enlisting more organizations in fact-checking and dissemination of accurate information, and/or 4) enhancing the public's capacity to scrutinize statements and do its own quality control. It was possible to reach more people if social media firms and major media stepped up to help. According to Juan Carlos Galindo, who headed the registry, social media companies had a major role to play. "We have to keep pushing for a real commitment from tech companies to the fight against disinformation," he said. Murcia said the decision to work only with fact-checking media to avoid politicizing the initiative limited its impact. "If I had to do it again, I would create a broad, national media alliance with large media outlets to amplify the impact," he said. Figuring out how to form such an alliance when news outlets had strong partisan leanings would require additional thought, however. Arianna Espinosa, who worked closely with Galindo, said the registry should have done more to reach specific groups exposed to false information. "I would have liked to have more resources and a team within the registry to create specific dissemination strategies directed at certain groups [such as] young people and older adults-notably, via WhatsApp," said Espinosa. More extensive collaboration with civil society organizations that could monitor election-related communication and help fact-check information would also have helped increase the impact of the registry's strategy. Developing these arrangements required considerable time and effort, however. Conversations had to start well in advance of an electoral cycle. Navigating organizational interests and cultures was never a simple process. For example, the impediments to #VerdadElecciones2019 included the registry's inability to offer academic partners a mutually beneficial agreement to produce fact-checking content, the disconnect between the registry's goals and the capacities and policies of fact-checking websites La Silla Vacía and ColombiaCheck, and the registry's choice to develop its plan without collaborating with other institutions such as the Electoral Observation Mission, a civil society organization that promoted civil and political rights in Colombia. Civic education and digital literacy were essential parts of a whole-of-society approach to the fight against misinformation, some #VerdadElecciones2019 principals emphasized. If more citizens had a better understanding of how the electoral process worked as well as a stronger culture of critical thinking and alertness regarding informational threats, the impact of false news would diminish and purveyors of disinformation would have less scope for action. At the end of 2019, Galindo's term ended. "When the new registrar took office, he was given information about the platform that had been created, but it seems that it was not a priority for him," said Murcia. "That's a shame because we lost the opportunity to keep working on development of the initiative and to maybe get to a point of interoperability with other agencies. Because this was based on artificial intelligence, I can't imagine how much it could have learned and how much information it could have processed." According to Felipe Sánchez Iregui, who led the registry's communication department after Galindo left office, the registry's strategy for combating disinformation simply changed. Instead of only contradicting fake news item by item, the registry's process focused on identifying the main issues that drew comment on social media and could generate confusion. The priorities were to (1) provide information on the registry's responsibilities in the electoral process and increase the transparency of its activities and (2) focus communication strategy on main topics, with an emphasis on giving voters accessible information that did not include technical terms and that used more images and infographics. While Sánchez and his team were still working on their strategy for the 2023 regional elections, they focused mainly on the production of videos and graphics for social media, with the hope that more voters would access them. From 2020 to 2023, the registry multiplied its Twitter and Facebook reach by three and its Instagram reach by two and started using such platforms as TikTok to reach new audiences.37 Measuring the impact of efforts to combat misinformation and disinformation required additional thought not just on the part of the registry but also globally. Espinosa said: "We evaluated the impact only through the reach of our posts: how many people would see the information and how many times it was shared. Those are good indicators, but we don't know how to identify the impact of the information on people and whether they changed their minds or even believed the information. The fact that they were sharing our information could mean either that they trusted it or that they denounced it. We still don't really know how to measure impact." References 1The registry is "responsible for the technical and operational dimensions of the electoral process in Colombia. Among its other responsibilities, the registry oversees the organization of electoral and citizen participation processes, carries out national civil registration policies, updates the voter roll and oversees the vote counting and result dissemination processes." For more information, see https://www.ifes.org/tools-resources/faqs/elections-colombia-2022-presidential-elections. 