What you need to know
In Taiwan's open society, government and political groups have embraced the use of Facebook as a barometer of public sentiment.
Look down any Taipei Metro car during rush hour, and it is immediately obvious that Facebook is a fact of daily life in Taiwan; 77 percent of the country claims to use the service, according to data-analysis firm Statista, among the highest penetration rates in the world.
But with great penetration comes great responsibility, and research conducted by The News Lens suggests data on Taiwan's citizens has been widely harvested by private and government entities.
Much of what happens with this data is a black box, but here are a few things we learned about data scraping in Taiwan:
A shortcut to sentiment
According to industry veteran TH Schee (徐子涵) of the Open Knowledge Foundation, a UK-based data transparency advocacy organization, the Sunflower Movement, a series of protests against the Taiwan government in 2014, acted as a catalyst for an intensification of data collection -- the government realized the importance of keeping a close eye on popular sentiment.
This helped engender a government culture that viewed the use of Facebook data as a positive; a way to respond to the emotions of the public.
Taiwan’s Digital Minister Audrey Tang (唐鳳) told The News Lens that the use of social media to track public sentiment was widespread; she said that at least a dozen government ministries are paying companies to conduct keyword monitoring -- scraping the internet, and Facebook groups and pages specifically, for the use of certain terms, or keywords.
Keyword tracking is anonymous and does not typically involve gathering users’ personal information. For example, private analytics firm Qsearch [whose tagline is "Reading the global mind"] was able to track the surge in public interest related to the toilet paper panic in early 2018, when news of an impending price hike created a run on toiletries. Savvy suppliers were able to take advantage of this knowledge through social media -- one company was even swift enough to put together an online raffle for toilet paper.
At least a dozen government ministries are paying companies to conduct keyword monitoring.
Documents seen by The News Lens that were produced by a contractor for a national government organization showed that Facebook data is a commonly used currency when it comes to measuring citizens' reactions to various subjects. Much of it is incredibly banal -- one report gathered from Facebook data detailed public opinion regarding betel nut use before and after an anti-betel nut campaign. Opinions regarding the chewy, intoxicating seed remained the same.
Analytics tools like this are pervasive in Taiwan, and are not inherently nefarious. One Facebook tracker, Social Insight, was even developed by the Institute for Information Technology (III), an NGO that works closely with the Ministry of Economic Affairs, to assist companies, as well as academic and research organizations.
Politics
But the biggest prize is political, and these tools are also in widespread use across Taiwan’s political system.
One anonymous staffer for the Ko Wen-je (柯文哲) Taipei mayor campaign in 2014 interviewed by researchers for Oxford University’s Computational Propaganda Research Project (PDF) claimed that the campaign had built web scrapers that gathered public Facebook data on 11-14 million Taiwanese Facebook users.
The staffer said in the interview that “this bot respected Facebook’s privacy terms and only crawled public pages, not individual user profiles. The crawler bot was still able to generate data on individual users, however, given that any Facebook user’s activity on a public page is also public.”
They were then able to use this data to improve the fledgling politician’s campaign messaging: “The intelligence gathered by this bot was also useful to gauge voters’ reactions to real-time events, such as mistakes Ko made, or campaign strategies.”
Ko-P, as Taipei's mayor is affectionately known, is hardly the only politician to have made use of Facebook data.
According the Open Knowledge Foundation's Schee, politicians’ Facebook pages have become big business, with campaigns spending up to US$400,000 per year to maintain a social media team that is also tasked with collecting and generating leads from data.
Shifting tactics
The face of social media is changing, and the upcoming November elections are expected to have a markedly different tone than previous years. Schee told The News Lens that with recent changes to Facebook’s policies, it will be harder to target specific voters.
In the wake of the Cambridge Analytica scandal, Facebook in April tightened its application programming interface (API) – the method by which data can be used by other programs – to reduce the amount of personal information that can be scraped by outside sources. The changes are wide-ranging– they no longer allow apps to read users’ Facebook comments through the API, while also restricting access to call and text history as well as apps’ access to events and groups, all methods by which data could previously be harvested en masse, and indeed probably was.
This data can be fed into an algorithm that determines which ads a user would be most responsive to; even basic demographic information can vastly improve the chances that someone will respond to a message.
Cambridge Analytica’s jumped on this train early, using data from up to 87 million Facebook profiles to build what they called “psychographic profiles” to target voters with ads. While they are far from alone in doing this, they gained massive international attention due to their unethical data collection methods and contentious choice of clients, including the 2016 U.S. presidential campaigns of Ted Cruz, Ben Carson and Donald Trump, groups supporting Leave in the UK’s Brexit referendum and the Barisan Nasional political party that recently lost power in Malaysia, among many others.
photo credit: REUTERS/Leah Millis/達志影像
Because of Facebook’s subsequent crackdown on user data, Schee predicts that Taiwan will see an upsurge in the role of content farms, which he suggested will replace microtargeting as a primary means of pushing propaganda.
“People can just plagiarize content from a credible website. This will be a common practice this year, outside the boundaries of Facebook.”
Schee said that there were already a lot of content farms springing up: “Just a few years ago, there were only one or two content farms, exclusively targeted at tech readers, but now you have lifestyle, politics, economics…”
The goal of these content farms is twofold; they can be massively profitable businesses, supported by online advertising revenues and near-zero costs. Once they achieve critical mass in terms of following, those controlling the sites can splice in politically charged content aimed at specific demographics.
