Augustine of Hippo lived a life of indulgence. As a young man, he took undue pride in his accomplishments, was a slave to his lustful desires, and even stole pears for the sinful thrill of it. But by his early 30s, the saint-to-be began investigating the nature of good and evil — a search that led him to the teachings of Christianity.
Wondering if he should convert, and fearing what such a decision would say about his life so far, Augustine sought solace and reflection in a garden. There, he heard a child in the neighboring house singing: “Take up and read; take up and read.” Interpreting the chant as a divine sign, he set about to find a book, and the first one he happened upon was the letters of the Apostle Paul.
Opening the book at random, his eyes fell upon Romans 13: “Let us walk honestly, as in the day; not in rioting and drunkenness, not in chambering and wantonness, not in strife and envying. But put ye on the Lord Jesus Christ, and make no provision for the flesh, to fulfil the lusts thereof.”
“I had no wish to read further, nor was there need,” Augustine wrote. “No sooner had I reached the end of the verse than the light of certainty flooded my heart and all dark shades of doubt fled away.” He closed the book — marking his place with “a finger between the leaves” — and went to tell the good news to his friend Alypius, who also converted.
Joel Miller opens his new book, The Idea Machine, with this famous scene from The Confessions because it sparked his own epiphany. Not a spiritual conversion, mind. What struck Miller during his recent reread was how Augustine marked his place with his finger. This seemingly unremarkable detail — a move any reader has made countless times — forced Miller to reevaluate books as not simply a vessel for ideas, but as history’s most successful “information technology.”
In marking his place, Miller realized, Augustine could do more than return to the passage later. By marking other books, he could collect and connect seemingly disparate ideas. He could put authors who never meet in conversation to spark new insights or reappraise old ones. And in writing The Confessions, Augustine gave himself a platform to refine his own ideas. All of which expanded Augustine’s thinking beyond the natural limits of human cognition.
“As a bearer of ideas alone, the book deserves our admiration,” Miller writes. “But reflecting on these additional facets of Augustine’s experience — which the form of the book made possible — we see that the book’s technological features and potentialities are equally important. Beyond what value the subject possesses, the object brings it forward and enables much else besides.” [Emphasis original.]
Giving the book its due
It may seem obvious that books are a technology, yet we rarely think about how they work. They just do.
Read most any list of history’s important inventions, and you’re unlikely to find the humble book. The printing press will absolutely be present, possibly taking the top spot. You may even spy paper in the running. While such innovations helped create the modern book — and we’ll discuss them soon — they are also distinct from the book as a technology in and of itself.
So, if books were so transformative, why don’t we think about them the same way we do microchips, microwaves, and the steam engine?
“It’s the success of the book which makes us not even think about it,” Miller tells me during an interview. “There is the old expression that familiarity breeds contempt. It actually breeds neglect more often than not. We fail to consider what a wonderful tool a book is and the many uses that it has.”
To help readers better understand all the book can offer, Miller analogizes it with an information technology that feels more, well, technological: the computer. Like a computer, books have their “hardware” — the ink, leaves, heavy board, and glue binding it all together — but they also have “software,” which are the ideas conveyed through them.
On the software side, books preserve ideas so they can be used at a later date — what Miller jokingly calls a “casual form of necromancy.” While books can’t contain any idea or experience with lossless fidelity, the variety and amount they can store is astounding. They can be written about law, science, medicine, and history. They can inform a reader about another person’s life or personal philosophy. They can even create vicarious experiences we would never enjoy in our daily lives, such as following the inner workings of an atom or exploring a fantasy world.
This describes the software at its simplest, but Miller adds that books do far more than preserve ideas. They make ideas more expressive by reaching people that writers could never hope to by word of mouth alone. They increase an idea’s specificity, developing it from a hunch to a concept to a full-on theory. And they do this over a period of time extending well beyond the original writer’s natural life.
As an example, Miller points to Charles Darwin’s On the Origin of Species. While sailing with the HMS Beagle, Darwin’s idea of evolution by natural selection was a hunch — an impressive hunch, but more intuition than idea. As he developed his thinking in his journals and letters to trusted colleagues, the idea became more expressive and specific. He then took the next step to write On the Origin, which sharpened the idea further and ensured it would be available for others to use long after Darwin’s passing.
And that’s exactly what happened. In the century and a half since its initial publication, the theory of evolution by natural selection has been refined by other thinkers, often by connecting it to ideas they found in other books. The big one is the laws of Mendelian inheritance, which provide a keystone for evolutionary theory today — and would have saved poor Darwin a heap of trouble had he known about Mendel’s work.
