How Every Computer Is a Chinese Computer (twirling the cord with Thomas Mullaney)

You're listening to Policy Prompt. I'm Vass Bednar, and I'm joined by my co-host, Paul Sampson. He's the president of the Center for International Governance Innovation. Policy Prompt features long-form interviews where we go in-depth to find nuances in the conversation with leading global scholars, writers, policymakers, business leaders, and technologists who are all working at the intersection of technology, society, and public policy. Listen now wherever you find your podcasts.

Good to see you, Vass. Have you ever wondered how a character-based language works on an English letter- based, you know, those Qwerty, Q-W-E-R-T that you see on your keyboard?

No, I almost want to say, to get to the other side, but I haven't given it much thought, I'm going to be honest with you.

But seriously, what was in your head about what a Chinese keyboard would look like? Today we're going to get into all this through the history of The Chinese Computer and what it could mean for the evolution of language in the digital form. And I hadn't really thought before either about just how big these changes might be.

No, absolutely. I mean, I knew nothing about the history and context of that development, and you're totally right. I didn't know what to expect, and I never really stopped to think about what it would mean to go beyond even trying to change the language on my keyboard. I think the only language I've added is emojis, and I felt really proud of myself for that.

Well, today we'll hear about the evolution of the Chinese computer through a fascinating history. Our guest is Thomas Mullaney, who's a professor of history in East Asian studies at Stanford.

He's incredible. He has seven influential books, among them The Chinese Typewriter, Your Computer is on Fire, and his latest that we've been chatting about, the Chinese Computer, which is the result of about 15 years of research. We are focusing this discussion on that last book, but those three form, I think a bit of a trilogy if you're into those.

Tom, thanks so much for joining us today and welcome to Policy Prompt.

Thank you very much for having me. It's wonderful to be here.

Today we're essentially talking about languages. We've already set that up, English and Chinese, but also the multitude of the world's languages. We want to discuss how languages are shaped by a modern technology-driven world and how challenging it can be to link them up to the keyboards that I want to say we know and love. But I'll just say that we know right now that our standard part of our lives. I wanted to start by asking both of you about your own exposure to languages, what you've grown up with or pursued in terms of developing your written and speaking skills? And maybe what you find fascinating about that. Who wants to go first?

Gosh, guests first.

I'll take it. No, so I've often, I guess the simplest way to put it is, the reason that I ended up studying Chinese in college was because Arabic didn't fit in my first year schedule, my big plan was to study Arabic, Chinese, Russian. if, I don't know, a genie in the bottle were to present itself undoubtedly, one of the wishes that I would ask for is the ability to speak and read all languages. I've always just gravitated. That and play all instruments. And I've just always gravitated towards the study of language. So, it's I've had maybe the chance to study off and on about a dozen languages, some more intensely than others, but it's infinite. I just love the realm of language, and something about the written word in particular, just I've always gravitated towards.

That's great.

Paul, what about you?

Well, I'd say a couple of things. I, One is I grew up on Vancouver Island where there's definitely only one language. It's very kind of rude-

What language is it?

Mostly English in a very slow, laid back manner. But I eventually got into languages and I got into French in university and did an exchange even in Chicoutimi, part of Canada, and then ended up studying in French actually in Geneva. And I did actually study for a brief while Chinese and some other languages, but I learned very little and it was actually the spoken language only, but I would note two things. One, in learning another language, it does open up a little bit of the brain that is otherwise not tapped into probably physically. But also, you learn these little cultural insights through another lens that is like, this is a word that doesn't really exist in English or in your language. It has a slightly different meaning. And so it kind of opens a window there that you don't otherwise have access to through your own language.

And then when you look at the written form of Chinese, it is just so different from English. Where we've got 26 letters. I looked up a stat and it was like, you need about 2,000 to 3,000 Chinese characters to read a book, like a typical book. In English, you actually need about 8,000 or 9,000 words to read a book. You need a lot of words. So, in a way that the Chinese characters are more efficient, but they're very complex and you need a lot of them. They're just utterly different things and it's hard to imagine how something that couldn't be more different in a certain way. But fascinating stuff. I agree with Tom that it's so cool to get into languages.

There's something about studying languages that also makes the familiar unfamiliar about your own language. I mean, take the word dinosaur, okay, we've said it however many half million times in a lifetime. It's just very second nature. And sure, we learn the etymology of it or Tyrannosaurus Rex or whatever it might be. But we kind of give it a pass. We say, "Whatever, that's my mother tongue, I know it so well." And then you learn another language and people, especially if you learn Chinese, people will revel over giving the literal definition of something. So computer, electronic brain and people, they'll really celebrating kind of a little bit fetish eyes or sort of other eyes that as well. But if you actually think back to your own language, you realize that you're surrounded by all of these words that if you actually thought of the etymology of the word, they are equally kind of funny if you stop and think about what something means.

That's the other thing too, is you get a very different perspective on home, whatever, either literally or figuratively when you through language take a step beyond passport control, so to speak. And then come back home and suddenly you actually can't come back home in some sense. the home that you took for granted is no longer there and you get to think reflectively critically about it for maybe the first time. Something about language learning, opening up your brain, but also opening up home to a different view.

