A Writer Who Never Left His Word Processor
There is something wonderfully anachronistic about imagining Seigo Matsuoka at work in the 2020s. Around him was a sophisticated intellectual and publishing organization. His Senya Sensatsu, or A Thousand Nights and a Thousand Books, had long existed on the web. He was surrounded by editors, computers, digital photography, online publishing, and the information culture that he himself had spent decades examining. Yet on his desk sat a Sharp Shoin, a Japanese word processor from another technological age.
Matsuoka was not keeping it as a decorative relic. In 2020, he described his workspace as having both a Sharp Shoin and a Dell computer on the desk, which he used alongside each other. Both were disconnected from communication networks, while messages reached him through his staff. A few years later, he wrote that he was still using the Shoin for his writing, despite the fact that the machines were no longer manufactured and had become increasingly difficult to repair.
The floppy disk was part of this arrangement. In an early account of the production of Senya Sensatsu, Matsuoka described finishing an installment and handing both the book and a floppy disk to a staff member. The staff would photograph the cover, proofread the text, establish links, and prepare the material for publication on the website. A major digital publishing project therefore had an unexpectedly old-fashioned point of entry: the author’s thoughts reached the organization on a small magnetic disk produced by a dedicated Japanese word processor.
Seen from the present, the workflow can appear slightly comical. One can imagine the accumulated technical anxieties surrounding it: aging drives, obsolete file formats, replacement machines, data conversion, and the possibility that one day a crucial piece of hardware would refuse to start. Yet there is also something familiar about the relationship between Matsuoka and the people around him.
Publishing history is full of stories about editors deciphering the nearly illegible handwritten manuscripts of celebrated novelists, learning the peculiarities of an author’s handwriting because someone had to transform those pages into printable text. In Matsuoka’s case, the manuscript was no longer handwritten, but the underlying arrangement had not entirely disappeared. The writer produced his text through the medium that suited him, while the people around him built the bridge between his private working environment and the public form of the work.
It would be easy to interpret his attachment to the Shoin as the stubbornness of an older intellectual who had failed to move with the times. That explanation does not fit very well. Matsuoka was hardly isolated from digital culture. His intellectual career dealt repeatedly with information, media, editing, networks, and the relationships among forms of knowledge. He had a computer beside the Shoin, understood the world into which his writing was being released, and participated in one of the most ambitious literary projects native to the Japanese web.
The more interesting question is therefore why he should have abandoned a tool through which he had already learned to think. Once a writing instrument becomes deeply incorporated into a person’s intellectual practice, it is no longer just equipment. The position of the keyboard, the rhythm of conversion, the appearance of text on the screen, the sequence of drafting and printing, and even the limitations of the machine can become part of a learned cognitive environment. What looks obsolete from the outside may remain entirely functional inside the mind of the person using it.
Matsuoka’s Shoin therefore raises a broader question. What exactly do writing tools do for us? Intellectual work may depend less on finding the objectively best technology than on constructing an arrangement in which reading, thinking, drafting, revising, and publishing can move together without unnecessary friction.
When Japanese Could Not Easily Be Typed
For writers working in alphabetic languages, the history of mechanical writing contains a strong element of continuity. The typewriter established the keyboard as a writing instrument long before the personal computer arrived. Mechanical typewriters became electric typewriters, electronic machines acquired memory and correction functions, dedicated word-processing systems appeared, and eventually software absorbed most of these functions into the personal computer.
The technological transitions were substantial, but one bodily relationship remained surprisingly stable. A person sat before a keyboard and pressed a key representing a letter. The machine changed, yet the basic grammar of typing remained recognizable. A writer who had learned to compose on a typewriter did not need to learn an entirely different relationship between language and the fingers when the computer arrived.
The history of Japanese writing was different because the writing system itself presented a mechanical problem. Modern Japanese combines hiragana, katakana, kanji, Roman letters, numbers, and punctuation. An alphabet can be represented on a keyboard with a manageable number of keys. Several thousand commonly used kanji cannot.
Japanese typewriters did exist, so it would be inaccurate to say that Japanese was literally impossible to type. The difficulty was that a machine containing thousands of characters could not offer the continuous relationship between thought and fingers that an alphabetic typewriter provided. Many Japanese typewriters required an operator to locate individual characters on a large character bed and select them mechanically. The process resembled typesetting as much as writing.
The difference becomes clearer if we imagine composition rather than transcription. When an English-speaking writer thinks of a word, the fingers can produce its letters in sequence. The keyboard maps relatively directly onto the written system. A Japanese writer thinking of a sentence containing kanji could not establish the same relationship merely by adding more keys. The machine needed some intermediate mechanism capable of turning a manageable form of input into the full written language.
Before that problem was solved, handwriting remained not only culturally important but technologically practical. Manuscript paper, pens, and pencils could handle the complexity of written Japanese without requiring a machine to understand the language. A writer could move naturally among kana, kanji, Roman letters, diagrams, corrections, marginal notes, and any other marks needed on the page.
This helps explain the extraordinary enthusiasm with which Tadao Umesao wrote about typewriters in his 1969 book The Art of Intellectual Production. One entire chapter, “From the Pen to the Typewriter,” moves through handwriting, Romanization, kana writing, kana typewriters, hiragana typewriters, and the broader problem of bringing Japanese into a mechanized writing environment.
To a reader today, some of this discussion may seem surprisingly intense. We type Japanese on laptops and smartphones without giving much thought to the engineering underneath. Umesao was writing when the problem was still visible. He was not discussing a minor preference in stationery. He was asking how Japanese intellectual work could participate in a larger transformation of information processing.
His concern extended beyond speed. He was interested in the whole technology of intellectual production: how to take notes, classify information, retrieve ideas, write manuscripts, and make knowledge available to others. If writing remained tied to laborious manual procedures, the entire system of knowledge work remained constrained by them.
The Japanese word processor changed that relationship. Toshiba’s JW-10, developed during the 1970s and shipped in 1979, is generally recognized as the first practical Japanese-language word processor. Its importance lay especially in practical kana-kanji conversion. A user could enter the phonetic reading of Japanese and allow the machine to select and display appropriate kanji, including choices among homophones. The first machine was enormous and extraordinarily expensive, but the underlying principle became one of the foundations of ordinary Japanese computer input.
This development did more than produce a convenient office appliance. It changed the relationship between Japanese and the keyboard. Thousands of possible written characters could be reached through a few dozen keys because the keyboard no longer needed to contain the writing system itself. It needed to provide a route into that writing system.
Once that route became familiar, typing could become a place for thinking rather than a mechanical stage that followed thought.
When the Machine Became a Place to Think
I remember this transition partly through my own experience with the Sharp Shoin. There was a period when the first draft still belonged to paper. I might write on the back of a flyer or on whatever sheet happened to be available, allowing an argument or a sequence of ideas to develop through handwriting. Only later would I sit down at the Shoin and turn those handwritten notes into a proper document.
In that arrangement, the word processor was not yet the true place of composition. It was closer to a sophisticated clean-copy machine. Thinking happened on paper, and the machine received something that thinking had already produced.
This pattern was hardly unique to Japanese writers. A scholar or journalist working in English might cover a yellow legal pad with handwritten notes before moving to a typewriter. Many writers continued drafting by hand even when typewriters were readily available. There has long been a sense, across writing cultures, that handwriting has a distinctive relationship with thought. The hand moves at a particular pace, deletion is inconvenient, and arrows, circles, fragments, marginal comments, and spatial arrangements can appear without requiring any command or interface.
The important change in my own practice came when I stopped needing that first handwritten stage. At some point I could sit at the Shoin with an unfinished thought and begin there. I no longer needed to know the complete sentence before touching the keyboard because I could discover the sentence through typing. Japanese and English research papers, reports, and other forms of writing could begin directly on the machine.
That shift sounds modest compared with the technological innovations surrounding it, but cognitively it was substantial. The word processor had moved from the second stage of writing into the first. Instead of writing by hand and then transferring the result into the machine, I had learned to think through typing.
This may help explain why people form deep attachments to particular writing environments. Once composition begins inside a machine, familiarity with that machine is no longer only familiarity with hardware. One learns where thought can hesitate, how easily a phrase can be replaced, what unfinished writing looks like on the screen, and how much text can remain visible while an argument is developing. The physical and visual environment begins to participate in the rhythm of composition.
For someone who has spent years or decades developing that relationship, changing the tool can carry a cognitive cost even when the replacement is technically superior. A familiar word processor may contain fewer functions, less memory, and a poorer display than a modern computer, yet the writer may have incorporated its limitations into a working method that no longer requires conscious attention.
The dedicated Japanese word processor eventually lost its reason for existing as a separate category of machine. Personal computers became powerful and affordable enough to absorb its functions, just as they absorbed calculators, filing systems, databases, desktop publishing, correspondence, photography, and eventually much of the communications environment.
For me, that transition was not difficult. By the time I moved fully into computer-based writing, the crucial transformation had already occurred. I had learned to compose through a keyboard. Whether kana-kanji conversion happened inside a Shoin or a Windows computer changed the machinery, but it did not change the fundamental relationship between my fingers and the emerging sentence.
Working in the IT industry also made computers an ordinary part of daily life. Windows became the workplace environment, while a MacBook eventually became my preferred personal environment. The dedicated word processor disappeared from my desk, but the search for the right writing surface did not end when the Shoin disappeared.
Returning to the Blank Surface
For many years, Microsoft Word seemed like the obvious place to write on a computer. It remains one of the most capable document-production tools ever created. It can handle long manuscripts, styles, references, tables, comments, revision histories, complex layouts, and countless other tasks. Yet capability and suitability are not always identical.
Over time, I found myself preferring plain-text and Markdown editors for actual composition. This was not because Word had become inadequate. In one sense, the opposite had happened. It offered more of the finished-document environment than I wanted during the earliest stage of writing.
A Word document already looks like a document. It has margins, fonts, pages, spacing, headers, styles, and the visual expectations of something moving toward completion. Even when the page is blank, the environment carries the language of finished documents. A Markdown file feels less committed. A heading can be marked with a simple symbol, emphasis requires no menu, paragraphs remain paragraphs, and the file itself is plain text that is relatively indifferent to the application displaying it.
That sparseness gives me an unexpected connection to the scraps of paper I once used before sitting down at the Shoin. The technologies could hardly be more different. A piece of discarded paper has no software at all, while Markdown belongs completely to the digital world. Yet they can produce a related psychological effect. Neither insists too early that I am producing a finished document. They allow words to exist before formatting has made too many claims about what those words are supposed to become.
This also helps explain why devices such as the Pomera and Freewrite continue to attract writers despite living in an age of computers that can do far more. Their limitations are part of their attraction. A laptop contains an extraordinary number of possible activities inside the same rectangle. One can write, search, check email, watch a video, open a message, consult social media, reorganize files, adjust settings, read the news, or disappear into another task without leaving the chair.
A machine designed primarily for writing changes that balance. It does not make concentration automatic, but it removes some of the exits. Matsuoka took this logic further by disconnecting his working computers from communication networks. His solution would be impractical for many people, especially those whose work requires constant collaboration, yet it reveals a useful tension in the history of technology. Progress is often measured by the number of functions added to a device, while intellectual work can sometimes improve when functions are deliberately removed.
There is no need to turn this into a campaign for distraction-free machines. My own preferred tool is still a fully connected computer, and I have no wish to return permanently to the Shoin. What I have retained is a desire for a writing surface that stays close to text. For me, Markdown provides that surface, while allowing the same text to move easily into other systems when the time comes to edit, publish, or archive it.
Then another interface began to change my relationship with writing more fundamentally.
When Writing Begins to Leave the Fingers
Typing once felt like liberation from the slowness of handwriting. It is faster, easier to revise, easier to copy, easier to search, and far more efficient for producing large amounts of text. A skilled typist can eventually reach a point at which the keyboard hardly seems to intervene because the fingers move while attention remains with the sentence.
Even so, typing still imposes a physical limit. Thoughts do not arrive one keystroke at a time. Speech offers another possibility because we can often explain an emerging idea aloud faster than we can type it. We can interrupt ourselves, return to an earlier point, insert an example, revise a claim in the middle of a sentence, and continue speaking. Conversation tolerates incompleteness because meaning does not depend entirely on perfectly formed sentences.
For years I regarded dictation as a logical next stage in my own history of writing. Handwriting had given way to typing, so it seemed reasonable to imagine that typing might eventually give way, at least partly, to speech. The idea was attractive long before the technology worked comfortably enough for the way I wanted to use it.
Earlier dictation systems demanded too much attention to the act of dictating. I had to pronounce words with unusual care. Recognition errors could interrupt the movement of thought. Punctuation often required special handling, while moving between Japanese and English introduced additional difficulties. Instead of allowing me to forget the interface, the system kept reminding me that an interface was present.
The result was a peculiar reversal. Dictation was supposed to free thought from the mechanical labor of typing, yet I sometimes found myself devoting more mental energy to satisfying the machine than I would have spent using a keyboard. The freedom I wanted from speech disappeared when speaking itself became a technical task, so I returned to typing.
Recent AI systems have changed that equation because they can operate at a different level from traditional speech recognition. The earlier ambition was largely transcriptional: convert an acoustic signal into the correct sequence of written words. That remains necessary, but contemporary AI can also work with the linguistic structure surrounding those words.
