Cooperation and Competition: Algorithmic News Recommendations in China’s Digital News Landscape
By Emilie Xie, Qiguang Yang, and Sun Yu
The flood of technology companies into the news ecosystem, including the emphasis on increasing attention and advertising, has driven publishers to restructure their processes and approaches to engagement and revenue. One of the most visible changes is the prevalence of algorithms in the news industry and their increasing ability to select the news that people see online. As news organizations and platforms transform into personalized news pipelines, questions are being raised about bias, invisible gatekeeping mechanisms, and the protection of consumer data. The problem with this is that not only can personalized algorithms potentially filter out a diverse range of content in favor of user-directed interests, but these filtering decisions can also mean a downplaying of (or overemphasis) on certain information. One country that relies on such filtering is China.
Over the past few years, personalization recommendation technology has fundamentally altered information distribution methods in China and the way the country’s platforms and publishers approach news consumers. In China, the news and technology industries have converged, with the values and priorities of technology platforms coloring publishers’ decision-making processes and business strategies. With funding and support from the Tow Center for Digital Journalism at Columbia University, researchers from Columbia University, Renmin University, and a reporter from the Financial Times explored a heavily regulated news ecosystem where technological developments have far outpaced the concerns of industry professionals.
Our research focused on the realities of the politically surveilled Chinese cybersphere, where algorithms continuously scan and censor material they deem unacceptable, such as politically sensitive information. And in a country home to the world’s largest number of internet users, who average twice the amount of time online as compared to their US counterparts, these extensive internet and media regulations are significant.
Our research attempted to explore and understand perceptions of and approaches to personalization among news organizations and technology companies in China. From September 2019 through August 2020, we conducted interviews with 26 individuals from 19 different Chinese media and technology organizations, culminating in a final report for the Tow Center to be published later this month. We selected China as a case study because of its widespread use of personalized news distribution services by technology and media companies alike. For example, in 2012, Chinese company ByteDance (of TikTok fame) launched one of its flagship products, Jinri Toutiao, a personalized news aggregator. Unlike technology companies such as Facebook and Google, which rely heavily on recommendation algorithms, Jinri Toutiao sells its approach as AI-centric. The company offers personalization based on users’ information as its core service. By 2018, the company’s popularity led to the proliferation of other personalized news aggregators such as Kuaibao and Yidian Zixun. Within the Chinese media studies field, the term “Toutiaoization” (“头条化 toutiao hua”) has emerged to refer to the extensive use of algorithmic recommendation technology by information distribution platforms to provide users with personalized and targeted content.
For the purposes of our study, we were primarily interested in understanding how Toutiao-esque personalization (as opposed to non-customization) affected a news organization’s business strategy and decision-making process. How important is personalization for the growth of their companies, and what did they think the future held for personalization within the Chinese digital news industry? In addition, we wanted to know how individualized algorithmic news distribution affected journalistic practices, and how much did competition with other organizations factor into such practices? We also asked our participants about the level of human intervention versus automation their organizations utilized, and how media-related restrictions, such as directives from the Chinese Cyberspace Administration, shaped the work they did. We then compared their responses from before versus after the outbreak of COVID-19 because during the pandemic, all organizations focused heavily on news related to the virus. In turn, we wanted to know if our respondents saw less value in personalized news recommendations.
Among our interviewees, we found that there was a clear acceptance of personalized news distribution as a core feature of the Chinese news media industry. They understood that all publishers and platforms engaged with it to varying degrees, from relying heavily on personalized news aggregators to experimenting with personalization on their own news apps. When asked why they thought organizations were pursuing personalization to distribute news content, our interviewees said that personalization helped counter information overload, enhance user targeting abilities, adjust users’ limited attention resources, and increase engagement and traffic. In China, private media companies cannot discount the competition coming from tech news platforms. And although state media are able to sustain themselves with government funding, private media companies look to commercial pursuits such as personalization to bolster their publications’ presence and freedoms.
On one hand, our findings indicate that Jinri Toutiao’s influence vis-à-vis algorithmic recommendations affected the way Chinese media organizations approached personalization, motivating other publishers and platforms to experiment with and implement similar strategies. On the other, news publishers emphasized the different advantages they had against the overwhelming traffic enjoyed by technology platforms and news aggregators, such as the importance of good journalism, variety of content, and their own organizations’ stricter standards for what passed as news. Moreover, news aggregation platforms were limited in their content creation privileges and subjected to other regulatory scrutiny. Regulation is built into the fabric of the Chinese news ecosystem and has become more all-encompassing over the last decade. These rules manifest in several ways, including the requirement that content of political importance be pinned at the top of an app’s home interface, thus rendering even personalized news aggregators unable to disregard manual intervention and curation. As a result, we found publishers relied on platforms for traffic, leading those platforms to turn to publishers for content, forming a relationship characterized by both cooperation and competition. The interconnectivity and reliance between platforms and publishers blurred boundaries and has led to a loss of news branding and differentiation. For example, some interviewees expressed frustration about consumers assuming all news they read on Toutiao was produced by Toutiao. It has also led to changing definitions of “media,” as news organizations ventured beyond traditional activities associated with newsmaking by doing things like organizing concerts.
The home interface of the People’s Daily news app. News articles of political importance are designated with a red “Pinned.”
It's important to note that in the months following the pandemic, our participants' work routines underwent relatively minor changes even with an increase in remote working and a heightened focus on pandemic-related news. There was, however, augmented levels of human intervention, including the prioritization of COVID-19 news in algorithms. Our interviewees said all platforms decreased their usage of personalized recommendations because the pandemic was the story that most people cared about. At the same time, and as the pandemic slowly wanes, all interviewees agreed personalization would likely only increase in the future and viewed it as an irreversible trend.
For the non-China scholar, this report presents several implications for the digital news landscape beyond China. News organizations are dipping in and out of the traditional understanding of news media in the race for attention and revenue. Ultimately, the tensions revealed in the report mirror the questions and debates found in other newsrooms around the world, including issues of editorial control, journalistic curation and integrity, and the roles personalized algorithms will play for the foreseeable future.