Totally, totally. And in fact, the word home in many languages is very revealing, like how they say the word home, but that could be a whole separate podcast. We here, we're looking at your book and as Vass said, kind of a trilogy of your work over a number of years. And it's been described as the first history of the Chinese language in the digital age in many ways. How did you get focused on this idea of looking at the history of digital technology in China as just being the thing to look at and drill into as so fascinating?

These three books literally started as one. They started as one book contract until the day I wrote to my editor and said, "How about two?" And my editor said, "How about three?" And I said, "Deal." And so, I had really stumbled across the question of Chinese in the, let's say the 19th, 20th, 21st century information age. And this question of the mismatch that every game-changing language technology from electric telegraphy onward, from Morse code onward, typewriters, Remingtons, and then into computers, personal computers, optical character recognition, whatever the big list has taken English as its starting point, has taken the Latin alphabet as its starting point. And this has a mixture of history. It's the history of American capitalism, it's the history of European colonialism. It's a variety of things.

But one thing that's true is this world, this digital world was certainly not built for Chinese or with Chinese in mind, a non-alphabetic writing system, 100,000 characters. And so I've always been fascinated by the history of print, the history of technology, but it was probably, I guess about 16 years ago now, when the question occurred to me of, have I ever seen a Chinese typewriter? I assume that one existed in the way that a French typewriter might exist or a German or so forth. But come to think of it, I've never seen one. And so, fast-forward 45 minutes and I'm lying, literally lying on my belly like a teenager on the phone with a new love.

I love that image of you.

Oh yeah, I was just lying on the floor of my office at Stanford looking at the Google patent document database.

Yeah, twirling the cord.

Exactly, exactly. You remember, I remember. That's exactly the feeling. And it was just this mixture of certainty. I know in some way, shape or form, I'm going to work on this, and utter uncertainty of, I couldn't put it intelligently as to why. I just know that this is what I'm going to do. And then at some point along the way, I realized that I had to split up the family in order to keep the family together. There was, to go from Morse code all the way to ChatGPT turns into one of those books that's, as the Chinese saying goes like, "Viewing flowers from horseback, just racing past it all." And that's what I wrote to my editor and I said, "Listen, I really do think that this has to live." And she was a big supporter. She said, "Absolutely. How about actually a three-book contract?" And that was-

I love this editor.

... the beginning. Oh, she was Katie Helke, then at MIT Press.

Nice.

Just a force of nature.

Policy Prompt is produced by the Center for International Governance Innovation. CIGI is a nonpartisan think tank based in Waterloo Canada. With an international network of fellows, experts, and contributors, CIGI tackles the governance challenges and opportunities of data and digital technologies, including AI and their impact on the economy, security, democracy, and ultimately our societies. Learn more at cigionline.org. Just wanted to ask you about a phone call you made to someone else, maybe also while you were on the floor of your office. But there are so many fascinating characters in this, I don't want to say story, but I did right, history, and there's a moment that of course I found particularly interesting. I always knew someone would call. A particular innovator says to you on the phone, what was that like and why did it take so long for someone to call him?

Thank you for that question, that's a really good question because, so the gentleman in question is the late Chan Yeh. He's a Taiwanese American engineer who co-founded the Ideographics Corporation right down in ... Was one of the first office complexes in Silicon Valley when we're still in the transition from orchards to Silicon. And he is the closest figure in the story. Maybe there's one other contender who at his apogee was like the Steve Jobs of East Asia of Taiwan certainly. The systems that he developed, the computational photo type setting system changed newspaper publishing. It utterly changed the tax administration, the yellow page industry communications and so forth. But when I finally, I'm a little bit of a researcher stalker. Once I figure out-

Being a researcher is a good excuse to be a stalker though, it's a good pivot for inner stalking techniques.

Well, you find someone's phone number, and what are you going to do? Are you not going to pick up the phone? So you pick up the phone and you say, "Hi, my name is, and are you by chance or do you know by chance?" And then you kind of take a gulp and see where it goes. The first words out of his mouth, I always knew someone would call. When I met with him his office was abandoned. We walked through the facility and the way I describe it in the book is it was like the opposite of Pompeii. It was as if everyone had just stood up and just disappeared. There were technical manuals still open to certain pages. It almost looked like someone was going to come back from the bathroom. But at one point this was the beating heart of an absolutely game-changing company in Chinese tech. And then like many companies, it was eventually overwhelmed by the software revolution.

So there's tons of neat stories in the book, but let me situate two things to help the listener a little bit here. So we're going to back up through some of the technologies, but before we do that, what does the current computer look like in China right now? The typical computer that somebody would use is their-

I've got it in front of me right now.

Yeah. So what does it look like?

It's MacBook Pro, it looks, you name it, I've heard of it. If you have a Windows machine, it's a Windows. If you have a Dell, it's a Dell. If you have a MacBook Pro, it's a MacBook Pro. So now we've reached this really interesting moment in history where either there is no such thing as a Chinese computer, or every computer is also a Chinese computer. But that is the result of phenomenal series of kind of failures and innovations and eccentric personalities. But now if you go to the Apple Store, you go to the Best Buy and you pick up an off-the-shelf machine, you bring it home, you boot it up, all you need to do now is go to the settings menu, typically to international or keyboard, whatever the menu is, and you simply check on not just one, but a variety of Chinese input system.