Repeat a similar process across countless readers, cultures, and eras, and the software of books has expanded and multiplied our ideas exponentially, helping to elevate “the human mind beyond its natural limits.”
“No human mind can compass the amount of information we produce in a year, let alone what we’ve accreted over, say, 3,500 years of civilization,” Miller says.
Every tech in the book
On the hardware side, the evolution of the book’s form has allowed the software to run more efficiently over time. Clay tablets sporting cuneiform are better than nothing, but they have certain disadvantages. They’re bulky, heavy, and scraping your nail across one will make your soul shudder. They are also cumbersome to sort and store. Just imagine, Miller says, a book the size of Middlemarch written on clay tablets.
Papyrus scrolls offset some of these problems: They are easier to produce and store, and papyrus is a far more flexible surface to write on. But they have their own drawbacks. For one, if you want to return to a particular passage later on, you can’t just open the bookmarked page. You have to unfurl the entire scroll up to the passage, and some scrolls could measure more than 100 feet long.
They also contained less text than modern books and were typically written in scriptio continua, a writing system in which words run together without punctuation and must be parsed syllabically. Scholars have variously referred to the result of this system as “a river of letters” or “a monolith of characters,” but to my eyes, it looks more like a cryptograph found on the back of a cereal box.
However you want to describe it, scriptio continua and the scrolls’ other difficulties made reading laborious. So laborious, Miller tells me, that few Greco-Romans had the time or privilege to learn, and even those elites would typically buy slaves to organize their libraries and read texts to them.
“Reading was an attainment of a certain type of person, and reading was never intended to be broadly applied,” Miller says. “Literacy in the Roman world would have been relatively high, [but] maybe only 5% of the people could read a public pronouncement or graffiti on a wall with a sort of workman’s literacy. The ability to read a philosophical text or decipher a poem [belonged to] a relatively tiny fragment of the population. And, by the way, that’s how they wanted it.”
The shift to the codex solved some of these problems. They are easier to store, can contain more text, and make passages easier to relocate. But it was paper and the printing press that gave the software its most important version update.
In his book, Miller cites scholarly estimates that between the 6th and 14th centuries, Western Europe produced roughly 6 million books, or 7,000 “and change” per year on average. In the 150 years after Gutenberg’s printing press, the number skyrocketed to 212 million, with some estimates placing it even higher. But the often unsung hero of this history is paper, which Miller says “solved the problem of scale” in a big way.
No human mind can compass the amount of information we produce in a year, let alone what we’ve accreted over, say, 3,500 years of civilization
Joel Miller
Parchment production is labor-intensive and time-consuming; it takes a lot of skinned animals to produce enough for one book. An 8th-century Latin translation of the Bible, called the Codex Amiatinus, required as many as 500 sheep to be slaughtered for its pages. While papyrus doesn’t require literal sacrificial lambs, it isn’t as durable as paper, and it is made from a single plant found mainly in Egypt. One bad harvest, and you’re out of luck.
Conversely, you can make reams of durable, sturdy paper relatively quickly and cheaply from the various fiber plant materials you can turn into pulp.
“If you look at every subsequent development in intellectual history in the West that depends on books, which is basically all of them, paper helped make that possible,” Miller says.
Consider science. Before paper and the printing press, it took a long time for scribes to make new copies of books. If a text was lost, damaged, or recycled for reuse in another book, it could take months to a year to produce another one — assuming it wasn’t the only copy available. Even if a copy existed and would-be scientists knew where it was, they may not be able to readily access it due to travel or other restrictions.
After the advent of paper, most scientists could own a personal copy of a particular treatise or monograph, and the copies would be nearly identical in content. Scientists working on the same problem could share information and solutions through books and letters more quickly, which proliferated the number of ideas available to work with. This abundance also allowed them to more easily check each other’s work and offer solutions or counterproposals.
“All of a sudden, science explodes in the West in a way it hadn’t before,” Miller says. “Instead of there being this mass scarcity of information — this scarcity to the software inside books — all the ideas, tools, and concepts conveyed in these books can be used to do more innovative work and push theories in all sorts of different directions simultaneously because you have lots of people working at the same time on similar problems.”
In search of a search engine
We just crammed far too many centuries into too few paragraphs, but this short history does hint at one of Miller’s themes in his book. As he put it, “The reality is that the book turned out to be so ridiculously useful that society adopted it. That’s how technological advancement always works.”
Another tradition is that new technology will have its opponents and naysayers, and while that may be difficult for us to imagine on this side of the literary revolution, it was true for the book as well. Socrates famously had misgivings about written philosophy; he worried that the medium weakens memory, leads to superficial understanding, and cannot accommodate new ideas or questions (as a dialogue partner can).
