Detecting Digital Fingerprints: Tracing Chinese Disinformation in Taiwan from Graphika, the Institute for the Future’s (IFTF) Digital Intelligence Lab, and the International Republican Institute comments
Taiwan is on the frontlines of the Chinese Communist Party’s (CCP) international influence operations, and what happens on the island often serves as a harbinger for how China will operate elsewhere. In 2018, the island’s local elections were subjected to myriad online disinformation campaigns that favored a Beijing-friendly agenda, attempted to undermine democratic integrity, and systematically attacked democratically elected politicians whose positions did not align with China’s strategic interests. However, despite the assertion of Chinese interference by several intelligence agencies and governments, clear evidence linking disinformation during the local elections to mainland Chinese actors has not been publicly shared.
This is not a unique scenario: governments around the world have discussed foreign interference campaigns without being able to share much public evidence to accompany these assessments. Many factors complicate the task of publicly sharing this type of evidence, among them privacy concerns linked to accounts that are often on private platforms, methodological concerns around standards of attribution in information operations (IO), and use of sensitive technical data in the process of analysis and attribution. Independent entities also face difficulties in their assessment of potential interference by China in the 2018 elections, notably because the Taiwanese online information space is unique and conducting a postmortem without consistent monitoring and real-time data collection is practically impossible.
Yet, when asked whether a foreign actor was likely to target Taiwan’s 2020 Presidential and Legislative election with disinformation, Wu Jun-deh, Director of the Cyber Warfare and Information Security Division at the Institute for National Defense and Security Research (INDSR), said “Of course, the answer is China.”
In June 2019, with the 2018 local elections as a point of reference, Graphika, Institute for the Future’s (IFTF) Digital Intelligence Lab, and the International Republican Institute (IRI) embarked on a research project to comprehensively study the online information environment in the lead- up to, during, and in the aftermath of Taiwan’s January 2020 elections, with an awareness of the 2018 precedents and an eye for potential similar incidents throughout this election cycle. Graphika and DigIntel monitored and collected data from Facebook and Twitter and investigated leads on several other social media platforms, including Instagram, LINE, PTT, and YouTube. IRI supported several Taiwanese organizations who archived and analyzed data from content farms and the island’s most popular social media platforms. The research team visited Taiwan regularly, including during the election, to speak with civil society leaders, academics, journalists, technology companies, government officials, legislators, the Central Election Commission, and political parties. The goal was to understand the online disinformation tactics, vectors, and narratives used during a political event of critical importance to Beijing’s strategic interests. By investing in the organizations investigating and combating Chinese-language disinformation and CCP influence operations, we also hoped to increase the capacity of the global disinformation research community to track and expose this emerging threat to information and democratic integrity.
Two months into our research, Twitter and Facebook released statements that, for the first time, directly linked the Chinese state to an online information operation taking place on their platforms. Twitter stated that they were “disclosing a significant state-backed information operation focused on the situation in Hong Kong, specifically the protest movement and their calls for political change.” Twitter found 936 accounts on its platform linked to mainland Chinese state actors. Facebook stated, “although the people behind this activity attempted to conceal their identities, our investigation found links to individuals associated with the Chinese government.” Facebook disclosed five accounts, seven pages, and three groups. Graphika conducted an independent investigation of the 3.5 million tweets produced by this set of accounts, finding a prolific yet unsophisticated cross-platform amplification network that promoted smear campaigns against the Hong Kong protest movement and opposition figures like Guo Wengui.
Twitter data included three datasets: (1) tweets from Twitter’s Chinese information operations archive relating to Taiwan; (2) a month-long stream of tweets relating to the January election; and (3) a Graphika map of the election discussion on Twitter. Our Facebook dataset included a Graphika map of the 2020 election, consisting of public pages relevant to the election in Taiwan, as well as additional analysis of 139,538 Facebook posts gathered using CrowdTangle. We also investigated content on several other platforms, including YouTube, Instagram, LINE, and PTT. More details on these datasets, including streaming queries, Taiwan keywords, and number of posts and users in each set, can be found in our Datasets and Methodology appendix.