But for that matter-

Farsi, Hindi, Cherokee, I mean the list, there are about 100 highly resourced languages out of the 7,000 alive today. But now every computer is an Arabic computer, is a Chinese computer, is a Devanagari computer. And the interesting and somewhat nefarious from an historical perspective is, Apple gets to take credit for that, or Dell and IBM. And it's almost as if, of course the Catholic universal computer, of course it's all things to all people, but what that actually represents is a very long grueling and hard-won battle by a whole transnational cast of characters who eventually got to this place where every computer is a Chinese computer. So, in their success, in a weird way it's become invisible.

That's cool. And yeah, they're definitely taking all the credit when they can, and they do, right, as you say. But so just to situate again a little bit something here that, wait, one of the tensions in the book, tension is the right word or not. But that there were all these technologies that came along, and to some degree it provoked a little bit of a crisis in China, whether it was the telegram, the typewriter, but it also provoked innovation at the same time. And you talked in the book about the character amnesia now being maybe a fear that, if I understood it correctly. The machine kind of guesses where you're going next and says, "I think this is what you're thinking." We see that in English as well where emails kind of finishes the sentence for you. And so is this still real for Chinese and other languages where this amnesia is a concern, but at the same time you've now got this device that works in your language. How is this tension playing out?

Let me just to back up a second. When you open your MacBook Pro or your Dell and you set your keyboard to Chinese, your computer will behave in a way that will be utterly unfamiliar to an English language speaker. So what will happen is you set it to Chinese phonetic input. And what you have done then is you have turned on what is known as an input method editor, an IME, which is this really interesting kind of software application, because it operates semi-invisibly between the user and whatever application the user is really using, a web browser, Microsoft Word, whatever. And what its job is to do in the case of Chinese, but also Japanese Korean, many other, is to intercept every keystroke that you enter on the keyboard. And instead of just sending that symbol to the screen, the way that the cat bit, the dog is just sending those symbols to the screen as soon as you touch them, it does work on those symbols. It uses those symbols, all of the keystrokes, to basically say, to put it anthropomorphically, what character is this person after?

Given the input they've given me, do they want this Chinese character or that Chinese character? Let me make my best educated guess as to which one they want and let me show them a series of options. And that's there's a pop-up menu that follows along, and then you choose the one that you want.

So that's a machine learning algorithm?

It effectively is a machine learning algorithm, because when you turn on the computer, like say you literally boot it up, you cold boot it for the first time, it's an air-gapped beginning computer. The guessing algorithm inside that input method, it's going to be basically just raw statistics. As in English, as in Chinese, some characters show up all the time, the, you, me, and some don't. Azerbaijan. So Azerbaijan is going to be much lower in the list of probability. But let's say that you're a consulate, an Azerbaijani consul, and you are writing the word Azerbaijan over and over and over again as part of your daily business. This is where the learning algorithm comes in and starts to promote that possibility over time, because the whole goal is try to juice every possible efficiency out of the system. And so, this is true, I'm not just talking about the last five, 10 years. This is how the first Chinese computer in 1959 was already built towards-

Which, is bonkers, honestly, it's kind of bonkers, I think.

Predictive text, auto-completion, all of that stuff has been part of Chinese computing now for 60 years.

Wow.

And in the case of English, the science of it has been around for a very long time, but it's in terms of a user-facing thing, it's what, maybe two decades old at the most in very limited fashion.

At most. Yeah. So, is that why typing in Chinese on a computer right now is much faster than in English?

Yes. Yes. That's why. That's the top speed is higher in Chinese than English because of-

I mean, just putting that in perspective too, right? Adding complexity to the interface makes it also faster. That's counterintuitive to me. And you say in the book, you offer words per minute speed kind of loses its meaning because of the many different interfaces. And of course, it feels like there's a natural link to some of the generative technologies you mentioned. Where does that situate the technology in relation to the speed of our human thought kind of writing and typing?

Well, I mean, so let's say that for most of the story of the book, I mean the book is really, if you had a timeline, it's failure, failure, failure, failure, failure, failure, failure, failure, failure, juggernaut. I mean, it's really this hockey stick at the end of the story. For most of the story, even with auto completion, predictive texts sort of learning in some sense, it's still slower. It was always comparatively slower. But here's the thing is that because, and this speaks to the issue of crisis, because there was this crisis of, how do we catch up with the speed of the alphabet and how do we compete with the speed of the alphabet that every conceivable time-saving idea was played with?

And then eventually, because I mean, literally all of us on our desk right now we have machines that according to the categorization system of say the 1980s U.S. Government would be considered a supercomputer that couldn't be sold to the communist block. I mean, we've got supercomputers on our desk right now using '80s and '90s definitions. In the case of Chinese input, Chinese input method editors harness that computational power. And when we switch to English, we basically have a glorified electric Remington typewriter on our table in terms of-

It's true yeah.

... how we interact with it. And that's the reason why the ceiling, the total ceiling, you take the two Olympic competitors of English typing Chinese input, the Chinese input will always be higher, because it's got computational power on its side. But because of that, the way in which 800, 900 million internet users and over a billion computer users worldwide and the Sinophone world more broadly are using, technically speaking no Chinese-speaking computer or new media user ever really touches Chinese in order to make Chinese. Because the QWERTY keyboard is ubiquitous in the Chinese-speaking world. So, the actual thing that a Chinese computer user is touching when they input are the letters of the Latin alphabet. And so, we have this interesting phenomenon now of, and this speaks to the character amnesia crisis, is people are wondering, "What does this mean?" Chinese still exists, it's thriving, it's thriving in the digital age, an enormous number of people on, Chinese speaking people are actually interacting with something quasi Latin alphabetic or quasi English at the same time. It's the first time in human history we've seen anything like this.

