Scantily clad women instead of protest posters: the Chinese government is suspected of flooding Twitter with suggestive content. A stress test for the converted platform.
Scores of Chinese Twitter accounts that had barely tweeted for months suddenly began sharing minute-by-minute images of scantily clad young women and escort service ads on Sunday. Just as Twitter began to emerge as a channel for coordinating protests in China.
The result: when hundreds of people demonstrated against the government in several Chinese cities on Sunday for the first time in decades, Twitter spread escort spam using search terms such as “Beijing”, “Shanghai” or “Shenzhen” instead of calls for protest. The Chinese government is suspected of intentionally flooding the platform with spam to disrupt the organization of anti-government demonstrations.
The profile of @Marjori81038508 shows how this works in detail – apparently a bot. He posts the same video every minute of an escort lady touching her lips lasciviously. As the Investigation by an anonymous Twitter analyst shows, the account only started mass distribution of the video on Sunday. Before that, Marjori81038508 didn’t publish anything, although “she” has had a profile on Twitter since May. Since the profile includes the major cities of Shanghai, Guangzhou and Wuhan in its name, it appeared under these search terms.
pic.twitter.com/XQa9oE1GRH
White only finish.— 北京外围/三-亚外围/上海外围/杭州外围/广州外围/南京外围/武汉外围/大连外围/济-南外围33 (@Marjori810) November 28, 2022
The strategy of escort spam is already known
In a brief search on Monday, the NZZ came across dozens of profiles that operate according to a similar pattern. That backs up the research by the Twitter analyst, who has over 200 accounts identifiedwho only started tweeting masses of tweets about escort services or online games last weekend.
It is not the first time that a suspected government-affiliated organization in China has used such a spam strategy, said a former Twitter employee told the Washington Post on Sunday. His team fought such escort spam both with algorithmic measures and manually. This doesn’t seem to be entirely successful at the moment, after all the spam was still widespread on the platform on Monday.
The question now arises as to whether this has anything to do with the major changes on Twitter in recent weeks. Since Elon Musk took over the platform about a month ago, thousands of employees have left the company and Twitter has changed fundamentally.
About three quarters of Twitter employees are gone
Musk dissolved supervisory boards, fired top managers and made himself the only director. After a few days, he dismissed a good half of the original 7,500 employees by email. Since then, further layoffs have followed in several waves: Musk fired some employees after publicly criticizing his decisions. He then sent out an ultimatum: employees could choose to be part of a new “hardcore” twitter, with longer hours and no home office option, or leave the company. Hundreds have apparently opted for the latter.
How many employees still work for Twitter today is difficult to say. American media assume that only around 2000 employees, i.e. around a quarter of the original workforce, are left.
Is it possible to downsize a company so drastically in a short space of time without jeopardizing ongoing operations? On the surface it looks like it, after all Twitter still works. This prompted some observers on Twitter to make malicious comments: the many employees apparently had nothing useful to do.
A social network on autopilot
Software developers opposed it. Some compare the current situation to an airplane being flown on autopilot. You can rely on it in normal operation, but you still don’t do without the human pilot, because he is needed in emergency situations. It’s the same with Twitter.
Millions of users are constantly interacting on the platform on various end devices, as well as advertisers and bots. You are constantly generating new content that needs to be stored somewhere and shown to the right people. This requires software and employees that decide in real time which content is undesirable and should be filtered out.
56) Spam! Spam is an *existential* threat. You need automated systems, AI-based clustering, entire teams doing manual review, and some *extremely* smart, driven, and creative engineers adept in adversarial thinking because you absolutely must be one step ahead at all times
— Mosquito Capital (@MosquitoCapital) November 18, 2022
Events such as the World Cup, attacks or earthquakes mean that many more users are online at the same time, which can overload the system. Security gaps that need to be patched can appear at any time, hackers try to penetrate the system. Or, as is now the case in China, unforeseen waves of protests can arise that are slowed down by spam.
Musk is dependent on China through Tesla
These are just a few scenarios in which personnel who know their way around and can react quickly are needed. Organically grown software systems like Twitter are complicated, the individual building blocks refer to each other. Developers document what they change. Nevertheless, there is a lot of implicit knowledge in such a system: knowledge that leaves the company with experienced employees – and when so many leave at the same time, newcomers can only build it up again with great effort.
According to that Tech magazine «The Verge» However, employees report that, among other things, the team responsible for maintaining the most important code modules is decimated – as well as the shift service that is available around the clock for emergencies.
The current case also illustrates that with a platform like Twitter, solving problems is not only of a technical nature. There are also political and moral choices to be made, in a delicate environment. Through his car company Tesla, Musk is dependent on China’s dictatorship, he produces and sells a large part of the vehicles there. He makes concessions to maintain relations: For example, he wrote a friendly opinion piece for the newspaper of the Chinese censorship authority. This, in turn, displeases the United States.
Making wise decisions as a platform in this environment not only requires technical knowledge, but also a deep understanding of the political context.
There’s no telling if things would have been different with the Chinese bots before Musk took over the platform. The former employee, who was interviewed by the Washington Post, suspects so. Anyone on Twitter who dealt with Chinese government influence has since left the company, he says.