ChinAI #138: Splinternet Superstitions

How China's IPv6 Rollout Challenges Splinternet Narratives

Greetings from a world where…

heat waves been fakin’ me out

…Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors). As always, the searchable archive of all past issues is here.

Feature Translation: China’s Progress on Large-scale Deployment of IPv6

In September 2018, Eric Schmidt, former Google chief executive, prophesied of an Internet splintered into two, with one led by the United States and the other led by China. One month later, the New York Times editorial board — citing Schmidt’s comments but doing the bare minimum to avoid regurgitating them — claimed there would be three Internets! Don’t forget about the one led by the EU, they remind us.

Here’s a modest proposal. What if we did the following two things:

  • Clearly conceptualize the Internet as a collection of technological layers rather than a singular technology

  • Measure the level of fragmentation across different levels using data rather than relying on impressionistic judgements

Fortunately, I can save you some work because the Daylight Lab at UC Berkeley’s Center for Long-Term Cybersecurity did just that. They divided the Internet stack into five layers:

  1. Link layer: the Ethernet protocol transfers data within a local network

  2. Network layer (focus of this week’s feature translation): facilitates data transfer between networks. IP addresses, for instance, are essential components of this layer. Because of the extraordinary growth of Internet-linked devices, IP addresses based on IPv4, the previous standard, are becoming scarce. So, countries are starting to switch to IPv6, a new IP protocol. Daylight Lab considers “greater IPv6 adoption as lower fragmentation, as countries remaining on IPv4 will experience exhaustion and possible outages.”

  3. Transport layer: ensures persistent connections, protection against network interference events (e.g. throttling, state-launched attacks)

  4. Application layer: includes web browsing and stuff that “standardizes communications across transport layer to create specific-purpose functionality”

  5. Regulation layer: includes data laws like those that restrict the use of encryption and cross-border data flows

After measuring proxies for fragmentation (e.g. IPv6 adoption rates) across all five layers, Daylight Lab’s findings challenge the received wisdom that “the Internet is largely bi-polar, split between ‘free’ and ‘closed’ Internets, and countries’ profiles will be similar within these basic groups.”

They specifically explored the narrative that “China’s model of the Internet has set a precedent, one which other Belt & Road countries follow” by examining data on Internet fragmentation in Laos, Indonesia, Mongolia, Pakistan, Djbouti, Argentina, and Sudan. These are all countries that have received substantial Chinese infrastructure investments, making them the most likely candidates to fit the Splinternet narrative. However, they found that these countries rank lower than China on many fragmentation indicators, including data locality, website locality, and network interference.

Context: All of that tees up this week’s feature translation of an article on the rollout of IPv6 on China. The June 2020 piece (link to original Mandarin) is by two researchers at the China Academy of Information and Communications Technology, a think tank under the Ministry of Industry and Information Technology that takes on dual roles as a brain trust and regulatory body.

Key Takeaways:

  • In November 2017 China issued an action plan on large-scale deployment of IPv6, which set the following benchmark: “By the end of 2025, the scale of China’s IPv6 network, users, and traffic ranks first in the world.” Chinese government entities and centrally-owned companies have led the way in IPv6 adoption: 86 percent of the websites of provincial-level governments and centrally-owned companies are accessible via IPv6. In addition, 78 of the top 100 websites and applications (by number of users) can be accessed via IPv6.

  • IPv6 penetration in China is growing and is probably higher than the Daylight Lab’s metrics. Daylight Lab uses Google data (the % of users that access Google over IPv6) to measure IPv6 adoption (2.25% adoption rate in China) but since Google is not used by many in China, this is probably not representative. According to data from the National IPv6 Development Monitoring Platform, as of November 2019, the number of active IPv6 users in China reached 273.9 million, which is around 30% of the Internet-using population. Again, according to the framework above, adoption of IPv6 at scale results in a more connected Internet, which goes against the Splinternet narrative, at least on one of the five layers.

  • China’s deployment of IPv6, however, still trails the leading countries: not all active users are always accessing websites over IPv6, so China’s true IPv6 penetration rate probably trails the U.S. rate (45%) by a significant margin. In addition, many of the aforementioned IPv6-accessible websites only support IPv6 access on their homepages, and IPv4 is still used for the links that actually generate the most traffic (e.g. streaming media, pictures, etc.) According to the CAICT researchers, other barriers to the diffusion of IPv6 in China include: issues with cloud service platforms and home wireless routers. Another example that cuts against the “China steals and scales U.S. tech faster and better” narrative.

  • What’s the connection to AI? IPv6 is an important infrastructural layer on which future AI applications will be built. As the article states, “with the rapid development of the Internet of Things, the Internet of Vehicles, and the Industrial Internet, China’s future demand for IPv6 addresses is still relatively large.” I want to stress that Internet protocols are very much out of my wheelhouse, so please share feedback, comments, and corrections on all these points.

FULL TRANSLATION: How is China’s Progress on Large-scale Deployment of IPv6

Thank you for reading and engaging.

*I’m behind on my reading, so no Four to Forward this week, but I’ll catch up next issue.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is a PhD candidate in International Relations at the University of Oxford and a Predoctoral Fellow at Stanford’s Center for International Security and Cooperation, sponsored by Stanford’s Institute for Human-Centered Artificial Intelligence.

Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).

Any suggestions or feedback? Let me know at or on Twitter at @jjding99