How do you type Chinese characters into a keyboard designed for the English language? Today we're joined by Thomas Mullaney to discuss his recent release, The Chinese Computer: A Global History of the Information Age. Mullaney's new book charts the complexities of integrating Chinese characters into early computing systems, offering a captivating exploration of the obstacles and innovations that contributed to China's role in the global digital revolution. You can purchase The Chinese Computer from MIT Press online and in bookstores now.

One of the things that you draw out in the book is this concept of hypography, if that's the proper pronunciation of it. But it almost never comes up in English, which is why I don't really know how to pronounce it. But can you give us an example of when this is used in English and just so that people can get their heads around what it is and why it's important and maybe it should be used more in English?

Yeah, it is a neologism, which normally I run screaming in the opposite direction of. So it was a decision to try to coin a word, but the Palm Pilot, if you remember those days, the palm alphabet-

I remember those, yeah.

So if you remember that the palm alphabet, that's a hypographic system. Because the goal of writing each of those symbols is not to write those symbols. It's to enable the computer to retrieve what you want, the letter A, B, C, D. And as soon as the computer or the device retrieves what you want, the palm symbol disappears. The other also in IDEs, various developers, obviously they don't just open a text file and start coding. They'll use various whatever development environments, and as they're typing it is suggesting different programs or libraries or so forth, and you can tab and it will just auto-complete and so forth. That is a hypographic kind of framework.

The biggest and most popular counterintuitively goes all the way back to the 19th century. Court Stenography. Think of the court reporter whose job it is to keep a transcript of the trial. There is no court reporter that uses a Remington style typewriter, because there is no typist on a classic typewriter who can keep up with the speed of speech. And so they use a special device called a Steno typewriter in which you are, it's a little bit complex to describe verbally, but effectively what you're doing is you are typing in code. Objection, overruled. The court rests, may everyone rise, whatever. To type the word overruled or objection, overruled, even a phrase is a single keystroke where you depress multiple keys at the same time. Long story short, the machine produces a transcript that if you just looked at it looks like gobbledygook, it's unintelligible. But the Steno typist knows that, "Oh, these two letters, if they appear at the same time, represent a different letter or represent a consonant."

What ends up happening is that the court reporter gets this tape that comes out of the machine, now it's digital, and it has to be translated into natural language English. Just like in Chinese input method editors, there's an extra step, and that would seem to make it slower, but it has this huge value added, which is if I'm capturing whole phrases with a single keystroke, I can easily keep up with the speed of speech. And so, we've actually seen hypography in English for a long time, but it's always been at the margins or in specialized work, let's say.

It was necessary to keep up with the environment that it was in. But if we need it, we'll make it happen. In a way there's no barrier.

There's no barrier. The barrier is, English is, I guess you could say that English is due for a good crisis. English has not really had a crisis along the lines of the Chinese language in modern history. There has been nothing approaching the crisis that confronted the Chinese writing system and language say in the late-19th, 20th century. And if there had not been that crisis in the 19th, 20th century, there is zero reason to believe that the average Chinese computer user would want this particular form of human-computer interaction. It is really crises born.

Are we heading into a new phase now where people use more characters, emojis, touch pads, symbols to express themselves and that the aging population is going to need simple interfaces? Or were we moving to a new interface revolution anyway, maybe even in the English-speaking world?

Well, the one, I mean classically, the three areas I guess where really mind-bending work is done, serious work is done in interface design is disability technology, non-Western writing systems, and then let's call them hyper-limited context. So, having to use a remote, a single-button remote control to enter the address on your television, or I've seen-

Stephen Hawking the way that he was interfacing.

[inaudible 00:30:07].

Yeah.

That's the origin of T9 input. T9 input before it was on Nokia phones was developed as the so-called the owl, which was meant to be worn as eyeglasses on the user's head, and then they would use eye-interrupt to select between, I think it was eight options around the rim of the glass. And then that information, eight letters is less than 26. So you feed that into effectively a disambiguation algorithm that is trying to make educated guesses against a back end of a library, an English language library encoded in the computer, and trying to use probability and contextual clues. And then along the way, the developers of that system said, "This is also of potential use to sighted individuals or people with full use of their motor functions." And that's the birth of T9. T9 began as a disability technology. So that's just sort of an interesting exception that proves the rule, which is by and large interface design, I'm going to upset some of your listeners, interface design in the anglophone world is, I don't want to insult anyone.

Boring.

Just say it, basic.

It's a tiny square foot plot inside a vast terrain of possibility. And it's very hard to imagine what's going to break it out of that, because it's not a technical problem that. The tech is easy, it's old. We know how to do a lot of this stuff. It's that the average mainstream English language user has been fed on a steady diet of, "You're the bee's knees, your language is the greatest thing since sliced bread. What could be easier than what you type is what you get?" So of course, why would they suddenly want to do all of this extra bells and whistles, even if it is faster? It's a technological culture problem now. And Silicon Valley just doesn't get that. They think it's a tech problem. They think it's like, oh, and they just keep reinventing the same stuff thinking that that's going to be the breakout hit.

