ITP Techblog

China’s Cyberspace Administration last week published general details of algorithms used by 30 of the country’s largest tech companies.

The information appeared in Mandarin on a government website and didn’t give any company secrets away. For instance, ByteDance, the owner of the social network TikTok, outlined how its algorithm determines a user’s likes and dislikes in order to recommend videos in its app.

Meituan, a major Chinese food delivery company, uses an algorithm to choose drivers and delivery routes in order to get food to customers most efficiently. None of that is surprising. But you can guarantee that if this information is being made public, the Chinese Government has access to a heck of a lot more detail about the algorithms that help determine what the Chinese buy, stream, consume and believe.

It was only a matter of time before Beijing wanted to look into the black box of precious algorithms the tech companies guard so jealously.

China's Big Tech clampdown

China is a couple of years into a major clampdown on its tech giants, which have flourished over the last decade with the likes of Tencent and Alibaba becoming nearly as powerful in their own right as Meta and Amazon.

The sector’s influence on citizens’ lives was considered a growing threat to the Communist Party’s authority. The resulting regulatory crackdown has limited what companies can do and introduced tougher data protection laws.

Anti-trust action has prevented already large tech companies from expanding into new categories and has even seen the sidelining of China’s tech hero Jack Ma, the co-founder of Alibaba, after he criticised the Communist Party’s actions in the tech space.

Auditing algorithms

I have some sympathy for China’s tech companies - they took up their Government’s call to build alternatives to the US tech giants but have been punished for being too successful. However, requiring some level of transparency into algorithms is also something I think government regulators should be pursuing everywhere.

As consumers of online services, we have no visibility into how algorithms make decisions about us. We just see the results. These algorithms are the secret sauce that have made Google the king of search engines and Meta, the master of social media. But there have been well-documented perverse consequences resulting from the application of these algorithms on a global scale. Many governments are considering regulating the use of algorithms. But regulation is useless unless they can see whether algorithms are complying with regulation.

Tiktok's voluntary move

TikTok has gone a step further in the US and recruited Oracle to audit its algorithms to prove that its algorithms aren’t being manipulated to suit the demands of the Communist Party. Huawei made a similar offer here a few years ago when it was prevented from building 5G mobile networks here. But its plan to set up a centre where government security experts could examine its hardware and software code in private was ultimately rejected.

Some type of auditing of algorithms is necessary and maybe the Chinese have settled on a model we can learn from. We could be in the early stages of an era of greater transparency around algorithms, offered voluntarily in some cases to win social licence, and required more generally by regulators as they look to exert some control over algorithms and artificial intelligence systems.

This can be done without spilling trade secrets. After all, as one cybersecurity expert pointed out on Twitter this week, governments are making data requests of companies all the time: