Humans have existed in their anatomically modern form for over 200,000 years. For most of that time, we relied entirely on spoken language to communicate knowledge, memory, emotion, and culture.
Written language, in contrast, is a recent invention - dating back just over 5,000 years, and only widely adopted in literate societies in the last few centuries. Even more recent is the printed book, popularized through the invention of the printing press in the 15th century.
This means that for more than 95% of human history, knowledge has been spoken - not read.
And yet, we now live in a world where nonfiction knowledge is almost entirely encoded through compressed, edited, formalized text - optimized for printing, scanning, citation, and structural clarity.
This structure makes sense for certain use cases - but when we take this compressed format and simply read it aloud, as in the case of most audiobooks, we encounter something deeply unnatural.
Written and Spoken Language Are Not the Same
Spoken language is messy. It is layered, rhythmic, interactive, and self-correcting.
When we speak naturally, we don’t use punctuation. We don’t “write” in our heads. Our sentences often don’t end before the next one begins. We rely heavily on intonation, rhythm, stress, and timing to signal transitions and emphasis. We interrupt ourselves. We leave things unsaid. We double back and reframe ideas mid-sentence.
This isn’t a bug. It’s how human speech works.
Linguistic studies have shown that spoken syntax is fundamentally different from written syntax. In fact, spoken language often follows a “paratactic” structure (short clauses, loosely connected), while written language tends to favor hierarchical structure and subordination.
Nobody talks in the way nonfiction books are written.
The Problem with Audiobooks
When we take a nonfiction book - already optimized for maximal density and clarity on the page - and simply record it being read aloud word-for-word, we create a hybrid artifact that feels strangely unnatural.
It sounds like human speech, but lacks everything that makes human speech natural:
- No repetition for reinforcement
- No informal scaffolding
- No adaptive pacing
- No real-time stress or emphasis
- No dialogic rhythm
It’s a simulation of reading, not a conversation - and not really listening either. It doesn’t benefit from either medium fully.
This is why many people struggle to follow complex nonfiction audiobooks. It's not a matter of intelligence or attention span. It's a mismatch between format and cognitive expectation.
The Podcast Explosion
In contrast, podcasts feel natural.
Why?
Because they mirror the default mode of human communication: informal, redundant, interactive speech.
In a podcast, when a host makes a key point, they might:
- Reframe it two or three different ways
- Use an analogy or story
- Ask a clarifying question
- Pause, return, elaborate
- Receive a reaction from a co-host or guest
This repetition and modulation is not inefficient. It’s how the human mind learns and processes abstract ideas in real time. It's cognitively ergonomic.
So while the content of many podcasts may be less "dense" than books, their format is far more aligned with how people actually absorb knowledge.
The problem isn’t attention spans. It’s the delivery format.
Knowledge Compression Is a Tradeoff
Nonfiction books - especially the popular, idea-driven kind - are inherently compressed.
They’re designed to deliver an argument or concept in the most efficient form possible. That means removing redundancy, smoothing over tangents, structuring argument chains with precision, and limiting the length of examples.
This works well for print - where the reader can re-read, highlight, pause, reflect, skip ahead, or scan sections.
But it works very poorly when consumed linearly, passively, and at human speech speed - as is the case with audiobooks.
Listening to nonfiction audiobooks feels like constantly catching up with a conversation that’s already moved on.
Even more so when you revisit the same book a second time: listeners often report noticing huge amounts of content they missed the first time.
Not because they weren’t paying attention - but because the form itself was overloaded.
Toward Something New: Not a Better Audiobook - a Different Species
The problem isn't the book. It's the assumption that we can just feed its highly structured, compressed content into a speaker and call that a listening experience.
We evolved to listen long before we ever learned to read.
Throughout most of history, knowledge passed through conversation, storytelling, dialogue, and ritual monologue. In many traditional cultures, this persists: when elders speak, the young listen - sometimes in silence for hours, absorbing meaning through rhythm, tone, repetition, gesture, silence.
We didn’t learn by being read to. We learned by being spoken to - with intention, with variation, with space.
What if we used AI not to read a book aloud, but to speak it the way a human would?
Imagine two voices - like two elders walking, explaining, thinking aloud. Ideas surfaced slowly, reframed, challenged, circled back. Nothing forced, nothing rushed. A natural human tempo, not the artificial speed of linear narration.
This isn’t a podcast. It’s not a book. It’s something else:
An AI-humanified conversation, based on written knowledge but delivered in the way we evolved to absorb it - by listening.
This wouldn't replace reading. But it could replace the audiobook as we know it.
Not a simulation of reading, but a reconstruction of real listening.
A new oral interface for nonfiction.
Less friction. More absorption. More human.
Conclusion: Let’s Talk
We don’t need a better narrator.
We need to bring the text back to life - with rhythm, with redundancy, with dialogue.
Not by reading to the listener, but by talking with them.
Audiobooks were a halfway solution.
Now we can finish the job.
Let’s build the real thing.
Let’s talk.