Social media influencers, our feeds are full of them, and does it ever look like a fun gig. But what's behind the carefree lives of these glamorous entrepreneurs? In a previous episode of Policy Prompt, we went behind the scenes with authors Grant Bollmer and Katherine Guinness to explore the profound impact of influencers beyond the glamour and hashtags. We discussed their book, The Influencer Factory: A Marxist Theory of Corporate Personhood on YouTube, where they unveil a striking new era in capitalism, the Corpenstine, where individuals morph into living corporations. We talk Mr. Beast, Mia Maples, Jeffree Star, and other non-human or sub-human influencers as we dive into the murky waters of modern economics that will reshape how you view social media and society. Check out that episode of Policy Prompt with Grant Bollmer and Katherine Guinness wherever you listen to podcasts. I mean, I actually didn't know the details of how a steno typewriter works, so I appreciate back there when you broke it out. I was like, "That's what they're doing in court."

That's awesome.

And I do like to mispronounce at least one word every show so I can cross it off my list. I wanted to zoom out for a second a little bit, just thinking about that history of innovation, that sweet spot between concept, proof of concept, being able to go to the market, all the stages that we sort of glamorize maybe when we're reinventing keyboards. In your book there are those early really stressful elements of a system being in development. There's prospective funding, communication going cold. There's the importance of gathering in-person once a year at a fair convention to check in on other people's innovations and also show off what you are working on. In startup culture now it's so normal to have angels and venture capitalists support the development of a product or service. Could you talk about any angels or their absence during this history of the Chinese computer?

Oh yeah, that's a really great question. Yeah, because the era of the Chinese computer and also the kind of antecedents to it is really the beginning of the birthplace of R&D, Bell Labs, like the corporate academic connection. It's not an accident that Claude Shannon and information theory is born at one of the first kind of academic corporate think tanks. And this is really that era. And so, I think of figures like Gao Zhongqin in the story who is based at IBM, sort of goes ... Well, he's in interaction with IBM, Lin Yu-tang is in interaction with IBM and Mergenthaler Linotype.

There's some elements that are the same where they need the prowess, the distribution channels, the precision engineering facilities of Mergenthaler to do this, or of IBM to do this. I mean, for any, I'm sure listeners will know, but Mergenthaler Linotype is a force of nature at this point. It's just with hot metal composing. They are the pallbearers of the Gutenberg age. They put Gutenberg in the ground. So you have outsized influence of a relatively small number of actors, you know, IBM, Mergenthaler, Olivetti and some others. And then you have this loose assemblage of inventors and eccentrics and entrepreneurs who are, they're kind of lone wolves in one way. They're in their room trying to figure this stuff out. But all roads lead to Rome at the end of the day. They have to get the likes of Thomas Watson Jr. to sign off on something.

And so I think you're referencing, especially the chapter on IBM. I found the internal private memoranda that IBM and Mergenthaler people are sending back and forth about their meetings with these Chinese inventors. And you just realize how precarious this is, that with a flip of a switch Gao's dreams are done and there's very little recourse if they said, "No, we don't think that this is where it's at." That stays true for a while. I mean, I would say through a good portion of the Cold War, I mean really maybe until we get to the '70s and where it's possible now to find other outlets, say in the East Asian, Japanese and Taiwanese context, when Japan is rebuilding its economy in this more tech focused way, consumer electronics way. And then only until we get to the '80s. Once we get to the '80s, I would say then the number of doors that a particular person can knock on to try to get that support starts to mushroom.

But the angel investor, this other rogue individual is, you don't see many of those figures at this time. They are all housed inside one or another corporate kind of or governmental framework. There's very few, I've personally come across very few who are hovering semi-independently and saying, "Okay IBM, okay, Gao, if the two of you get together, I will back this." That seems to be a late 20th and early 20th century 21st century thing.

One of the things I liked in the stories that you were telling in the book and the history was the back and forth on either side of the Pacific Ocean. In the United States you'd be talking about some inventor. I didn't expect that much there. It was mostly in the pursuit of expanding markets, selling computers that was driving the innovation. And then on the other side was in China this crisis of, how do we not lose the Chinese language or weaken it given the importance of standardized Chinese in the model? A fear of weakening that. And one of the stories that you told that was particularly fascinating was the journey of Xi Bingyi and the long detention during the Cultural Revolution to then becoming a leading developer during the Nixon reach out to China and all those photos and things.

But I couldn't get out of my mind that image of him in his solitary cell staring at the only words, if I understand correctly, that he had access to, because he didn't have books, leniency. It said, "Leniency is for those who confess severity for those who resist." And he stared at this all day and night and tried to think about how that would work in a computer. Is that right?

That is right. With one tiny adjustment, which makes it even in many ways more interesting is he was thinking of it, the computer part came second. And in the beginning he was thinking really of filing cabinets and library card catalogs and dictionaries, and he was thinking in order to maintain his sanity and his also sense of self and just boredom and et cetera, terror, horror, et cetera. But he was part of a deeper history, which I actually talk more about in The Chinese Typewriter of a thing dating back to the 1910s and '20s. Which is about the nerdiest thing you could possibly imagine that is incredibly important. It was called the character retrieval problem or the character retrieval crisis.

