The Hong Kong protests through the eyes of Weibo

Team Members

Sebastian Bertoli, Sameena, Thomas Hardiman, Sirak Tegegn

Introduction

The Hong Kong Protests

On Friday 27 September 2014 large student protests joined by the Occupy Central with Love and Peace movement began in Hong Kong. The cause of the protests was Beijing's decision to vet for candidates in the 2017 election for Hong Kong’s chief executive. This would mean that the future candidates running for election will effectively be preselected by the central Chinese government. In addition there were also fears that the ‘one country, two systems’ rubric that has been in place in Hong Kong since the 1980s might eventually be reverted. The protesters demand was electoral freedom, a continuing concern in Hong Kong since it passed from British control to Chinese control in 1997.

The protests quickly escalated over the weekend of the 28th of October arguably partly caused by Police forces’ use of pepper-spray and tear gas which drew widespread condemnation and thousands of additional demonstrators in the street. On the 3th October 2014 chaos erupted when government loyal mobs attacked protesters. In the following weeks the number of protesters decreased but they nonetheless remained in control of the barricades they created across the city. Talks as well as additional protest action however did not bring any progress in fulfilling the protesters demands.The main protest actions came eventually to an end in the second week of December when police dismantled the main protest site and protesters were hauled off.

The Hong Kong protest movement caught the public imagination as the ‘umbrella movement’ due to protester’s use of the now emblematic rain umbrella to defend themselves from the teargas used against them by governmental forces. Time magazine has called the protests ‘the most significant political protest since 1989’ (Barber n.p.).

Social Media and the Protests

Throughout the protests Chinese authorities intensified patrolling and control on social media platforms. In fact major news sources reported at the wake of the protests that photo sharing platform Instagram had been blocked in mainland china supposedly to curtail the spread of imagery of the protests. Additionally many posts showing support for the Hong Kong protests or containing sensitive keywords referring to the protests were removed from the Twitter-like microblogging platforms Tencent Weibo and Sina Weibo. In fact the South China Morning Post reported on the basis of statistics it had gathered from researchers at the University of Hong Kong that the deletion rate of posts on the Chinese Weibo’s had increased fivefold between Friday 26 September 2014 and Sunday 29 September 2014 (Boehler).

Tencent Weibo

Tencent Weibo (launched in 2010) is together with its competitor Sina Weibo (launched in 2009) part of the two biggest micro-blogging platforms in China (Lukoff). Arguably, with approximately 220 million active users per month and over 500 million registered users it is also one of the biggest micro-blogging players in the world (Millward). Its features are comparable to those of Twitter which includes the ability to create a ‘tweet’, to ‘retweet’ as well as to ;mention’, ‘reply to’ or to ‘follow’ a particular user. However they are called differently: a ‘Tweet’ is called a ‘weibo’, a ‘retweet’ is a ‘forward’ whereas a ‘hashtag’ is called a ‘topic’. In addition Tencent Weibo has some additional features such as the ability to ‘comment’ on a post that makes it a more feature-rich and complex micro-blogging platform compared to Twitter. Moreover while the limit of 140 characters remains the same much more content can be conveyed Chinese compared to English (Canaves 77).

Comparative studies between Chinese social media and western social media platforms have found that major difference in the way content is shared on these platforms (Yu et al.; Daifeng et al.). For instance it was found that the social networks on Tencent Weibo are more complex and that there is a greater tendency that users forward posts (the equivalent to retweeting). Reposting ratio increase dramatically during ‘hot events’ like Hong Kong protests, where studies have found that they reach on average 65.9% (Guan et al. 6). In a nutshell Chinese micro-blogging platforms might be comparable to Twitter but they should not be seen as merely being a Chinese version of Twitter. Both functionality, usage and user activities are different and as such have to be taken into account when studied.

Research Questions

The aim of this research is twofold. Firstly, to analyze the unfolding of the Hong Kong protests through the microblogging platform Tencent Weibo in order to understand the popular public sentiment with regards to the protests, and uncover the extent to which Chinese censorship directives might affect how the protests are reflected on Tencent Weibo. The second part of the research focuses on comparing the posts on Tencent Weibo with the news coverage of the protest in regional (South China Morning Post), mainland China (Xinhuanet) and international news media (BBC). This comparison will bring out the differences in the way the protests are covered across different media, and study the influence of pro-government agencies on these various media platforms.

Hypothesis for research objective 1: The web of Chinese censorship is strong and pervasive and has been successful in eradicating the pro-protest sentiments from being expressed on the Tencent Weibo platform.

Hypothesis for research objective 2: Except for the international news media, all other news media platforms and the Weibo platforms are heavily influenced by pro-government agencies.

