ChinAI #263: A History of the Chinese Computer

Reviewing Thomas S. Mullaney's history of Chinese language computing

Greetings from a world where…

Waterloo sunset’s fine

…As always, the searchable archive of all past issues is here. 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).

Book Review — The Chinese Computer: A Global History of the Information Age

At a time when Chinese quantum computers and Chinese AI systems and Chinese electric vehicles absorb all headline buzz and pundit bloviation, Professor Thomas S. Mullaney’s text draws our attention to a humbler and arguably more important technology: the Chinese input method editor (IME).

The Chinese IME is a technology of translation: it converts QWERTY alphanumeric keyboard inputs into Chinese characters. Today, I can easily switch over to my Chinese IME (by pressing the caps lock key as a shortcut) to type out the Chinese phrase for input method [输入法] by pressing the three keys “s”, “r”, “f” and then “1” (to select first option of characters from the resulting pop-up menu). In tracing efforts to develop Chinese IMEs back to the 1940s, Mullaney reminds us of this translation task’s scale and weightiness. How to make a QWERTY keyboard produce tens of thousands of characters in a language full of homophones that have the same pinyin spelling?

Incredibly, today, the fastest Chinese typists can produce 221.9 Chinese characters per minute. This exceeds the record in English, which sits at 174 words-per-minute. Somehow, Chinese IMEs have gotten “faster than the mode of textual production they were built to compensate for: English and the longstanding model of one-key-one symbol, what-you-type-is-what-you-get (p. 219).” What makes this development all the more astonishing is that for most of the twentieth century, Chinese computing was slower such that “the markedly slower performance of Chinese raised disturbing questions about Chinese cultural and civilizational ‘fitness’ (p. 216).”

How did we get from then to now? For Mullaney, the answer lies in the nature of IMEs as a compensatory technology: “The IME has never been a revered technology, no matter its potential or proven track record. From its inception to the present day, input has tended to be understood as an inherently compensatory technology, one meant to assist or ‘work around’ the challenges confronted in the computational processing of Chinese text (p. 11).” Chinese computing is the middle school boy with bad acne who is not very good at sports. Chinese IMEs are the sense of humor he develops in order to belong.

In six carefully-researched chapters that span roughly six decades, Mullaney takes us through the set of “work-around” technologies that have turned Chinese into, perhaps, “the fastest writing system of the digital age (p. 154).” This is a global history text in that developments consistently traverse national borders, bringing in characters from high-level Chinese government actors, the Taiwanese military, IBM, Japanese linguists, and the Pentagon.

The Chinese Computers embeds its facts and insights within sensitive stories about people who often get left out of the history books. The first chapter is about IBM’s electric Chinese typewriter, which was invented by Chung-Chin Kao in the 1940s. But it is really about Lois Lew (featured on the book’s cover), an IBM plant worker without a high school education, who became the typewriter’s human IME. To demonstrate the machine’s capabilities at events in China, Lew had to memorize thousands of codes that linked four-digit numbers to Chinese characters — all without the help of code books, let alone pull-up menus. Lew is now in her 90s. She swims at the YMCA three hours a week and regrets not buying IBM stock. When Mullaney interviews her about the typewriter’s code manual, she can still recall the four-digit codes: (You, 0275, 你; He, 0178, 他; Me, 0314, 我).

At times, Mullaney’s interest in peripheral characters left me a bit unsatisfied, but this is a quibble rather than a quarrel. For example, chapter four profiles Zhi Bingyi, a Taiwanese engineer who developed an IME that broke down Chinese characters into component parts. After learning all about Zhi’s efforts to woo the Chinese government, the chapter’s conclusion briefly tells us that Chinese government agencies sidelined foreign systems in favor of developing indigenous ones. I would have liked to see this section delve deeper into the role of the Chinese state as well as the four main leading enterprise groups which had very close government ties (Stone, Legend, Founder, and China Great Wall Computer). Qiwen Lu’s China’s Leap into the Information Age, a notable gap in the references list, would provide a helpful starting point.