This was an episode that expanded over many years in the teens and '20s, mainly librarians, educators, publishers, teachers, and linguists. And what they were saying was, "Listen, we've got a major problem. Because when one of our students goes to the library in China and they're trying to find a book in the catalog, it takes them 45 seconds to find the card, because we don't have alphabetic order. We have all of these different systems for how you look up a character in a dictionary." If they're in Germany or if they're in English, or if they're in France, they find it in five seconds. So if we do the calculation, if we extrapolate this 40 second delay across every instance of information retrieval across every user across time, we are literally behind the rest of the world informationally. We are operating. It's the reason that high-performance traders want to be closer to the index because they're literally slower if they're a few blocks further away with.

And so, it prompted dozens and dozens of people to try to imagine a new way of organizing a Chinese dictionary, a Chinese card catalog. And most of this stayed there. But two people in particular, one is Jibing Yi and this other Lin Yu-tang, they made a leap. They made a leap and said, "Okay, wait a minute. What if we take this idea of information retrieval, finding a book and a catalog, finding a name in a list, and we migrated it to the realm of inscription, of writing technology." That's the birthplace of input, of input method editors. What Chinese input and Japanese and Korean input what they are, and indeed what ChatGPT is, is writing for the purpose of finding, not writing with the purpose of inscribing.

The inscription comes later. The inscription comes only after, "Yes, that's the character I want. Please commit it to the paper, commit it to the screen." And so, really the way that an average Chinese computer uses their MacBook Pro is as an information retrieval system, not as, again, a glorified electric typewriter that's just meant to commit ink to a page. And so Jibing Yi was on the verge of probably insanity at any given moment and confronting this poster, propaganda poster. By the way I describe it in the book, is he turned his mind into a kind of particle accelerator and just was smashing Chinese characters together into pieces.

And then saying, "Okay, what if we take these different pieces? Let's assign a letter of the Latin alphabet to each one, and let's come up with a kind of Latin alphabetic code, a four letter code for every Chinese character." But he had a dictionary in mind, but then when he got out of prison, he was still under house arrest, but when he went out of prison, things were changing, the question of computing, mini computing, micro computing, and he also made the leap. He said, "Wait a minute. That thing that kept me from going insane about dictionaries is actually what is needed. Because look, how else are we going to find Chinese characters using a QWERTY keyboard? My system might work." And the rest, you could not imagine a trajectory through life like his, it's just the non-linearity of history embodied.

You mentioned invisibility earlier, and we sort of touched on it in terms of the, I always knew someone would call. Can you imagine? Can you imagine being the one to say that to someone else too? You also point out in the book that all of the large scale devices that were prototyped, that were built and that were developed aren't really in any museum collections.

They're on the bottom of landfills or scrap for parts.

Why do you think that happened? Why are we losing this history of product development and linguistics advancement? I mean, I'm so appreciative in your history that there are, I will say I love a few pictures in a book.

Yeah, me too.

You bring forward the kind of, I don't want to say propaganda, but the advertising, the advertising for these things that really helped me as the reader to imagine it. But wow, the fact that we can't see them anymore, we'd have to, I guess just use a 3D printer to try to have a proxy. How does that make you feel?

Well, I try not to let it cut too deep, but of course, so on the one hand, at a large scale, this is global. Is we do a miserable job, just humanity does a miserable job of keeping the records of failures. We do fairly well of successes that became failures like Wang Laboratories or something like that. The papers are in the business school special collections at Harvard. So it's a case study in business school there and stuff. So when something reaches a peak and then collapses, maybe we document that, but whatever part of the portfolio of an angel investor's portfolio that doesn't make it, that stuff doesn't make it into the archives generally speaking. And prototypes have a miserable track record for making it into the archives. And this isn't just, I want to underscore, this is not because I have some sort of fictional idea of a world where nothing is lost. It's more that if we are just constantly archiving success, it is a completely distorted view of how this actually works.

How innovation actually works, how change over time, the history of technology in particular it ... I love different various histories of technologies. I love a good explainer video on YouTube, but it's really a march through, well, first we had this and then they had this, and it's a stairwell that you move through. History is not a stairwell. It's this collision course that's exploding, it's churning and out of it come these moments. But we do a very bad job of recording in general. China has the additional problem where modern Chinese technology is systematically denigrated. It's not taken seriously, including in China. It's not just, the Chinese typewriter in its day and age was understood as kind of a absurdity, a silly, a monstrosity. And so that naturally, I think reverberates into, there is no museum of Chinese typewriting. There are dozens and dozens of collections on the Western typewriter all around the world.

Same thing with early computing. It was always as sort of the laughable cousin of the real story, and therefore, why bother with this or that? And we have lost irreplaceably in many cases the history we would want to have at our fingertips.

Yeah, it's on so many technologies and inventions. I'm not that sad there's no prototype Palm Pilot Museum or those kind of things.

Why?

Well, because there will be just too many.

Why? Did you just feel like you got scammed by one?

No, I had a Palm Pilot, but it just didn't last long enough. It was replaced by something else so quickly. I'd just gotten used to it. I'm still kind of mad about that, I kind of liked it. I was using it. I did. And then it was gone. That thing is ancient. Get rid of it.

We have time for a couple more questions, and one thing we wanted to make sure that we asked you was about, you've had a lot of side projects over time. You squeeze a lot of hours-

We say that with admiration, not be like-

You squeeze a lot hours.