Events in Hong Kong went on to shape Taiwan’s information environment in myriad ways. In the early phase, a number of narratives emerged alleging that, for instance, the Taiwanese government was secretly providing large-scale financial backing to Hong Kong protesters at the expense of the Taiwanese taxpayer. As the protests and the Hong Kong government’s crackdown continued, Taiwanese public opinion about the island’s own relations with China radically shifted, making it untenable for China- friendly politicians to maintain a stance on cross- Strait relations that would be acceptable to Beijing. The predominant narrative in June 2019 was that Taiwan’s presidential and legislative elections would be so tightly fought that a disinformation campaign could decide the result, but a shift in opinion meant that by fall 2019 the results of the presidential election in favor of the incumbent, Tsai Ing-Wen, seemed a foregone conclusion, and only the legislative seats were seriously contested. The Taiwanese information space during the 2020 presidential and legislative elections—a popular incumbent, no credible opposition, and a plummeting view of China—was thus radically different from the 2018 local elections, making it implausible that the same narratives would play out.
On January 10, the day before the vote, COVID-19 started to shape the Taiwanese information environment. A rumor circulated online that a new type of SARS had reached Taiwan and that it would be unsafe for citizens to vote in person. In the weeks and months after the election, disinformation associated with COVID-19 was regularly seeded in Taiwan, much of which was directed at undermining the government’s
Key Findings are -
Finding 1
Disinformation related to COVID-19 was used to discredit the Taiwanese government and had links to mainland China.
Mainland Chinese accounts pushed COVID-19 disinformation targeting Taiwan on Facebook and Twitter. These accounts revealed their Chinese origin by overlap with previous Chinese netizen-led disinformation campaigns targeting Taiwan and a poor grasp of linguistic differences between Taiwanese and Chinese Mandarin. response to the virus and sowing distrust. Although the initial prompt for this study was election interference, COVID-19 and other incidents described in this report make clear that disinformation in Taiwan is a persistent threat not limited to election cycles.
Given that malign foreign actors typically exploit existing weaknesses and vectors for influence in a country’s information environment, we have analyzed disinformation related to Taiwanese democratic integrity regardless of whether it can be directly linked to foreign actors. We found a number of indicators suggesting coordinated disinformation campaigns targeting Taiwan, some of which were foreign in origin, particularly in the post-election period. The most clear-cut of these was a sustained, post-election COVID-19 disinformation campaign that showed signs of coming from the Chinese mainland and used Malaysian content farms to disseminate and amplify false information about the virus.
Finding 2
Disinformation was targeted at undermining democratic actors writ large, not just the election, and increased in the months immediately following the election.
In addition to foreign campaigns focused on COVID-19, a network of domestic Taiwanese accounts drove a cross-platform campaign falsely alleging Tsai Ing-wen’s Ph.D. dissertation was fake. These inauthentic Facebook, Twitter, and Instagram accounts promoted a petition to the U.S. government to investigate her Ph.D.’s authenticity. These posts often included instructions and links to YouTube videos instructing Chinese speakers not fluent in English how to navigate signing a petition on petitions.whitehouse.gov.
Finding 3
Content farms in Malaysia promoted Han Kuo-yu, the Kuomintang (KMT) candidate for president, and criticized Tsai Ing-wen, the Democratic Progressive Party (DPP) candidate and incumbent president.
This network of content farms coordinated production and distribution of these stories in the lead up to the election. Often the stories displayed links to mainland China through Chinese vocabulary choices, similarities with content from attributed Chinese government information operations, running stories copied from PRC state-owned media outlets, or by running disinformation attributed to the Chinese government. After the election, this network promoted several false stories alleging that COVID-19 originated in the U.S.
Finding 4
Disinformation frequently targeted the voting process and Taiwan’s Central Election Commission (CEC) before the election. The Taiwan FactCheck Center catalogued several examples of disinformation on LINE and Facebook that targeted central aspects of Taiwan’s democratic process, including the voting process and the CEC, some of which alleged CIA intervention to swing the result. At least one of these stories displayed signs of mainland Chinese authorship through vocabulary choice.
Finding 5
One presidential candidate’s Facebook page appeared to benefit from false inflation, gaining a suspicious and abrupt increase in Facebook followers one month before the election.
James Soong, a third-party candidate for president, gained nearly 500,000 Facebook followers in a period of days the month before Taiwan’s election. Soong received the highest rise in followers out of all 268 official candidate and party Facebook pages we observed in the month leading up to the election. This represented a 356% increase in followers and occurred over a span of about 72 hours. Soong’s suspicious gain of nearly half a million followers over four days is unlikely to be organic and warrants further investigation.
Finding 6
Disinformation was a cross-platform problem during the election.
We observed disinformation on six social media platforms—Facebook, Instagram, LINE, PTT, Twitter, and YouTube—and on dozens of domains. Far from being limited to one platform, disinformation was present on all social media platforms studied.
Finding 7
Taiwan’s domestic digital marketing industry plays a large role in political disinformation on the island. The industry is primarily commercially motivated.
