top of page

The Cave Wall and the Machine Mind


On 12 September 1940, in a France that had signed an armistice three months earlier, four teenagers from Montignac lowered themselves into a hole in the ground. Marcel Ravidat had found the opening four days before and gone back for help: Jacques Marsal, Georges Agnel, Simon Coencas. What they dropped into was Lascaux. Horses, aurochs, stags, a whole ceiling of them, and nobody had looked at that ceiling since the last ice age.


Lascaux Cave Paintings
Lascaux Cave Paintings

Sixteen years later, almost the same week on the calendar, a much less romantic room.


11 September 1956. MIT, second day of a symposium run by the Special Interest Group in Information Theory. George Miller gave the paper on the seven-item ceiling of working memory. Noam Chomsky laid out generative syntax. Allen Newell and Herbert Simon presented a program called the Logic Theorist, which had proved thirty-eight theorems out of Principia Mathematica and found a shorter proof for one of them than Whitehead and Russell had. Years later Miller did something rare for a scientist. He named the day. Not the year. The day. 11 September 1956, when the mind and the machine stopped being two subjects and became one.


A cave wall in one September. A machine mind in another. Two days apart on the calendar, sixteen years apart in the world.


That is the subtitle of my book, and I did not arrange it. I noticed it while deciding when to publish. There you go, September just fits perfectly and here we are.


The book

The Handprint and the Algorithm: From the Cave Wall to the Machine Mind is out now.

It opens in a different cave. About 36,000 years ago, in the Ardèche gorges, somebody put a hand flat against rock, took a mouthful of ochre and water, and blew. The pigment went everywhere except where the hand was. Then the cliff above the entrance came down and sealed the room for longer than farming, writing and the printing press have existed. In December 1994, three cavers squeezed through a gap and found the hand exactly where it had been left.


The message arrived. It had never been addressed to anyone.


Every tool we have built since is a version of that afternoon. Writing. The press. The camera. The phonograph. And now machines trained on close to the entire written record of our species. My argument is that this is one line, unbroken, and that we keep missing it because every generation is certain its own invention is the first real one.


Every previous panic was right. Not "understandable given the times." Right. Socrates was right that writing would damage memory, and it did. John Philip Sousa was right that recorded sound would empty the parlors, and it did. Reading those old panics now is uncomfortable for a reason nobody expects, because they were mostly correct about the cost and completely wrong about the ending. Not one of them was the ending.


So, this is not a doom book and it is not a hype book. Both of those are easier to write and neither one survives contact with the history and facts.


It runs in four parts. The first asks what building thinking machines taught us about our own minds, which turned out to be considerably more than it taught us about the machines. The second gets specific about what these systems do all day, including the two things they do worst: inventing facts with total confidence, and having no yesterday. The third is the invoice. Work, money, art, energy, and what you think your job is. The fourth is where the questions get harder than the technology, and it covers alignment, consciousness, and what we might owe a thing that asks us not to switch it off.


There is exactly one chapter that explains how the machinery works, and it does it through rock music rather than linear algebra. You can skip it and the rest still holds.


Somewhere in the middle of building these things, something inconvenient happened. We set out to build a mind and we built a mirror. The book is about what the mirror shows.


Two languages


The book exists in English and in Portuguese, as A Marca da Mão e o Algoritmo: Das pinturas nas cavernas às máquinas que pensam.


The Portuguese is not a translation in the lazy sense. Quotations from Borges and from Milton were rendered from the original languages rather than passed through my English. Films carry their Brazilian release titles. Three chapters have a Brazilian clause the English does not, added only where the local fact changes the shape of the argument rather than to decorate it. The back cover and the author page were written natively in each language, because a translated author bio always sounds translated, and mine has jokes in it.


Both editions are available now.


There is more at igoralcantara.com/handprint: the chapter list, the sources behind every chapter, and the material that did not fit between two covers.


Two Septembers, sixteen years apart. Four boys with a lamp and three men with a theorem prover, all of them finding something in the dark that had been sitting there waiting.


The marks have started making themselves. The only question left is whose hand is on them.


Read more about my latest book here: https://igoralcantara.com/handprint


 
 
 

Comments


© 2026 Data Voyagers

  • Youtube
  • LinkedIn
bottom of page