... you're doing some other random stuff, but there's innovator. You've got some failures on the side. Sorry, Paul, I keep teasing you. This happens. This happens.

Just hanging out. Yeah, he's just hanging out with all his extra time. But one of the things that you have been working on is this big project to assist and preserve digitally disadvantaged languages that are declining or on the verge of extinction in an area of growing dominance of a few languages, especially English.

Yes.

What the question for you is about, how many languages are we talking about here in this kind of disadvantaged category? And is technology both the friend and foe? But maybe the most interesting thing of all would be can you give one or two examples of a language that you're working on or thinking about that's in this category that may be salvageable and savable and promotable?

Yeah, thank you very much for that. Yeah, so the project, it's a presidential initiative at Stanford called Silicon that just got off the ground. And in many ways is the culmination of the exact type of work that I know I've been working on for about 20 years, and my co-PIs as well, Chinese was really one of the top three, the first digitally disadvantaged language alongside say Arabic and Hindi perhaps in terms of really large world languages that were systematically blocked from technologies, until they weren't. But then it's very new waters because it's not history anymore, it's the present and the future. But no, I mean by most estimates, there are 7,000 plus living languages, not including historic scripts. And the Unicode Consortium, the Unicode is of course the gateway to text in digital spaces as this non-profit, international standard-bearing organization with character and coding, they categorize 90 plus percent as digitally disadvantaged.

So we mean we're somewhere in the order of magnitude of 6,000, 6,500 languages that to one degree or another is either completely borrowed from digital space, it's not even encoded. Or you could post a blog about it, but there's no user interface, user forget auto complete machine translation, nowhere near large language models or AI. And yes, so it's a dance, because technology is both the accelerant of a lot of these pathways, but also is something that can be leveraged, and of course is the pathway into culture and arts and global economy and so forth at the same time. So, it's a double-edged sword, and we have to ... We're finding ways to navigate that. But I can give you an example of, in our pilot phase we've launched a practitioner program and a student internship program now, and then we're going to be ramping up things as we move along. But our view is, I don't work on character encoding. I don't design fonts, I don't design keyboards personally. But we are tapped into and connected into these amazing ecologies of practitioners and experts and language experts who are.

And we're trying to identify and connect and support and augment their work, and in some cases help them overcome certain kinds of blockages, so that the work they're in many cases already doing goes from taking 15 years to 10 or five. Because the honest fact of the matter is, one of my colleagues put it and put it very bluntly is if no one does anything, language extinction will take care of this issue on its own.

And so there is a group that we're working on, I'll give you an example. There is a large language community predominantly in sort of the area of Iran, a Kurdish minority, the Locke. And by and large, the majority of letters that they use in their writing system is identical with their neighbors. And so it was already in the set, but we are collaborating with this amazing type designer, this Iranian American type designer, who is developing just a small, what they call a glyph extension. It's basically just a few more letters or symbols or glyphs that can be, we can augment the existing font and more or less like voila, suddenly a fuller scale participation in the digital space by speakers of Locke is possible. Because you can't include what you can't see. If you don't have a font for something, there's no inclusion.

We are working with another colleague who's doing something comparable, developing a font that works with this consortium of ethnic groups predominantly in Ghana. And we're talking about a community of maybe three, four, five million people, and he's one of the first, he's a Ghanaian type designer, and he's working on one of the first digital fonts that would help sort of span these ethnic groups in the language groups. And a type of font, it's serious work, but at the same time you could think of it, certain kinds of font as like an MA thesis. It's something that takes multiple years, but with concerted effort and with free time it's not something that takes 20 years or 10 years. It's something that is very achievable and within range. But here's the problem, and then I'll mute myself. Here's the problem is that everything we've talked about, Chinese especially, that comes from an era of the '80s and the ;90s of software internationalization where the big goal was market share.

I mean, we're talking about the Chinese market, we're talking about Indian market, we're talking about the Middle Eastern market. So there was a ton of interest in overcoming these forms of exclusion. But what's been left behind in this wake is predominantly these 6,000 plus languages, to be honest, for whom no one is coming urgently.

There's no market driver, there's no market driver.

There's no market driver, because each individual language as a whole, if someone's sitting and looking at the bottom line at corporation X, Y, or Z, is saying, "This on its own doesn't justify a massive outlay." And so it's predominantly been left to the actions of brilliant, incredible committed, but very resource restricted designers, keyboard designers, font designers, character and coding proposal designers. Or to piecemeal sort of philanthropic efforts, which are amazing. But again, pale in comparison to the story of the '80s and '90s. Because it's hard to internationalize, it's hard to do these things. It takes a lot of concerted effort. That's really what at Silicon we're trying to think through-

Thanks for telling us that. Yeah, thanks for telling us that. And if people want to find out more, I think they can find you and find links to that project and-

We'll sneak them into our show notes too.

Sure, definitely.

Maybe we can conclude here. I mean, you end the book. I'm sorry that this is a bit of a spoiler, I guess, by saying that, "Writing has changed, our frameworks for understanding it must change as well."

Yes.

How is writing changing today?

Well, I do like giving away, because people can ... I don't mind if someone-

Sorry, we can delete it if that's annoying. I did that. It really stuck with me.

No, I like it. I like if someone buys the book because they want to read the full argument, but I do want them to know the argument.