Scope & limitations of this research

There were two limitations concerning the data that was available and considered for this study. Firstly this particular study focuses only on one (Tencent Weibo) out of the three popular micro-blogging platforms in Hong Kong - the other two being Sina Weibo and Twitter. Since Hong Kong operates under the directive of ‘one country two systems’, unlike mainland China it has access to all the western media platforms and websites. Hence a comparison of the public sentiment on the protests across all the three platforms would have been insightful. While this research study is part of a larger project that did indeed carry out research on Twitter and Sina Weibo a cross platform analysis was due to logistic reasons not possible at the time of writing. Secondly although the protests started already in the last week of September and lasted to approximately December only data pertaining to October was available.

Methodology

The research methodology for this study is divided into two parts: The first one relates to the data from Tencent Weibo for the month of October and the second one relates to the data from the different news media for key dates in October.

Data from Tencent Weibo

Social media sites are some of the first sources of live information for the public, as well as the means to spread the word amongst the people (Parker; Fan and Osnos). Hence this data was carefully collected using the three step process explained below:

1) Data Collection: This was done by searching for particular protest related keywords using the public search API on Tencent Weibo. The date range was specified as 1 – 31 October. Below is a description of the keywords that were used in the search:

2) Data cleaning: The data-set containing 4729 posts was cleaned with help of the online tool ‘Discovertext’. This involved mainly the deduplication of entries (40 duplicate entries were removed).

3) For the data analysis we identified four broad themes based on their content and categorized the posts accordingly.

1. Posts relating to on-ground protest events (20% of the entire dataset)

2. Posts with opinions on the government and the Chief Executive of Hong Kong (10% of the entire dataset)

3. Posts with ‘anti-occupy’ message (40% of the entire dataset)

4. Posts that have no relevance to the protests (30% of the entire dataset)

Once the posts were categorized we discarded those posts that were not relevant for our study and created a table of top three reposts for each of the seven key days (see Figure 1) we decided to analyse. The translation work of the posts written in Mandarin was done in a semi-automated fashion via Google Translate and with the aid of a native Mandarin speaker that checked over the results. Top reposts were specifically used because they gave a fair idea on the popularity of the public sentiment on that day. Additionally a list of the overall top three posts for the entire time period were gathered, in order to understand the general interest of the population surrounding the protests. For a full list of top three reposts for each key date and for the overall month, refer to this spreadsheet. As a part of the analysis, we also checked which of the posts were still available online and which had been subsequently deleted. Depending on how many and which posts were eliminated, it became easier to understand if the pro-government authorities had any influence in eliminating them.

Data from news media

The first point for collection of this data was to identify the news journals that could give a complete picture of the protest coverage. Hence it was decided to include an International, National (Mainland China) and Regional (Hong Kong) based news media so that the study had the advantage of knowing the point of view that each of these media platforms carried towards the protests. Thus, the three news journals included in this study are South China Morning Post for regional Hong Kong news, Xinhuanet for news from mainland China and BBC for international news coverage.To keep the study crisp and well rounded, we drew up an event timeline of the protests in the month of October and picked those dates in which some key events took place. Posts from Weibo and articles for these particular dates were then compared to arrive at the findings. Below is the event timeline for the same. For detailed news items covered in each news journal, see here.

Figure 1 – Event timeline of Hong Kong Protests. News courtesy BBC, South China Post, Xinhuanet

Findings

This section details the findings based on the data analysis process mentioned earlier.

Overall findings for the month:

The below graph represents the number of protest related posts on Tencent Weibo for each day in the month of October 2014. The blue blocks represent posts that are still currently available online, and the grey blocks represent the posts that are no longer available. It can be seen that there is a huge surge in activities towards the later part of the month when the talks between the government and protesters were held.

Figure 2: Timeline of posts on Tencent Weibo



Top Re-posts:

Top reposts was specifically used because they gave a fair idea on the popularity of the public sentiment on that day. Additionally a list of the overall top three posts (see figure 3) for the entire time period were gathered, in order to understand the general interest of the population surrounding the protests.

Figure 3: Overall Top 3 protest related reposts for the month

As can be seen from the figure above, 2 out of the three popular posts are against the protest movement in Hong Kong. An example of the anti-occupy post is shown below.