2 See https://hora724.com/senador-jaime-amin-denuncia-no-tarjetones-la-consulta-derechista-biffi/ 3 See https://colombiacheck.com/chequeos/video-con-denuncia-de-tarjetones-de-la-derecha-agotados-es-del-2018; https://archive.is/SDGGF; https://hora724.com/senador-jaime-amin-denuncia-no-tarjetones-la-consulta-derechista-biffi/. 4 See examples at https://medium.com/dfrlab/electionwatch-misinformation-was-everyday-business-in-colombian-election-7b46ab536d9d; https://medium.com/dfrlab/electionwatch-bees-cause-buzz-in-colombia-a7ad06e0be7a; https://medium.com/dfrlab/electionwatch-fraud-claims-in-colombia-51e141f57745. 5 See https://www.moe.org.co/. 6 See the report at https://moe.org.co/wp-content/uploads/2018/11/Irregularidades-Electorales-en-Colombia-Informe-Pilas-con-el-Voto_Digital.pdf. 7 See decree 1010 of 2000 on organization of the National Civil Registry. Available at https://www.funcionpublica.gov.co/eva/gestornormativo/norma.php?i=64913. 8 See https://www.oecd.org/gov/gov-at-a-glance-lac-country-factsheet-2020-colombia.pdf. 9 See https://www.registraduria.gov.co/-Informacion-de-la-entidad-670-.html. 10 Source: DANE, Encuesta de Cultura Politica 2019. Available at https://www.dane.gov.co/files/investigaciones/ecpolitica/bol_ECP_19.pdf. 11See https://web.karisma.org.co/8790-2/. 12 See https://www.cigionline.org/articles/misinformation-shaping-colombian-election/. 13 See https://www.eltiempo.com/elecciones-colombia-2018/presidenciales/candidatos-presidenciales-firmaron-pacto-por-la-no-violencia-en-elecciones-2018-209888. 14 See articles 4 and 5 of the Decree 1010 of 2000. Available at https://www.funcionpublica.gov.co/eva/gestornormativo/norma.php?i=64913. 15 See Daniel Arnaudo et. al., Combating Information manipulation: A Playbook for Elections and Beyond," International Republican Institute, September 28, 2021 available at https://https://www.iri.org/resources/combating-information-manipulation-a-playbook-for-elections-and-beyond/ and https://marketing4ecommerce.mx/facebook-activa-megafono-electoral-otra-herramienta-las-elecciones-2018-mexico/ ; https://centralelectoral.ine.mx/2018/02/05/facebook-e-ine-anuncian-colaboracion-para-elecciones/. 16 See ISS case study on Sweden at https://successfulsocieties.princeton.edu/publications/sweden-defends-its-elections-against-disinformation-2016-%E2%80%93-2018. 17 Electoral witnesses were citizens who represented political parties, movements, or certain groups of citizens. In accordance with Law 1475 of 2011, electoral witnesses monitored the voting process. They could file complaints and request intervention by relevant authorities if they noticed irregularities. See https://www.registraduria.gov.co/Testigos-electorales-2042.html. 18 See https://www.registraduria.gov.co/IMG/pdf/20191125-informerendicioncuentas2019corte30septiembre.pdf. 19 According to the European Union's website, the standard Eurobarometer "is conducted twice a year and focuses on monitoring key trends relevant to the European Union as a whole, European Commission priorities as well as contemporary socio-political events. It allows analyzing long-term trends in attitudes related to European affairs. The Standard Eurobarometer surveys are generally conducted in a face-to-face interview format in all EU Member States and some additional countries and territories." The results of the Spring 2019 Standard Eurobarometer can be found at https://europa.eu/eurobarometer/surveys/detail/2253. See also https://data.europa.eu/data/datasets/s2183_464_eng?locale=en. 20 Source: GAD3. 21 The 2018 code of practice on disinformation developed a framework of self-regulation whereby major technology firms could fight disinformation. The code included measures to make political advertising more transparent, to close fake accounts more effectively, to distinguish bot-based and human interactions, to provide tools for reporting of disinformation, and to prioritize authentic information in feeds. See the document at https://digital-strategy.ec.europa.eu/en/library/2018-code-practice-disinformation. 22 Ángel Ocampo Rodríguez, "Registraduría y partidos políticos se suman al pacto contra la desinformación," RCN Radio, August 21, 2019. Accessed at https://webcache.googleusercontent.com/search?q=cache:upREwUsN620J:https://www.rcnradio.com/politica/registraduria-y-partidos-politicos-se-suman-al-pacto-contra-la%20-disinformation&cd=5&hl=en&ct=clnk&gl=fr&client=firefox-b-d. 23 The International Fact-Checking Network (IFCN) is an initiative created by United States-based Poynter Institute, with the stated goal of promoting fact checking in journalism. The IFCN created its own code of principles to help establish standards for fact-checking methods. The IFCN also organizes fellowships, training, and conferences. See https://www.poynter.org/ifcn/. 24 See https://www.javeriana.edu.co/documents/12789/11121862/P%C3%A1g.+10+y+11.+Estrategia+contra+noticias+falsas+en+elecciones.+HJ+agosto+2019+WEB.pdf/ad252cfe-407f-4988-a857-cd9f29c79fc5. 25 On July 30, 2019, the registry launched a call for tenders to provide a "comprehensive IT solution for the security of the 2019 electoral process." The call for tenders included a variety of IT services aimed at preventing and monitoring events that could put the security of the electoral process at risk. Within the requirements, the call for tenders mentioned a "solution to identify and counter disinformation strategies that affect the electoral process." Indra-Minsait submitted its proposal in August and competed against four other companies for the contract. The final contract was awarded to Indra-Minsait and included a variety of services related to the security of the vote. It was signed on September 13 and cost the registry 5.2 billion pesos (approximately US$1.5 million). The detailed cost of its antidisinformation component was not made public. Detailed information about the tendering process can be accessed on Colombia's public tendering platform at https://www.colombiacompra.gov.co/secop-ii. 26 See https://www.elinformador.com.co/index.php/politica/214045-las-estrategias-para-combatir-las-noticias-falsas-y-la-desinformacion. 