If you want to know more, you have to buy the book.

But the simplest way to put it is the keyboard base computer and the whole idea of human computer interaction was again designed with the English language in mind. And it was designed really with the typewriter in mind, the electric typewriter. The keyboard that's in front of all of us right now was "meant to be used in a particular way." More than 50% of global computer users use this same keyboard that's in front of all of us right now in a way it was not meant to be done. That's crazy. They sort of hacked it globally over the course of a century, hacked it.

But because the keyboard looks and feels the same, if I'm in Shanghai or Tel Aviv, or if I'm in Beirut and I see someone at a cafe pop open their laptop, it is totally possible for the English speaker in me to be deluded into the sense of, "Ah, they have one too. Good for you. You also have this universal machine in front of you." But lo and behold, if I actually looked a little bit longer and paid attention, I would notice, wait a minute, this computer user is using it in a way I don't use it. And so quietly, the dominant model for human computer interaction has actually, it's not changing, it's already flipped.

But English speakers, Latin are still operating in this sort of sleepwalking delusion that it's the same as it ever was. And so that's why I call it the ... I make a cheeky remark. I call it the Incunabula period, which is a technical term for the first 50 years of the Gutenberg Revolution. Remember, when the Gutenberg Revolution, they didn't have terms for this stuff either. They were in the first 50 years. They're like, "What is all this stuff?" We are in the first 50 years of something, a different epoch of writing, and we're still using terminology from the pre preceding era to describe it.

Well, thanks so much, Tom, for joining us-

Thank you very much.

... today and for walking through all this with us. Fascinating. I think it's opened our minds to a lot.

Thank you.

I very much appreciate it. Thank you.

Paul, I want to be honest with you. I mean, I'm always honest with you, let's be real, but-

Let's hope so.

... there were parts when I was reading this book where I was freaking out a little bit on the inside because of the technicalities. Because of how deeply researched it is, and I didn't know how we were going to bring it to life together, but I had so much fun talking to Thomas.

Yeah, it was great. I mean, the book is tough sometimes when it talks about the actual Chinese characters and then the actual technology, and you're like, "That's two things that are really hard to put together." But it does tell a fascinating story with lots of characters. And that bigger picture of this thing has evolved. In a way the bottom line was, all computers have become that Chinese computer that now opens up the world of the main languages, but not the small ones that are still totally excluded from that whole universe. And we're probably moving to much more creative interfaces over time, but there's a straight jacket around the dominance of this little keypad, that letter pad that we have in front of us.

Hopefully. It's always hard to imagine that change. It's hard to imagine anything better than what we've been mostly used to, despite you and I retiring our Palm Pilots. But I also really appreciated Thomas talking about failure. It makes me think. There was an art exhibit once at the Toronto Harbourfront, the Museum of Failure, and it was a traveling kind of exhibit. But just showing failed attempts at innovation. And it was very charming and sometimes funny, but also a reminder that we're always trying to build interesting things, and sometimes it just doesn't work. But sometimes you also need support to make that work.

Yeah, totally. I mean, if you're truly innovating, you're failing a lot of the time. And so he articulated a lot of that. A lot of those have been lost to history, like that extra step where somebody tried this and didn't work out and it was-

Heartbreaking.

... and the machine's gone. The actual prototype is long gone, right? But it was striking to me how many there were. I had no idea that there were so many steps and so many back and forth iterations. But it makes sense when you think about it, right?

Of course.

There had to be so much work to get to this magical step of integrating it all.

I also appreciated his sharing a kind of accidental entry point into something that's obviously become a sort of feature or fascination of his career. And I think I felt there were parallels to innovators and disruptors too. In terms of how you are just, there's just a pull for a particular project or initiative or topic that you just can't get away from.

Yeah, totally. I think you were talking about a Hollywood script as well coming out of this.

Okay, this is true.

This thing could be a movie, right? The characters and the ideas.

I don't mean to be an emoji girl, but I mean, Apple should buy the rights to this. This is a miniseries. This is, it's fascinating. The characters are compelling. It's stressful, there's tension, and it's not a 12-month race. Someone tried to caution me the other day. They're like, "It's a marathon, not a sprint." And I kind of thought, honestly, marathon seems a lot harder. I'd rather just sprint if I could. But framing that marathon, but keeping that tension is kind of cool.

It's a mystery and it's spoiled, because you know the outcome. He is a Chinese computer, a computer that works fully Chinese, but getting there is not obvious. It's like those movies that start with the murder scene and then work backwards. But it's a long journey. Bit of a mystery to get there.

Well, this was a great pick. I would not have touched this book otherwise, and that's on me, so-

Shout out to creative director, Som Tsoi on that. The idea of this book. Som, thanks a lot for finding this one.

Thanks, Som. Thanks, Paul. Thanks everyone.

Thanks.

Policy Prompt is produced by me, Vass Bednar and Paul Sampson. Tim Lewis and Mel Wiersma are our technical producers. Background research is contributed by Reanne Cayenne, marketing by Kahlan Thomson, brand design by Abhilasha Dewan, and creative direction from Som Tsoi. The original theme music is by Josh Snethlage, sound mixing by François Goudreault. And special thanks to Creative Consultant Ken Ogasawara. Please subscribe and write Policy Prompt wherever you listen to podcasts, and stay tuned for future episodes.