QWERTY Did Not Conquer Chinese Computing

Western computing history loves to tell itself a stupid little fairy tale: we built the universal machine, everybody else adopted it, the end. Cute. Shame about the facts.

One of the best things I read today was about how Chinese computing basically had to hack the entire idea of a keyboard just to exist. Chinese script was treated for decades as a “problem” for telegraphs, typewriters, and early computers, mostly because those machines were designed around alphabetic languages and then had the nerve to blame everyone else for not fitting. Very elegant bit of colonial ergonomics there.

According to the Computer History Museum, MIT professor Samuel Caldwell and his team built an early Chinese computer called Sinotype in the 1950s, treating character entry as a search problem instead of trying to force characters into alphabet logic. Later, in the late 1960s and 1970s, engineers went gloriously off the rails with hardware experiments. In an adapted excerpt from his new book for IEEE Spectrum, historian Thomas S. Mullaney describes Chan-hui Yeh’s IPX system, a Chinese keyboard with 120 levels of shift that could theoretically enter 19,200 characters. Not twenty-six letters. Nineteen thousand two hundred. Absolutely deranged. Also brilliant.

And it was not alone. IEEE notes that other designs ranged from 256-key to 2,000-key systems, with some using styluses or stranger interfaces instead of normal keyboards. For a while, Chinese computing was a full-contact hardware design sport. The industry did not yet know that the boring beige rectangle would eventually win by sheer market momentum.

The twist is that QWERTY did win the hardware war, but not the software war. As China opened up in the 1980s and 1990s, cheap mass-produced personal computers flooded the market, and they were overwhelmingly QWERTY-based. So the real innovation moved into input methods. The excellent World of Chinese piece lays out the lineage: Cangjie, invented in Taiwan in 1976 by Chu Bong-Foo; Wubi, invented in 1983 by Wang Yongmin; and eventually the rise of pinyin input, despite early engineers hating it because phonetics are ambiguous as hell.

They were not wrong. One syllable can map to a pile of characters, which means the machine has to guess what you mean. That sounded inefficient until engineers realized that people do not write isolated characters, they write words and phrases. Once software started handling multi-character prediction, the supposedly bad idea became practical. Mullaney’s argument, repeated in both CHM and World of Chinese, is the part I love most: autocompletion and predictive text were pioneered in Chinese computing because the language forced the machine to get smarter.

So here is the take. QWERTY was never universal. It was provincial hardware that got dragged into universality by everyone it failed to serve. Chinese users did not passively adapt to the computer. They redefined the human-computer interface so thoroughly that now the rest of the planet takes IMEs, suggestions, and predictive text for granted. That little candidate bar on your phone, the one finishing your thoughts like an overeager intern, owes a lot to engineers solving a problem Silicon Valley did not even know how to describe.

Mullaney says 51 percent of the global population now uses computers in ways those computers were never originally designed to support. Good. The machine deserved to be bent.

Sources: Computer History Museum | IEEE Spectrum | World of Chinese