A spoken draft may contain repetitions, abandoned clauses, corrections, fillers, and ideas presented in an inconvenient order. Human conversation handles these imperfections constantly because listeners interpret meaning through them rather than treating every spoken fragment as a finished sentence. AI can increasingly perform a related kind of linguistic recovery.
This changes the writer’s responsibility during dictation. I still need to speak with enough intention that the system has something meaningful to work with, but I no longer need to behave as though I am reading a completed manuscript aloud. I can speak while the thought is still taking shape. Instead of requiring perfectly formed prose before the words leave my mouth, AI-supported dictation allows speech to remain closer to thinking aloud and lets some of the linguistic organization occur afterward.
Faster is not automatically better. Handwriting contains useful friction, and typing does as well. The delay imposed by the fingers can force a thought to wait long enough to become clearer. A writer who speaks faster than he can judge his own claims may produce more language without producing better thought.
For that reason, the history of writing technology should not be understood as a race toward zero friction. A more useful question is which kinds of friction help and which merely obstruct. AI-supported dictation removes some friction that I no longer need, while leaving reflection, structure, evidence, revision, and judgment for stages in which slowing down may be an advantage.
AI has also introduced something more consequential than improved dictation. Earlier writing tools could receive, store, display, and transform my words. A conversational system can respond to them.
When Conversation Enters the Writing Process
A notebook receives what I put into it. A typewriter records my keystrokes. A word processor allows me to rearrange them. A Markdown editor provides a flexible textual surface, while a dictation system turns speech into text. None of these tools, by itself, asks what I meant or proposes a different way of understanding the idea I have just expressed.
A conversational AI can do that, and this changes intellectual work in a way that is difficult to describe as another improvement in writing software. The interface is no longer limited to inscription. It becomes responsive.
I can begin with an intuition that has not yet become an argument. I can explain it conversationally, repeat myself, introduce an example, notice a contradiction, and ask whether there is a historical parallel. A response may expose a distinction I had not articulated. I can reject that distinction, modify it, or use it to sharpen the original thought. Another question follows, and the subject develops through exchange.
This does not remove authorship. If anything, it makes the location of authorship more interesting. The essential intellectual decisions still belong to the person doing the thinking: which question deserves attention, which response is convincing, which analogy goes too far, which fact needs verification, which interpretation belongs in the final piece, and which conclusion should remain unresolved. AI can participate in these movements without having to determine them.
For my own work, dialogue has increasingly become an intermediate medium between thinking and writing. I may speak an idea rather than type it. The spoken material becomes part of a conversation, the conversation exposes structures and connections, research can enter where necessary, and only later does the accumulated material become an essay.
This is different from sitting before a blank Word document and beginning at the title. It is also different from handing an AI a topic and asking it to produce an article. The part I find intellectually productive is the conversation itself. A thought moves outward before it is settled, encounters another linguistic system, and returns in a changed form. I respond to what comes back, sometimes by agreeing, but often by correcting, qualifying, resisting, or pushing further. Writing can emerge from that movement without the conversation replacing the writer’s responsibility for the finished work.
Seen from this perspective, the distance from Umesao to Matsuoka to AI becomes shorter than it first appears. Umesao experimented relentlessly with the infrastructure of intellectual production. Cards, notebooks, filing systems, typewriters, manuscript practices, and the physical arrangement of information interested him because thinking did not occur in isolation from its tools.
Matsuoka created his own elaborate ecology of reading, editing, writing, books, printed pages, computers, staff, networks, and the aging Shoin. From the outside, the arrangement could look technologically inconsistent. From inside the practice, it evidently worked.
My own arrangement now looks different again. It contains a MacBook, Windows computers at work, Markdown, ordinary web tools, digital research environments, AI-supported dictation, and extended conversations with AI. There is no reason these systems should resemble one another. Their common feature is more fundamental: each represents an attempt to construct an environment in which intellectual activity can move from perception to memory, from memory to association, from association to language, and from language into a form that can be shared.
Every Thinker Builds a Workshop
Looking back, I can describe my own history as a sequence of technologies. There was handwriting on whatever paper happened to be available. Then came the Sharp Shoin. The handwritten first draft eventually disappeared, and the word processor itself became the place where writing began. The personal computer absorbed that environment. Microsoft Word gave way, for much of my drafting, to the lighter surface of Markdown. Dictation became attractive, then disappointing, and later useful again once AI improved the relationship between speech and written language. Conversation with AI has now become another component of the process.