Chapter five was my favorite chapter: “The Search for Modding China: Printers, Screens, and the Politics of Peripherals.” As personal computer imports to China were projected to grow from just 600 in 1980 to 130,000 in 1985, Chinese customers faced the key obstacle of Chinese-language compatibility. Mullaney details, “Western-built printers couldn’t natively output Chinese characters, Western-built monitors couldn’t display them, and Western-designed operating systems couldn’t deal with the intricacies of Chinese input methods (p. 156).” Computer memory systems, for instance, assigned 5 bytes of memory per symbol for digital English font, whereas one Chinese character required 32 bytes. Dot-matrix printing methods outputted Chinese characters that were “twice the height of English words (p. 169).”

The heroes of this chapter’s account are engineers who came up with modifications to Western-manufactured printers, monitors, and operating systems. To make these systems more legible in Chinese, they rewrote printer drivers and hacked program subroutines. Over time, in the early 1990s, Western-built computers began to incorporate these modifications into their core architectures, but Mullaney urges us not to forget the work of these early “modders”:

“Instead the history of early Chinese personal computing has been retrospectively imagined as one in which the Western-built computer had always been language-agnostic, neutral, and welcoming. All but forgotten is the fact that…most of these problems were first solved, not by the likes of IBM, Microsoft, and Apple, but my modders. Thanks to these modders, Western-manufactured personal computers became usable, and thus useful, to one-sixth of the global population (p. 182).”

This is essential reading for ChinAI subscribers. I highly recommend that you buy the book, and let me know what you learned! Thanks to Tom for sharing an advanced copy of the book with me. Published by MIT Press, the book comes out on May 28, 2024. Pre-order here!

ChinAI Links (Four to Forward)

Must-read: Keep your enemies safer: technical cooperation and transferring nuclear safety and security technologies

My latest article is now out in European Journal of International Relations:

Even during the Cold War, the United States and the Soviet Union cooperated on nuclear safety and security. Since accidental or unauthorized nuclear detonations anywhere threaten peace everywhere, it seems straightforward that states more experienced in developing nuclear safety and security technologies would transfer such methods to other states. Yet, the historical record is mixed. Why? While existing explanations focus on the political costs and proliferation risks faced by the transferring state, this article argues that specific technological features condition the feasibility of assistance. For more complex nuclear safety and security technologies, robust technical cooperation is crucial to build the necessary trust for scientists to transfer tacit knowledge without divulging sensitive information. Leveraging elite interviews and archival evidence, my theory is supported by four case studies: US sharing of basic nuclear safety and security technologies with the Soviet Union (1961–1963); US withholding of complex nuclear safety and security technologies from China (1990–1999) and Pakistan (1998–2003); and US sharing of complex nuclear safety and security technologies with Russia (1994–2007). My findings suggest the need to examine not only the motivations behind nuclear assistance but also the process by which it occurs and the features of the technologies involved, with implications for how states cooperate to manage the global risks of emerging technologies.

I’ll do a fuller breakdown of the article next week. ***Note: for those without access to the journal, the full manuscript (without typesetting and copyedits) is available here on my personal site.

Should-read: China bets on industrial AI

Mary Hui, in her weekly a/symmetric post, analyzes industrial AI as a “route of differentiation” for China’s competitive AI strategy. Pulling together a lot of Chinese sources, she highlights that integrating small-scale LLMs into industrial applications may reduce the computing power requirements.

Should-read: Troubling the Water

For the Made in China Journal, Yangyang Cheng criticizes new restrictions on U.S.-China scientific exchanges in critical technologies by tracing the history of these exchanges.

Should-read: The Cloud Under the Sea

In The Verge, Josh Dzieza unearths the undersea cables that power the internet, with some great details about how the maintenance of these cables has been neglected.

Thank you for reading and engaging.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.

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).

Also! Listen to narrations of the ChinAI Newsletter in podcast format here.

Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99