“There’s always a group of people that oppose [technological advancement] because they’re not clear on the trade-off,” Miller says. “They understand what they’re losing, but not what they’re gaining. Or worse, there’s a kind of narcissism — they don’t care or think about what other people may gain.”
Toward the end of our conversation, I asked Miller what his study of the history of books taught him about the technological shifts in our own time, specifically the information overload online and burgeoning AI scene.
Looking back, Miller points out that information overload has been with us for quite some time. “It’s deeply entrenched in the history of the book,” he says, “knowing that there is simply too much to know and constantly coming up with schemes, techniques, and methods to corral all the information. Humans are dang good at creating information, and we don’t think about the costs of creating information because we have so much to gain from it.”
For instance, the Royal Library of Ashurbanipal, named for the Neo-Assyrian king who amassed the collection in the 7th century B.C., housed tens of thousands of clay tablets. These contained information on laws, omens, rituals, annals, treaties, medicine, and magic, and with little more than waxen tablets, archivists had to devise a system to keep track of it all. If the Royal Library was “an internet of clay and sandal leather,” then this metadata system was its search engine.
The same is true for the Library of Alexandria, where librarians tagged scrolls, stored similar subjects together, and maintained vast catalogs. Some readers may even remember thumbing through card catalog cabinets in their elementary school libraries to find books using the Dewey Decimal System. All evolutionary stages of the search engine before it went digital.
“[The economist] Tyler Cowen made a comment in his book Stubborn Attachments that once an idea has been generated, it can be used at zero marginal costs by everyone else, but that’s only true if you can find it,” Miller says. “Gregor Mendel’s work was lost for decades until somebody ran across it and decided it was useful.”
One hope Miller has for generative AI is that large language models (LLMs) will help us grapple with this age-old problem more effectively.
While the internet allows us to share and discover ideas more quickly than ever — no more schlepping to the nearest monastery — it has also exponentially increased how much information we have to sift through to find the ideas we need. Digital search engines try to wrangle all that data, but they still leave us a lot to sift through, and their ranking systems can be manipulated.
Miller argues that LLMs have the potential to perform this task more efficiently. They could filter through unfathomable amounts of data to find the ideas users need — helping us to better connect our ideas and make the software of books even more powerful.
“It’s an underappreciated reason for LLMs,” Miller says. “They aren’t just for cheating on your history finals.”
In the book of memory
But let’s not forget those trade-offs. Socrates may have missed the mark on books, but as Miller points out, he wasn’t entirely wrong about them either. We do “outsource some of our memory to the page,” a trend that has only deepened with smartphones. They can also be one-sided. If we’re not careful to read widely on a particular subject, we can trick ourselves into believing we have a firmer grasp on it than we actually do.
When considering LLMs, Miller warns we must be mindful of similar trade-offs. He worries that we risk abdicating our critical thinking skills if we come to rely on AIs to assess arguments or solve every problem. We may also come to trust these machines to the point that we no longer check the accuracy of the answers they provide.
“If we de-skill that part of ourselves, we will have lost something essential actually functioning in a society,” Miller says.
And while our conversation focused on books as an informational technology, Miller stresses that they are an experiential technology, too.
Through literature, we vicariously embody the emotional lives of others. We see the world through their eyes, feel the crushing force of their defeats, and partake in their joys. Books provide people a means to share their common humanity — something that, for all their parroting, LLMs simply can’t do.
“An LLM can’t atomize human experience and present it back to us in a way that’s compelling,” Miller says. “A novelist can create an immersive experience that is emotionally, psychologically, intellectually, and morally rich. That cannot be duplicated by an LLM at this point, and maybe never, because an LLM has no human experience to draw on. All it has is the artifacts of those experiences.”
It’s another facet of the book’s power inherent in Saint Augustine’s story. While The Confessions is ultimately a treatise on theology and philosophy, it is also his story. Through it, readers experience his licentious youth, his regrets, and his search for the truth. They come to know his triumphs and losses and struggles. Augustine made full use of the technology of the book to share with others the breadth of his mind and soul. And in doing so, he has continued to inspire us for centuries.
“Who knows how [technology] is going to develop, but an LLM cannot be a human, and it cannot do the work that only a human can do,” Miller adds. “That privileges the book in a way that we might underappreciate right now. At the same time, it gives us a place to hope about how our interactions with the book may develop in the coming years.”
This article The most successful information technology in history is the one we barely notice is featured on Big Think.