Taiwan is the world’s biggest target for foreign disinformation, according to The V-Dem Institute, and its society reflects this. Its National Science and Technology Council is currently working on the Trustworthy AI Dialogue Engine (TAIDE) to combat disinformation coming from chatbots. Citizens can attend civil defense training courses at the likes of Kuma Academy to learn how to spot disinformation. And, frequently, diplomatic exchanges focus on sharing methods for countering disinformation, with Taiwan’s joined-up political warfare system praised by some for its efforts to keep out Chinese influence.
However, conspicuously lacking in some of the forums for monitoring or discussing disinformation is an explicit definition of how it should actually be classified. This presents a number of issues. For one, you can have everyone agree that fake news has a disruptive effect on democracy when in fact you’re all thinking about different things. And more importantly, if you don’t discuss who gets to decide what fake news is, undue power can be ceded to whoever takes on the task, stifling democratic debate.
The starting point for this problem is that much of what is now labeled as fake news is not simply fake facts, it’s so-called false “narratives.” Narratives that are skewed disproportionately in one direction or another, working through lies of omission and disproportionality, as much as outright lies.
“The main challenge in the information ecosystem has shifted from identifying fake news to disclosing misleading information narrative manipulation through distorting effects,” said Hwang Chao-hwei (黃兆徽), chief content officer at Taiwan AI Labs Foundation, at a recent Asia Centre forum in Taipei (speaking through AI software to demonstrate its potency at recreating likenesses).
One example Hwang offered for this was the egg shortage in Taiwan last year. Hwang said a 5% shortage of eggs in Taiwan was amplified by non-Taiwanese troll groups, particularly on Facebook, which then exacerbated the shortage by causing panic buying. This appears relatively cut and dry. The scale of the response made the shortage worse, which seems a reasonable definition of a “skewed” narrative. And Hwang’s AI Labs Foundation was able to tie down the source of that to bad faith actors on Facebook, accounting for clearly larger proportions of egg-based discussion than other groups.
But not every case looks so simple. Hwang’s next example was of narratives “criticizing U.S. President Joe Biden and questioning U.S. foreign policies.” Depending on how it is assessed, such a broad category could place a huge amount of different “narratives” within the bracket of “skewed” or malicious, cutting off many legitimate criticisms and political perspectives.
This is not just a hypothetical concern in Taiwan. “It is certainly a danger that the government acts as the arbiter of what is fake news, or what constitutes a legitimate political narrative or not,” New Bloom editor Brian Hioe told Domino Theory by email. “There are times in which the Tsai [Ing-wen (蔡英文)] administration has ventured into questionable territory, for example, as in attempts to dismiss the outbreak of racism against the prospect of Indian blue-collar migrant workers joining the workforce as simply the result of Chinese disinformation rather than reckon with that this was an outbreak of racism from the Taiwanese population.” He also noted that “Rosy perceptions of the U.S. that fail to account for its own domestic failings or human rights violations committed by the U.S. abroad also contribute to a view of criticisms of the U.S. as potentially being disinformation.”
For Hioe, one structural balancing mechanism against this issue becoming more potent comes from the positioning of the current Democratic Progressive Party (DPP) government in relation to China. “[P]an-Green [aiming for more distant relations with China] administrations have an imperative on distinguishing Taiwan from China on the basis of its free speech and protection of democratic freedoms. As such, the actions that they would take against disinformation tends to be rather limited,” Hioe said. He noted that when TV channel CtiTV had its broadcast license revoked, the Tsai administration “stuck to regulatory violations” rather than challenging its affiliations with China.
However, issues remain. Hioe suggested that the opposition Chinese Nationalist Party (KMT) may not have the same incentives toward restraint should it take power. He noted its current legislative push for new powers to prosecute government officials, members of civic organizations and others as useful evidence to support this case. He also noted that the KMT is attempting to “revive institutions used in the process of Taiwan’s democratization to target non-KMT politicians [such] as the Special Investigation Division and put this under the control of the legislature.”
So, if government is no neutral arbiter, who can be trusted to decide what constitutes a “false narrative,” and on what grounds?
For T.H. Schee (徐子涵), an expert on internet and public policy in Taiwan, public institutions are a key thread in pushing out factually accurate information in the first instance. Speaking via email, he suggested efforts should be put into investing in public media and reinforcing existing public information channels like official websites. He also said open and public channels outside of social media should be prioritized, and endorsed the work on the TAIDE AI dialogue mechanism.
But this still leaves room for marking out existing disinformation, and at the moment this work has been taken on by a number of fact–checking NGOs. How do they acquire legitimacy when analyzing and labeling concepts as slippery as “false narratives”?
For Eve Chiu (邱家宜), of Taiwan FactCheck Center, the answer is in limiting analysis to outright facts. “[W]e check no opinions but [only] facts,” she said via email. “[F]alse information is usually partly false, therefore we dissect the narratives and do debunking on fact-related parts but let go of the views and opinions. We do our best to find evidence to prove or disprove the fact related parts like statistics, testimonies of witnesses, an authentic video, etc. Our fact check reports can be refined by new evidence and we do it very often according to the feedback from the audience.”
For Filip Noubel, managing editor of Global Voices, an NGO whose Civic Media Observatory investigates how narratives are constructed in media, legitimacy comes partly through the quality of the process. Speaking via email, he said his NGO puts substantial research into “understand[ing] the shape, arguments and origins as well as modes of transmission of narratives” in a “non-partisan, open-minded way in the spirit of an investigation.” They then look into both who these narratives harm and who benefits from them, as well as the impact they have on “larger concepts of democracy, elections, plurality, tolerance, national or environmental security.”
A second plank of legitimacy then comes through opening up the results to wider discussion. “[A]n essential element is to open the data and research as we do and to present it to the public and to the media so everyone can verify the analysis of information items and in the process become more literate in narrative detection. To sum up, it must be a collective, transparent and participatory process,” Noubel added.
Of course, this does not mean that narrative checking can be fully immunized from making difficult decisions. Judgements about proportionality still ultimately have to be made, as National Taiwan University Graduate Institute of Journalism director Hung Chen-ling (洪貞玲) acknowledged recently while speaking on the issue. Noting that “our freedom of speech is directed by and is at the mercy of giant platforms and business interests,” she went on to explicitly tie freedom of speech to national security. “When a democracy is constantly threatened by a neighboring authoritarian country through information manipulation [on digital platforms], the talk of freedom of speech needs to be done in a new light to see how a democracy’s freedom of speech and national security can be protected,” Hung said.
She is not alone in this kind of view. AI Labs’ Hwang Chao-hwei has said the ultimate goal of freedom of speech is to protect human rights.
Many might disagree with either of those framings. But they are at least an honest, transparent approach that can be debated. The worst of all worlds would be where these lines exist but aren’t expressed out loud. Where decisions on the limits of free speech are made without debate — and thus without widespread democratic consent.
Leave a Reply