Figure 4: An example of anti-occupy post

According to the studies conducted by Open Net Initiative, all Internet platforms in China have to comply with censorship regulations. Censorship is enforced not just by censoring keywords, or deleting offensive or harmful content, but also through the presence of groups such as “50 cent party members”. This is a group of netizens paid by the government to create pro-China comments in exchange for half a Renminbi. It is estimated that 250,000 to 300,000 people belong to the 50-cent party (Sterbenz) and are used to sway the public opinion in the “right direction” in times of unrest. Hence there may be a chance that the anti-protests posts may have come from such a source. The findings reported in the next section will help clarify this further.

Deleted Posts

We found that 18% of the posts collected on the key-dates are not available anymore. This means that when trying to access these posts via a web-browser a 404 HTTP response code was generated.

Figure 5: % Number of posts deleted from Weibo

This prompted us to look closer at the content of the deleted posts and in particular to look at those posts that had become part of large threads within Tencent Weibo with high posting-activity. The removed post below from the 21st of October 2014 for instance was a verbatim re-post or ‘forward’ of a thread containing 10,707 contributions. The original user who wrote the message asks for ideas on how to boycott (resist) the ‘unscrupulous artist’. The user is referring to Chinese artists that have publicly supported the Occupy Central movement such as Hong Kong actors Tony Leung Chiu-Wai and Chow Yun-fat as well as Hong Kong born singer Andy Lau (# Addy). The post also mentions the Communist Youth League of China, a 89 million strong youth movement and sub-party of the Communist Party of China.

Another post that was part of a larger and very active thread revolving around the Hong Kong protests can be interpreted as a call to arms to resist the Occupy Central movement. From the research we carried out it emerged that posts like the one below were likely to be removed from the platform. All posts removed from the most active threads were calls to action in support of the anti-Occupy-central or ‘blue ribbon’ mobs, which violently attacked protesters (Fitzpatrick).

Hence it can be concluded that what is likely to be censored is not necessarily content that is critical of Chinese authorities but it is any content that could, even if pro-government, ignite collective activities, and as such lead to more eventual unrest. (Gary et al. 1).

Comparison of the top reposts with the events happening on the ground: A comparison of the event timeline with the top reposts can be accessed here. This table shows that the discussions on Weibo have been in line with the offline events taking place in Hong Kong. Since the analysis consists of reposts, it is an indication of the popularity of the sentiment of people related to a certain event. Over the month it is consistently seen that posts relating to the ‘anti-occupy’ movement, for instance posts against the protest and pro-democracy sentiment are way more popular and have been reposted more than the protest ones. There can be two reasons for this kind of behaviour: One, that the public of Hong Kong is truly against the pro-democratic protests and is inclined to be a part of mainland Chinese regime. The second could be the fact that the Internet censorship teams of the Chinese governments have manipulated the posts on the platform.

From the data gathered, it seems more likely that the second reason is more valid here. The reason being that most anti-occupy posts seem to be posted by ‘bots’ - this is proved by the fact that all of them are posted within seconds of each other and everyone of them are reposted exactly the same number of times. The posts from Oct 21 - 25 indicate this. Also, for some of the key dates of the protests like 21, 15 etc. there were no relevant protest related posts present which the Chinese censorship teams may have deleted. Hence this analysis indicates strong evidence of the data having been manipulated by the pro-government teams.

Framing of content on different media: This is a comparison of how the reports have been covered across the different media under study here - Social Media (Tencent Weibo), International News Broadcast (BBC), Hong Kong News Broadcast (South China Morning Post) and Mainland China News Broadcast (Xinhuanet). The aim is to compare the sentiment and any perceived bias in the way news gets reported on these different platforms.

The silent protest on China National day or the day the Hong Kong chief executive issued a careless remark about the protests - are important days in the history of the protest. Nevertheless the fact that the dataset does not contain any posts relating to these events, suggest that the censors and pro-government authorities may have deleted posts of such nature. On the other hand when the initial anti-occupy supporters tried to disrupt the protests, the tencent weibo dataset carries all the posts in favour of the anti-occupy movement. In fact from the dates between 21 - 26 October, there were several anti-occupy posts that emerged on tencent Weibo, all posted within seconds of each other and each of them being reposted exactly the same number of times. For example anti-occupy posts on 21st were re-posted 10 thousand times, on 25th 11 thousand times and so forth. This is an indication that 'bots' were potentially involved in posting and reposting of these messages on Weibo, thus confirming beyond doubt the involvement of the authorities in manipulating the public sentiment on a public opinion platform.

A look at the published news articles indicates that the BBC and South China Morning Post have a neutral tone of reporting on the protests. They are not biased towards any party. However Xinhuanet articles were in favour of the Chinese government. For example, the coverage of violence during protests is the same in BBC and South China Morning Post where they report about arrests and civilian casualties, whilst the Xinhuanet article looks mostly at police casualties. Another example that supports this view is the way the meeting between the student representative and the government was reported. The BBC and South China Morning Post reported that the event was ‘fruitless’ while Xinhuanet article contained information on how the government was approaching it peacefully.

Thus both the social media platforms and the mainline newspapers seem to be heavily influenced by the censorship regulations and pro-government authorities.

Conclusion

The aim of this research was to analyse how the Hong Kong protests were covered and reported on the microblogging platform Tencent Weibo. This trend was further compared and analysed with the reporting of the events in other national and international media journals against key protest related events that took place in October 2014.

From the research findings described in the above section it can be inferred that the dataset is slightly skewed with a lot of irrelevant data being included in this set. From among the relevant data it is quite clear that the anti-occupy message or the anti-protest message resonates strongly on the platform. However this is not the ‘real’ sentiment of the public or the platform. It has been seen clearly that most of the anti-protest messages were planted by the pro-government authorities, to shift the mood on the platform. And this was further achieved by eliminating majority of the posts of key protest related events from the platform as seen from the findings on deleted posts.

These incidents are a strong indicator of just how pervasive and strong the censorship activities in China are. The censorship regulatory teams have not just ensured the deletion of the ‘offensive’ or ‘harmful’ posts but they have also successfully tried to shift the sentiment in favour of the government by ensuring that anti-occupy messages gain the most popularity - 11 thousand reposts for anti-occupy posts versus a maximum of 600 reposts on a protest related post.

The same trend reflects even in the journalistic media - Hong Kong based news media and International news media seem to report more unbiased and neutral views giving a fair hearing to both sides. However the mainland China news media has made deliberate efforts to twist the story against the protesters in Hong Kong.

China leads the world in how seriously it takes its censorship laws and to what extent it can go to uphold them. The following statement by Faris & Villeneuve explains this behaviour in precise terms, ‘Claiming control of the Internet has become an essential element in any government strategy to rein in dissent—the twenty-first century parallel to taking over television and radio stations [...] the constant blocking of a swath of the Internet has become part of the everyday political and cultural reality of many states.’(Faris & Villeneuve, 2008).

By claiming the Internet and the various social media platforms, the Chinese government essentially has all the roads to expressing negative public opinion barred effectively.

Discussion

The results have shown that censorship and framing of the news is modified to suit the political agenda of the Chinese government. Despite not having a conclusive dataset from Sina Weibo, the data from the Tencent Weibo proves that our hypothesis has been proven and shows a correlation between real life events and activity on the social media sphere. For future studies, based on the social media usage in China and Hong Kong, it would be interesting to view censorship across both of the Weibo platforms and Twitter. In addition, it would be interesting to include results across the time period of the Hong Kong protests, from September to December and follow how the trends develop.

Bibliography

Boehler, Patrick. “Record Censorship of China’s Social Media as References to Hong Kong Protests Blocked.” South China Morning Post. N.p., n.d. Web. 15 Jan. 2015.

Cai, Weiyi. 'Looking At Hong Kong’S Current Protest Through Weibo'. N.p., 2014. Print.

Canaves, S. “China’s Social Networking Problem.” IEEE Spectrum 48.6 (2011): 74–77. IEEE Xplore. Web.

Haewoon Kwak, Changhyun Lee, Hosung Park, Sue Moon (2010) What is Twitter, a social network or a news media? Proceedings of the 19th international conference on World Wide Web: 591-600.

Li, Daifeng et al. "What is the Nature of Chinese MicroBlogging: Unveiling the Unique Features of Tencent Weibo." arXiv preprint arXiv:1211.2197 (2012).

Millward, Steven. “Tencent Weibo Reaches 540 Million Users.” Tech in Asia. N.p., 22 Jan. 2013. Web. 14 Jan. 2015.

Parker, Emily, Jiayang Fan, and Evan Osnos. 'Social Media And The Hong Kong Protests - The New Yorker'. The New Yorker. N.p., 2014. Web. 17 Jan. 2015.

R. Faris and N. Villeneuve (2008). " Measuring Global Internet Filtering." in R. Deibert et al. (eds.), Access Denied: The Practice and Policy of Global Internet Filtering. Cambridge, MA: MIT Press, 5-27.

Sterbenz, Christina. 'China Banned The Term '50 Cents' To Stop Discussion Of An Orwellian Propaganda Program'. Business Insider. N.p., 2014. Web. 15 Jan. 2015.

Yu, Louis, Sitaram Asur, and Bernardo A Huberman. "What trends in Chinese social media." arXiv preprint arXiv:1107.3522 (2011).

Open Net Initiative - https://opennet.net/research/profiles/china