My mental model of where we stand relative to deep history can be summarized as compounding progress. I assume we are the beneficiaries of all the progress that precedes us - not only each evolutionary unlock for life, but each hard won cultural victory over hunger, coldness, sickness, wild animals, superstition, plumbing, and car accidents.
The ‘past’ is an answer to the questions you ask. Here I ask questions that classify the world and easily move from one category of compounding growth to another as if these were continuous. This is would require a lot of work to defend in an academic context, but, roughly, I think it is the right thing to do, and nitpicking is a distraction.
The value of thinking about history this way is, for me, a tool to understand how to act with respect to the future. It takes vigilance to believe the acceleration of growth will continue (because we are so easily beguiled into playing a game defined by the past).
Consider all words in any language. If we take the year 300 AD as a touch point (for reasons we can set aside), how much was written down in all languages? About 98 million words in total. Maybe that grows a bit if you count memorization as a kind of writing (I’ve added in Sanskrit below).1
Writing introduced extraordinary advantages relative to memorization for memory and cross-generational transmission. The sum total of everything that had ever been committed to writing in human history in more or less any language in the year ~300 was about 98m words.
Total written words ~300 CE
| Language | Word Estimate | Notes |
|---|---|---|
| Greek | ~57,000,000 | Literary, philosophical, scientific, early Christian texts |
| Akkadian | ~10,500,000 | Cuneiform corpus (letters, literature, omens, laws) |
| Latin | ~10,000,000 | Literature, inscriptions, legal texts |
| Egyptian | ~6,000,000 | Hieroglyphic/hieratic/demotic (not Coptic) |
| Classical Chinese | ~5,700,000 | Up to Han dynasty (before 220 CE) |
| Sanskrit | ~3–5,000,000 | Vedic + early classical (conservative cutoff) |
| Sumerian | ~3,000,000 | Primarily literary/religious cuneiform |
| Syriac | ~500k–1,000,000 | Early Christian texts to 300 CE |
| Hittite | ~700,000 | |
| Hebrew | ~350,000 | Hebrew Bible + inscriptions up to 300 CE |
| Aramaic | ~100,000 | Inscriptions, DSS, biblical, papyri |
| Coptic | ~100,000 | Early Coptic biblical & Gnostic texts |
| Old Japanese | ~100,000 | Kojiki & Man'yōshū (just post-300 CE) |
| Avestan | ~85k–100,000 | Zoroastrian scripture (Avesta) |
| Old Korean | <10,000 | Hyangga and inscriptions (scarce) |
| Old Persian | ~7,000 | Royal inscriptions (cuneiform) |
I’ve read meaningful double digit percentage of the above texts – after a lifetime of scholarship they are a portable home to me, as they have been for scholars for many centuries.
There are 500m tweets per day on x.com – more than 80 years of YouTube videos are uploaded per day. The core activity in which humans distilled their learning and created a portable representation of their culture has grown to a scale that no human will ever be able to comprehend.
In 2024, I was meeting with a dear friend from Oxford. A scholar, deeply steeped in the world of texts, he is a novelist, and friend of many novelists and artists. He had never seen chatGPT – I casually asked chatGPT to write, in the style of John Coetzee, an additional chapter to one of his books. John is a Nobel laureate and my friend has been his official translator from time to time.
He read it carefully – having previously worked through every word with care, “Well, I am the last man.”
Total Hominin Population Over Time
Acceleration story in numbers
| Event / Boundary | Approx. Year | Time Between |
|---|---|---|
| Earth forms → First simple life (prokaryotes) | 4.54B → ~3.5B years ago | ~1.0B years |
| First simple life → Oxygenic cyanobacteria | ~3.5B → ~3.0–2.7B years ago | ~500–800M years |
| Cyanobacteria → Great Oxidation Event (GOE) | ~3.0–2.7B → 2.4B years ago | ~300–600M years lag |
| GOE → First eukaryotes | 2.4B → ~1.8–2.1B years ago | ~300–600M years |
| Eukaryotes → First multicellular organisms | ~2.0B → ~1.2B years ago | ~800M years |
| Multicellular organisms → First animals | ~1.2B → ~600M years ago | ~600M years |
| First animals → First land vertebrates | ~600M → ~360M years ago | ~240M years |
| Land vertebrates → First mammals | ~360M → ~200M years ago | ~160M years |
| Mammals → Neanderthals / early humans | 200M → ~400k years ago | ~199.6M years |
| Early humans → Symbolic systems | 300k → 100k years ago | ~200k years |
| Symbolic systems → Cities | ~100k → 7000 BCE | ~93k years |
| Cities → Writing | 7000 → 3400 BCE | ~3,600 years |
| Writing → Printing | 3400 BCE → 1450 CE | ~4,850 years |
| Printing → Industrial Revolution | 1450 → 1760 | ~310 years |
| Industrial Revolution → World Wide Web | 1760 → 1990 | ~230 years |
| World Wide Web → Social media | 1990 → 2004 | ~14 years |
| Social media → LLMs reach 100M users | 2004 → 2023 | ~19 years |
I depend here almost entirely on Enrique Jiménez’s 2017 Sofja Kovalevskaja Prize Speech given in Berlin in 2017. Enrique advocates for the digitization of texts and updates prior research by Michael Streck, „Großes Fach Altorientalistik: Der Umfang des keilschriftlichen Textkorpus." Deutsche Orient-Gesellschaft Mitteilungen (2010 Berlin) 142: 35-58. ↩︎