It would be easy to arrange these stages into a story of progress, but that would give too much authority to the chronology of machines. If technological progress alone determined the quality of intellectual work, Matsuoka should have discarded his Shoin as soon as a newer computer became available. He did not, and there is little reason to think that his refusal prevented him from producing work. Through that peculiar mixture of old equipment and contemporary publishing infrastructure came the immense body of Senya Sensatsu. The machine was old, while the intellectual production remained alive.
That distinction matters because we are surrounded by technologies designed to make older technologies appear inadequate. New devices promise greater speed, integration, intelligence, storage, connectivity, and convenience. Many of those improvements are real, and I have benefited from them throughout my working life. Intellectual work, however, has another measure of usefulness.
The questions are practical but also personal. Does the tool help me begin and stay with a difficult thought? Can an idea remain rough until I understand it, or does the environment encourage premature polishing? Can I retrieve what I need and move between fragments and larger structures? Does the technology interfere at the wrong moment? Can I revise without losing the original impulse? Above all, can I think through it?
No universal technology answers those questions for everyone. A fountain pen may be the right instrument for one person, while another may need a mechanical keyboard and several monitors. Someone else may prefer a Pomera that makes browsing inconvenient. A novelist may dictate while walking. A researcher may live inside Emacs or Obsidian. Another writer may be happiest in Microsoft Word. I am increasingly comfortable speaking thoughts into an AI-assisted environment and later bringing the developed material into Markdown.
A mature approach therefore does not require imitating the visible habits of admired intellectuals. Buying the same card boxes as Umesao would not reproduce Umesao’s mind. Finding an old Shoin would not turn someone into Matsuoka. Using the same AI system as another writer will not produce the same intellectual results.
The useful lesson lies beneath the particular tools. Umesao took the mechanics of knowledge work seriously enough to examine them. Matsuoka developed a system that fit the peculiar scale and style of his reading and writing. Their practices were personal without being arbitrary because they emerged from years of noticing where thought moved easily and where it became obstructed.
That remains useful in an age whose tools they could not have fully imagined. Each of us can build an intellectual workshop containing old and new technologies at the same time. Paper may remain useful for spatial thinking, Markdown may provide the freedom of an unfinished surface, speech may capture thought before the fingers can, and AI may introduce dialogue and transformation. A laptop can connect these activities to research and publication without requiring every stage of thinking to take place in the same interface.
The workshop does not need to look technologically coherent to someone else. It needs to support the movements of thought that matter to the person working inside it. Matsuoka’s combination of books, printed pages, Shoin, floppy disks, computers, staff, and website may have looked strange from outside, yet together they formed a functioning intellectual system.
There is still something amusing about imagining a celebrated intellectual in the twenty-first century finishing another immense meditation on books, saving it onto a floppy disk, and handing it to people responsible for getting those words onto the internet. I can imagine the staff maintaining the necessary machinery with the care once given to an eccentric master’s manuscripts. If anyone occasionally wished that Sensei would move to a modern computer, the thought would have been understandable.
Changing the Shoin, however, might also have changed something that did not need to be changed. The surrounding system could adapt instead. The floppy disk crossed the boundary between Matsuoka’s private writing environment and the public digital world, while editors and computers handled what came afterward. The reader encountered the finished work without needing to know what kind of keyboard had produced it.
This may be the most useful perspective from which to understand writing technology. The history of writing is filled with attempts to alter the distance between thought and expression. Pens made marks portable. Typewriters mechanized letters. Japanese word processors found a practical route from phonetic input to thousands of characters. Personal computers made revision, copying, and storage extraordinarily easy. Markdown stripped some of the digital surface away again. Dictation began moving writing out of the fingers. AI can now listen to unfinished speech and participate in the development of an idea.
None of these technologies abolishes the work of thinking. They rearrange where that work happens and redistribute the friction among different stages.
Umesao searched for a better route from Japanese thought to mechanically produced text. Matsuoka found a machine that became sufficiently intimate that he saw little reason to abandon it. My own path has taken me from handwritten scraps to the Shoin, from the Shoin to computers, from formatted documents to plain text, and from typing toward speech and dialogue with AI.
There will probably be another change, although I do not know what it will look like. I no longer think the interesting task is to predict the final writing technology because there may never be one. New tools will remove old constraints while introducing new forms of dependence. Some older practices will disappear, while others will return when we discover that the friction we tried to eliminate had been performing useful cognitive work.
The more durable task is to continue examining the relationship between thought and its instruments. We write with tools, but after enough use the relationship becomes more intimate than that description suggests. A tool that has been absorbed into a working practice can influence the rhythm in which ideas appear, the point at which they become sentences, the ease with which they can be revised, and the routes by which one thought encounters another.
Given enough time, we learn to think through our tools.
Leave a Reply