Content from The News Lens' Taiwan edition is an occasional target for plagiarizing content farms, with some articles even being turned into bizarre computer-narrated YouTube videos, and content from the English edition has also appeared on content farm sites without permissions or attributions.
But how much of this online content are people actually reading? According to a 2017 study by the Reuters Institute, just 16 percent of Taiwanese people said social media was their main source of news, but 31 percent relied on online news sites, higher than any country in the region except South Korea. This is paired by extreme skepticism of the news media – only 36 percent said that they trusted the news they consumed, and 17 percent said that the news was free from political and business influence.
The propaganda question
In an open society like Taiwan, social data is as easily available to foreign actors as it is to local players.
One example: In 2016, photographs circled back from China of a nuclear-capable bomber, apparently cruising in the skies close to Yushan, Taiwan’s highest mountain.
Taiwanese media quickly picked up the story, running with the notion that a Chinese military aircraft had penetrated Taiwan's airspace uncontested, leading to swift denials by Taiwan's military.
As it turned out, the photo was real, but the mountain was not, in fact, Yushan. The question is, how did the misinformation attached to the image gain so much traction?
Tang told The News Lens that when it comes to propaganda, the most trustworthy content gets shared the most: “If the original message comes from Taiwan, for it to go viral it needs a lot of local contributions.
“[So] for foreign actors, it's much easier if you just sponsor a local team and do locally-originated content instead of trying to supply your own message,” she said.
According to Tang, “It's very difficult to infect people with a viral message and expect people not fact check. In that sense, I think Taiwan's very healthy. Of course, there are pockets of population who are educated in a more authoritarian era, before the martial law was lifted. They tend to look at messages printed in some form that were spoken with some authoritative voice or are credited to some authoritative figure and be more inclined to believe it.”
Taiwan was under martial law between 1949 and 1987 (1992 on Kinmen and Matsu), during which freedom of speech and the press was greatly curtailed.
“In Taiwan there is a generational and educational gap, but we see the prevalence of social media and there is a lot of cross-generational learning taking place as well, and I think even older generations are catching on that not all that they see is real.”
Legality
Bills such as the Digital Communication Act (DCA), currently being debated in Taiwan’s Legislative Yuan, intend to tighten restrictions around the use of data by private companies. Such legislation complements the already in force Personal Information Protection Act.
The legislation aims to bring Taiwan loosely in line with the General Data Protection Regulation (GDPR) passed in the European Union, according to Tang. The GDPR came into effect around the world on May 25.
The GDPR, which greatly expands the right of citizens to control their own data, is expected to have wide-ranging ramifications in Taiwan, as the regulation applies to any business that does business in the EU or employs European citizens.
It is likely that your personal information is still up for sale in less reputable corners of the internet.
The dawn of the GDPR does not mean that your data is safe or has disappeared from the internet for good, and it is likely that your personal information is still up for sale in less reputable corners of the internet. According to Schee, a list of Facebook users and their information is useful, even if the data is outdated or people have closed their accounts: “They can combine the data with their own database; people can buy private data from the black market – it's very valuable.”
The Cambridge Analytica mystery
SCL Group – the parent company of SCL Elections, which became Cambridge Analytica in 2013 – once asserted that they worked on a national election campaign in Taiwan on their website, but the information was deleted in May 2017.
A statement on the SCL Group website read: “Working for a major Taiwanese political party, SCL undertook all research, strategy, content creation and implementation for a national political campaign. This required SCL to not only balance delicate local concerns, but also display sensitivity to larger geo-political issues.”
The statement dates back to September 2010, and was last seen on the company’s website May 2017.
While the page has since been taken down, some associated data was still available on SCL’s Amazon CloudFront web server at the time of publication.
Several deleted versions of the website were preserved on archive.org’s Wayback Machine, a project of U.S.-based nonprofit digital library The Internet Archive, as well as on private web crawler archive.is.
SCL Group (formerly Strategic Communication Laboratories) also claimed to be active in several other countries, including Indonesia, the Philippines, India and Thailand. Those countries are still featured on Cambridge Analytica’s election website, while Taiwan has quietly disappeared.
Taiwan was not specifically mentioned in a cache of documents submitted by Canadian whistleblower and former Cambridge Analytica employee Christopher Wylie to parliament in the UK. The documents can be found here.
Photo Credit: AP/達志影像
The majority of the data used by Cambridge Analytica to build their "psychographic profiles" was primarily collected through the Facebook app “Thisisyourdigitallife” by an academic researcher in 2014, four years after SCL initially claimed it worked on Taiwan’s elections.
Questions
What were SCL's activities in Taiwan and what data was collected from Taiwanese citizens?
While the statement regarding SCL's activities in Taiwan was unchanged from the time of its posting in 2010 to its disappearance in May 2017, it also remains uncertain when and if SCL Group and Cambridge Analytica ceased operations in the country.
Representatives from the Democratic Progressive Party (DPP), Kuomintang (KMT) and SCL Group did not respond to requests for comment. A representative from Cambridge Analytica referred The News Lens to their Twitter profile for information.
The News Lens is interested in any tips related to this story, especially which politicians or parties SCL has worked for in Taiwan. We can be reached by email.
Read Next: Cambridge Analytica – The Singapore Connections
Editor: David Green