27 See https://twitter.com/Registraduria/status/1186667811751497730?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1186667811751497730%7Ctwgr%5Ef5493a591219280ee26f6a30b990b5ee50f53bb1%7Ctwcon%5Es1_c10&ref_url=https%3A%2F%2Fwww.ambitojuridico.com%2Fnoticias%2Fgeneral%2Felectoral%2Fcaja-de-herramientas-para-las-elecciones-de-este-domingo. 28 See https://colombiacheck.com/chequeos/gaitan-no-dijo-hasta-las-4-de-la-tarde-vota-el-pueblo-despues-vota-la-registraduria. 29 See https://colombiacheck.com/chequeos/no-es-cierto-que-se-trasladara-el-puesto-de-escrutinios-en-buenaventura-como-escribio. 30 See the original tweet at https://twitter.com/petrogustavo/status/1188450390767013888/photo/1 and the registry's post at https://twitter.com/Registraduria/status/1188504269009371143/photo/1. 31 See https://twitter.com/Registraduria/status/1188486553804390400/photo/1. 32 See https://twitter.com/Registraduria/status/1188593829903032326/photo/1. 33 See https://twitter.com/Registraduria/status/1188289705059766272/photo/1. 34 Through this program, Facebook relied on independent fact checkers to review and rate the accuracy of posts on its platform. Each time a fact checker categorized a piece of content as false, Facebook limited the piece's distribution, labeled it accordingly, and notified people who tried to share it. More information on the program can be found at https://www.facebook.com/formedia/mjp/programs/third-party-fact-checking. 35 See https://wapp.registraduria.gov.co/electoral/Elecciones-2019/. 36 As an example, Gustavo Petro's tweet accusing the registry of fraud on October 27, 2019, was liked by almost 5,000 users and shared 2,889 times, whereas the registry's tweet contradicting the claim was liked by only 43 users and shared only 29 times. See https://twitter.com/petrogustavo/status/1188450390767013888/photo/1 and https://twitter.com/Registraduria/status/1188504269009371143/photo/1. Other examples can be found on the registry's twitter account at https://twitter.com/search?q=(from%3Aregistraduria)%20until%3A2019-10-28%20since%3A2019-10-26&src=typed_query&f=top. 37 Data provided by the registry's communications department. Innovations for Successful Societies makes its case studies and other publications available to all at no cost under the guidelines of the Terms of Use listed below. The ISS Web repository is intended to serve as an idea bank, enabling practitioners and scholars to evaluate the pros and cons of different reform strategies and weigh the effects of context. ISS welcomes readers' feedback, including suggestions of additional topics and questions to be considered, corrections, and how case studies are being used: iss@princeton.edu. Terms of Use Before using any materials downloaded from the Innovations for Successful Societies website, users must read and accept the terms on which we make these items available. The terms constitute a legal agreement between any person who seeks to use information available at successfulsocieties.princeton.edu and Princeton University. In downloading or otherwise employing this information, users indicate that: a. They understand that the materials downloaded from the website are protected under United States Copyright Law (Title 17, United States Code). This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/. b. They will use the material only for educational, scholarly, and other noncommercial purposes. c. They will not sell, transfer, assign, license, lease, or otherwise convey any portion of this information to any third party. Republication or display on a third party's website requires the express written permission of the Princeton University Innovations for Successful Societies program or the Princeton University Library. d. They understand that the quotations used in the case study reflect the interviewees' personal points of view. Although all efforts have been made to ensure the accuracy of the information collected, Princeton University does not warrant the accuracy, completeness, timeliness, or other characteristics of any material available online. e. They acknowledge that the content and/or format of the archive and the site may be revised, updated, or otherwise modified from time to time. f. They accept that access to and use of the archive are at their own risk. They shall not hold Princeton University liable for any loss or damages resulting from the use of information in the archive. Princeton University assumes no liability for any errors or omissions with respect to the functioning of the archive. g. In all publications, presentations, or other communications that incorporate or otherwise rely on information from this archive, they will acknowledge that such information was obtained through the Innovations for Successful Societies website. Our status and that of any identified contributors as the authors of material must always be acknowledged and a full credit given as follows: Author(s) or Editor(s) if listed, Full title, Year of publication, Innovations for Successful Societies, Princeton University, http://successfulsocieties.princeton.edu/. GLOBAL CHALLENGES: ELECTION DISINFORMATION Alexis Bernigaud Innovations for Successful Societies (c) 2023, Trustees of Princeton University